├── test
├── __init__.py
├── test_mt5.py
└── test_correlation.py
├── mt5_correlation
├── __init__.py
├── gui
│ ├── __init__.py
│ ├── mdi_child_log.py
│ ├── mdi_child_help.py
│ ├── mdi_child_diverged_symbols.py
│ ├── mdi_child_status.py
│ ├── mdi_child_divergedgraph.py
│ ├── mdi_child_correlationgraph.py
│ └── mdi.py
├── mt5.py
└── correlation.py
├── .gitignore
├── definitions.py
├── .idea
├── vcs.xml
├── misc.xml
├── inspectionProfiles
│ └── profiles_settings.xml
├── other.xml
├── modules.xml
└── mt5-correlation.iml
├── requirements.txt
├── LICENSE
├── mt5_correlation.py
├── config.yaml
├── README.md
└── configmeta.yaml
/test/__init__.py:
--------------------------------------------------------------------------------
1 |
--------------------------------------------------------------------------------
/mt5_correlation/__init__.py:
--------------------------------------------------------------------------------
1 |
--------------------------------------------------------------------------------
/.gitignore:
--------------------------------------------------------------------------------
1 | /venv/
2 | /.idea/workspace.xml
3 | /*.log
4 | /*.log.?
5 |
--------------------------------------------------------------------------------
/mt5_correlation/gui/__init__.py:
--------------------------------------------------------------------------------
1 | from mt5_correlation.gui.mdi import CorrelationMDIFrame
2 |
--------------------------------------------------------------------------------
/definitions.py:
--------------------------------------------------------------------------------
1 | import os
2 |
3 | ROOT_DIR = os.path.dirname(os.path.abspath(__file__))
4 | HELP_FILE = fr"{ROOT_DIR}\README.md"
5 | LOG_FILE = fr"{ROOT_DIR}\debug.log"
6 |
--------------------------------------------------------------------------------
/.idea/vcs.xml:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 |
5 |
6 |
--------------------------------------------------------------------------------
/.idea/misc.xml:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 |
--------------------------------------------------------------------------------
/requirements.txt:
--------------------------------------------------------------------------------
1 | pandas==1.2.1
2 | matplotlib==3.4.2
3 | MetaTrader5==5.0.34
4 | pytz==2021.1
5 | scipy==1.6.0
6 | logging==0.4.9.6
7 | markdown==3.3.4
8 | wxpython==4.1.1
9 | mock==4.0.3
10 | numpy==1.20.0
11 | wxconfig==1.2
--------------------------------------------------------------------------------
/.idea/inspectionProfiles/profiles_settings.xml:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 |
5 |
6 |
--------------------------------------------------------------------------------
/.idea/other.xml:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 |
5 |
6 |
7 |
--------------------------------------------------------------------------------
/.idea/modules.xml:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 |
5 |
6 |
7 |
8 |
--------------------------------------------------------------------------------
/.idea/mt5-correlation.iml:
--------------------------------------------------------------------------------
1 |
2 |
3 |
4 |
5 |
6 |
7 |
8 |
9 |
10 |
11 |
12 |
13 |
14 |
15 |
16 |
--------------------------------------------------------------------------------
/LICENSE:
--------------------------------------------------------------------------------
1 | MIT License
2 |
3 | Copyright (c) 2021 Jamie Cash
4 |
5 | Permission is hereby granted, free of charge, to any person obtaining a copy
6 | of this software and associated documentation files (the "Software"), to deal
7 | in the Software without restriction, including without limitation the rights
8 | to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
9 | copies of the Software, and to permit persons to whom the Software is
10 | furnished to do so, subject to the following conditions:
11 |
12 | The above copyright notice and this permission notice shall be included in all
13 | copies or substantial portions of the Software.
14 |
15 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
16 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
17 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
18 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
19 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
20 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
21 | SOFTWARE.
22 |
--------------------------------------------------------------------------------
/mt5_correlation.py:
--------------------------------------------------------------------------------
1 | """
2 | Application to monitor previously correlated symbol pairs for correlation divergence.
3 | """
4 | import definitions
5 | import logging.config
6 | from mt5_correlation.gui import CorrelationMDIFrame
7 | import wxconfig as cfg
8 | import wx
9 | import wx.lib.mixins.inspection as wit
10 |
11 |
12 | class InspectionApp(wx.App, wit.InspectionMixin):
13 | # Override app to use inspection.
14 | def OnInit(self):
15 | self.Init() # initialize the inspection tool
16 | return True
17 |
18 |
19 | if __name__ == "__main__":
20 | # Load the config
21 | cfg.Config().load(fr"{definitions.ROOT_DIR}\config.yaml", meta=fr"{definitions.ROOT_DIR}\configmeta.yaml")
22 |
23 | # Get logging config and configure the logger
24 | log_config = cfg.Config().get('logging')
25 | logging.config.dictConfig(log_config)
26 |
27 | # Do we have inspection turned on. Create correct version of app
28 | inspection = cfg.Config().get('developer.inspection')
29 | if inspection:
30 | app = InspectionApp()
31 | else:
32 | app = wx.App(False)
33 |
34 | # Start the app
35 | frame = CorrelationMDIFrame()
36 | frame.Show()
37 | app.MainLoop()
38 |
--------------------------------------------------------------------------------
/mt5_correlation/gui/mdi_child_log.py:
--------------------------------------------------------------------------------
1 | import definitions
2 | import wx
3 | import wx.html
4 |
5 | import mt5_correlation.gui.mdi as mdi
6 |
7 |
8 | class MDIChildLog(mdi.CorrelationMDIChild):
9 | """
10 | Shows the debug.log file
11 | """
12 |
13 | __log_window = None # Widget to display log file in
14 |
15 | def __init__(self, parent):
16 | # Super
17 | mdi.CorrelationMDIChild.__init__(self, parent=parent, id=wx.ID_ANY, pos=wx.DefaultPosition, title="Log",
18 | size=wx.Size(width=800, height=200),
19 | style=wx.DEFAULT_FRAME_STYLE)
20 |
21 | # Panel and sizer for help file
22 | panel = wx.Panel(self, wx.ID_ANY)
23 | sizer = wx.BoxSizer()
24 | panel.SetSizer(sizer)
25 |
26 | # Log file window
27 | self.__log_window = wx.TextCtrl(parent=panel, id=wx.ID_ANY, style=wx.HSCROLL | wx.TE_MULTILINE | wx.TE_READONLY)
28 | sizer.Add(self.__log_window, 1, wx.ALL | wx.EXPAND)
29 |
30 | # Refresh to populate
31 | self.refresh()
32 |
33 | def refresh(self):
34 | """
35 | Refresh the log file
36 | :return:
37 | """
38 | # Load the help file
39 | self.__log_window.LoadFile(definitions.LOG_FILE)
40 |
41 | # Scroll to bottom
42 | self.__log_window.SetInsertionPoint(-1)
43 |
44 |
--------------------------------------------------------------------------------
/mt5_correlation/gui/mdi_child_help.py:
--------------------------------------------------------------------------------
1 | import definitions
2 | import markdown
3 | import wx
4 | import wx.html
5 |
6 | import mt5_correlation.gui.mdi as mdi
7 |
8 |
9 | class MDIChildHelp(mdi.CorrelationMDIChild):
10 | """
11 | Shows the README.md file
12 | """
13 |
14 | def __init__(self, parent):
15 | # Super
16 | mdi.CorrelationMDIChild.__init__(self, parent=parent, id=wx.ID_ANY, pos=wx.DefaultPosition, title="Help",
17 | size=wx.Size(width=800, height=-1),
18 | style=wx.DEFAULT_FRAME_STYLE)
19 |
20 | # Panel and sizer for help file
21 | panel = wx.Panel(self, wx.ID_ANY)
22 | sizer = wx.BoxSizer()
23 | panel.SetSizer(sizer)
24 |
25 | # HtmlWindow
26 | html_widget = wx.html.HtmlWindow(parent=panel, id=wx.ID_ANY, style=wx.html.HW_SCROLLBAR_AUTO | wx.html.HW_NO_SELECTION)
27 | sizer.Add(html_widget, 1, wx.ALL | wx.EXPAND)
28 |
29 | # Load the help file, convert markdown to HTML and display. The markdown library doesnt understand the shell
30 | # format so we will remove.
31 | markdown_text = open(definitions.HELP_FILE).read()
32 | markdown_text = markdown_text.replace("```shell", "```")
33 | html = markdown.markdown(markdown_text)
34 | html_widget.SetPage(html)
35 |
36 | def refresh(self):
37 | """
38 | Nothing to do on refresh. Help file doesnt change during outside of development.
39 | :return:
40 | """
41 | pass
42 |
--------------------------------------------------------------------------------
/config.yaml:
--------------------------------------------------------------------------------
1 | ---
2 | calculate:
3 | from:
4 | days: 10
5 | timeframe: 15
6 | min_prices: 400
7 | max_set_size_diff_pct: 90
8 | overlap_pct: 90
9 | max_p_value: 0.05
10 | monitor:
11 | interval: 10
12 | calculations:
13 | long:
14 | from: 30
15 | min_prices: 300
16 | max_set_size_diff_pct: 50
17 | overlap_pct: 50
18 | max_p_value: 0.05
19 | medium:
20 | from: 10
21 | min_prices: 100
22 | max_set_size_diff_pct: 50
23 | overlap_pct: 50
24 | max_p_value: 0.05
25 | short:
26 | from: 2
27 | min_prices: 30
28 | max_set_size_diff_pct: 50
29 | overlap_pct: 50
30 | max_p_value: 0.05
31 | monitoring_threshold: 0.9
32 | divergence_threshold: 0.8
33 | monitor_inverse: true
34 | tick_cache_time: 10
35 | autosave: true
36 | logging:
37 | version: 1
38 | disable_existing_loggers: false
39 | formatters:
40 | brief:
41 | format: '%(asctime)s - %(name)s - %(levelname)s - %(message)s'
42 | datefmt: '%H:%M:%S'
43 | precice:
44 | format: '%(asctime)s - %(threadName)s - %(name)s - %(levelname)s - %(message)s'
45 | datefmt: '%Y-%m-%d %H:%M:%S'
46 | handlers:
47 | console:
48 | level: INFO
49 | class: logging.StreamHandler
50 | formatter: brief
51 | stream: ext://sys.stdout
52 | file:
53 | level: DEBUG
54 | class: logging.handlers.RotatingFileHandler
55 | formatter: precice
56 | filename: debug.log
57 | mode: a
58 | maxBytes: 2560000
59 | backupCount: 1
60 | root:
61 | level: DEBUG
62 | handlers:
63 | - console
64 | - file
65 | loggers:
66 | mt5-correlation:
67 | level: DEBUG
68 | handlers:
69 | - console
70 | - file
71 | propagate: 0
72 | charts:
73 | colormap: Dark2
74 | developer:
75 | inspection: true
76 | window:
77 | x: -4
78 | y: 1
79 | width: 1626
80 | height: 1043
81 | style: 541072960
82 | settings_window:
83 | x: 354
84 | y: 299
85 | width: 624
86 | height: 328
87 | style: 524352
88 | ...
--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
1 | # mt5-correlation
2 | Calculates correlation coefficient between all symbols in MetaTrader5 Market Watch and monitors previously correlated symbol pairs for divergence
3 |
4 | # Setup
5 | 1) Set up your MetaTrader 5 environment ensuring that all symbols that you would like to assess for correlation are shown in your Market Watch window;
6 | 2) Set up your python environment; and
7 | 3) Install the required libraries.
8 |
9 | ```shell
10 | pip install -r mt5-correlation/requirements.txt
11 | ```
12 |
13 | # Usage
14 | If you set up a virtual environment in the Setup step, ensure this is activated. Then run the script.
15 |
16 | ```shell
17 | python -m mt5_correlations/mt5_correlations.py
18 | ```
19 |
20 | This will open a GUI.
21 |
22 | ## Calculating Baseline Coefficients
23 | On the first time that you run, you will want to calculate the initial set of correlations.
24 | 1) Open settings and review the settings under the 'Calculate' tab. Hover over the individual settings for help.
25 | 2) Set the threshold for the correlations to monitor. This can be set under the Settings 'Monitoring' tab and is named 'Monitoring Threshold'. Only settings with a coefficient over this threshold will be displayed and monitored.
26 | 3) Calculate the coefficients by selecting File/Calculate. All symbol pairs that have a correlation coefficient greater than the monitoring threshold will be displayed. A graph showing the price candle data used to calculate the coefficient for the pair will be displayed when you select the row.
27 | 4) You may want to save. Select File/Save and choose a file name. This file will contain all the calculated coefficients, and the price data used to calculate them. This file can be loaded to avoid having to recalculate the baseline coefficients every time you use the application.
28 |
29 | ## Monitoring for Divergence
30 | Once you have calculated or loaded the baseline coefficients, they can be monitored for divergence.
31 | 1) Open settings and review the settings under the 'Monitor' tab. Hover over the individual settings for help.
32 | 2) Switch the monitoring toggle to on. The application will continuously monitor for divergence. The data frame will be updated with the last time that the correlation was checked, and the last coefficient. The chart frame will contain 3 charts which will be updated after every monitoring event:
33 | - One showing the price history data used to calculate the baseline coefficient for both symbols in the correlated pair;
34 | - One showing the tick data used to calculate the latest coefficient for both symbols in the correlated pair.
35 | - One showing every correlation coefficient calculated for the symbol pair.
--------------------------------------------------------------------------------
/test/test_mt5.py:
--------------------------------------------------------------------------------
1 | import unittest
2 | from unittest.mock import patch
3 | import mt5_correlation.mt5 as mt5
4 | import pandas as pd
5 | from datetime import datetime
6 |
7 |
8 | class Symbol:
9 | """ A Mock symbol class"""
10 | name = None
11 | visible = None
12 |
13 | def __init__(self, name, visible):
14 | self.name = name
15 | self.visible = visible
16 |
17 |
18 | class TestMT5(unittest.TestCase):
19 | """
20 | Unit test for MT5. Uses mock to mock MetaTrader5 connection.
21 | """
22 |
23 | # Mock symbols. 5 Symbols, 4 visible.
24 | mock_symbols = [Symbol(name='SYMBOL1', visible=True),
25 | Symbol(name='SYMBOL2', visible=True),
26 | Symbol(name='SYMBOL3', visible=False),
27 | Symbol(name='SYMBOL4', visible=True),
28 | Symbol(name='SYMBOL5', visible=True)]
29 |
30 | # Mock prices for symbol 1
31 | mock_prices = pd.DataFrame(columns=['time', 'close'],
32 | data=[[datetime(2021, 1, 1, 1, 5, 0), 123.123],
33 | [datetime(2021, 1, 1, 1, 10, 0), 123.124],
34 | [datetime(2021, 1, 1, 1, 15, 0), 123.125],
35 | [datetime(2021, 1, 1, 1, 20, 0), 125.126],
36 | [datetime(2021, 1, 1, 1, 25, 0), 123.127],
37 | [datetime(2021, 1, 1, 1, 30, 0), 123.128]])
38 |
39 | @patch('mt5_correlation.mt5.MetaTrader5')
40 | def test_get_symbols(self, mock):
41 | # Mock return value
42 | mock.symbols_get.return_value = self.mock_symbols
43 |
44 | # Call get_symbols
45 | symbols = mt5.MT5().get_symbols()
46 |
47 | # There should be four, as one is set as not visible
48 | self.assertTrue(len(symbols) == 4, "There should be 5 symbols returned from MT5.")
49 |
50 | @patch('mt5_correlation.mt5.MetaTrader5')
51 | def test_get_prices(self, mock):
52 | # Mock return value
53 | mock.copy_rates_range.return_value = self.mock_prices
54 |
55 | # Call get prices
56 | prices = mt5.MT5().get_prices(symbol='SYMBOL1', from_date='01-JAN-2021 01:00:00',
57 | to_date='01-JAN-2021 01:10:25', timeframe=mt5.TIMEFRAME_M5)
58 |
59 | # There should be 6
60 | self.assertTrue(len(prices.index) == 6, "There should be 6 prices.")
61 |
62 | @patch('mt5_correlation.mt5.MetaTrader5')
63 | def test_get_ticks(self, mock):
64 | # Mock return value
65 | mock.copy_ticks_range.return_value = self.mock_prices
66 |
67 | # Call get ticks
68 | ticks = mt5.MT5().get_ticks(symbol='SYMBOL1', from_date='01-JAN-2021 01:00:00', to_date='01-JAN-2021 01:10:25')
69 |
70 | # There should be 6
71 | self.assertTrue(len(ticks.index) == 6, "There should be 6 prices.")
72 |
73 |
74 | if __name__ == '__main__':
75 | unittest.main()
76 |
77 |
78 |
--------------------------------------------------------------------------------
/mt5_correlation/mt5.py:
--------------------------------------------------------------------------------
1 | import pandas as pd
2 | import MetaTrader5
3 | import logging
4 |
5 | # Timeframes
6 | TIMEFRAME_M1 = MetaTrader5.TIMEFRAME_M1
7 | TIMEFRAME_M2 = MetaTrader5.TIMEFRAME_M2
8 | TIMEFRAME_M3 = MetaTrader5.TIMEFRAME_M3
9 | TIMEFRAME_M4 = MetaTrader5.TIMEFRAME_M4
10 | TIMEFRAME_M5 = MetaTrader5.TIMEFRAME_M5
11 | TIMEFRAME_M6 = MetaTrader5.TIMEFRAME_M6
12 | TIMEFRAME_M10 = MetaTrader5.TIMEFRAME_M10
13 | TIMEFRAME_M12 = MetaTrader5.TIMEFRAME_M12
14 | TIMEFRAME_M15 = MetaTrader5.TIMEFRAME_M15
15 | TIMEFRAME_M20 = MetaTrader5.TIMEFRAME_M20
16 | TIMEFRAME_M30 = MetaTrader5.TIMEFRAME_M30
17 | TIMEFRAME_H1 = MetaTrader5.TIMEFRAME_H1
18 | TIMEFRAME_H2 = MetaTrader5.TIMEFRAME_H2
19 | TIMEFRAME_H3 = MetaTrader5.TIMEFRAME_H3
20 | TIMEFRAME_H4 = MetaTrader5.TIMEFRAME_H4
21 | TIMEFRAME_H6 = MetaTrader5.TIMEFRAME_H6
22 | TIMEFRAME_H8 = MetaTrader5.TIMEFRAME_H8
23 | TIMEFRAME_H12 = MetaTrader5.TIMEFRAME_H12
24 | TIMEFRAME_D1 = MetaTrader5.TIMEFRAME_D1
25 | TIMEFRAME_W1 = MetaTrader5.TIMEFRAME_W1
26 | TIMEFRAME_MN1 = MetaTrader5.TIMEFRAME_MN1
27 |
28 |
29 | class MT5:
30 | """
31 | A class to connect to and interface with MetaTrader 5
32 | """
33 |
34 | def __init__(self):
35 | # Connect to MetaTrader5. Opens if not already open.
36 |
37 | # Logger
38 | self.__log = logging.getLogger(__name__)
39 |
40 | # Open MT5 and log error if it could not open
41 | if not MetaTrader5.initialize():
42 | self.__log.error("initialize() failed")
43 | MetaTrader5.shutdown()
44 |
45 | # Print connection status
46 | self.__log.debug(MetaTrader5.terminal_info())
47 |
48 | # Print data on MetaTrader 5 version
49 | self.__log.debug(MetaTrader5.version())
50 |
51 | def __del__(self):
52 | # shut down connection to the MetaTrader 5 terminal
53 | MetaTrader5.shutdown()
54 |
55 | def get_symbols(self):
56 | """
57 | Gets list of symbols open in MT5 market watch.
58 | :return: list of symbol names
59 | """
60 | # Iterate symbols and get those in market watch.
61 | symbols = MetaTrader5.symbols_get()
62 | selected_symbols = []
63 | for symbol in symbols:
64 | if symbol.visible:
65 | selected_symbols.append(symbol.name)
66 |
67 | # Log symbol counts
68 | total_symbols = MetaTrader5.symbols_total()
69 | num_selected_symbols = len(selected_symbols)
70 | self.__log.debug(f"{num_selected_symbols} of {total_symbols} available symbols in Market Watch.")
71 |
72 | return selected_symbols
73 |
74 | def get_prices(self, symbol, from_date, to_date, timeframe):
75 | """
76 | Gets OHLC price data for the specified symbol.
77 | :param symbol: The name of the symbol to get the price data for.
78 | :param from_date: Date from when to retrieve data
79 | :param to_date: Date where to receive data to
80 | :param timeframe: The timeframe for the candes. Possible values are:
81 | TIMEFRAME_M1: 1 minute
82 | TIMEFRAME_M2: 2 minutes
83 | TIMEFRAME_M3: 3 minutes
84 | TIMEFRAME_M4: 4 minutes
85 | TIMEFRAME_M5: 5 minutes
86 | TIMEFRAME_M6: 6 minutes
87 | TIMEFRAME_M10: 10 minutes
88 | TIMEFRAME_M12: 12 minutes
89 | TIMEFRAME_M15: 15 minutes
90 | TIMEFRAME_M20: 20 minutes
91 | TIMEFRAME_M30: 30 minutes
92 | TIMEFRAME_H1: 1 hour
93 | TIMEFRAME_H2: 2 hours
94 | TIMEFRAME_H3: 3 hours
95 | TIMEFRAME_H4: 4 hours
96 | TIMEFRAME_H6: 6 hours
97 | TIMEFRAME_H8: 8 hours
98 | TIMEFRAME_H12: 12 hours
99 | TIMEFRAME_D1: 1 day
100 | TIMEFRAME_W1: 1 week
101 | TIMEFRAME_MN1: 1 month
102 | :return: Price data for symbol as dataframe
103 | """
104 |
105 | prices_dataframe = None
106 |
107 | # Get prices from MT5
108 | prices = MetaTrader5.copy_rates_range(symbol, timeframe, from_date, to_date)
109 | if prices is not None:
110 | self.__log.debug(f"{len(prices)} prices retrieved for {symbol}.")
111 |
112 | # Create dataframe from data and convert time in seconds to datetime format
113 | prices_dataframe = pd.DataFrame(prices)
114 | prices_dataframe['time'] = pd.to_datetime(prices_dataframe['time'], unit='s')
115 |
116 | return prices_dataframe
117 |
118 | def get_ticks(self, symbol, from_date, to_date):
119 | """
120 | Gets OHLC price data for the specified symbol.
121 | :param symbol: The name of the symbol to get the price data for.
122 | :param from_date: Date from when to retrieve data
123 | :param to_date: Date where to receive data to
124 | :return: Tick data for symbol as dataframe
125 | """
126 |
127 | ticks_dataframe = None
128 |
129 | # Get ticks from MT5
130 | ticks = MetaTrader5.copy_ticks_range(symbol, from_date, to_date, MetaTrader5.COPY_TICKS_ALL)
131 |
132 | # If ticks is None, there was an error
133 | if ticks is None:
134 | error = MetaTrader5.last_error()
135 | self.__log.error(f"Error retrieving ticks for {symbol}: {error}")
136 | else:
137 | self.__log.debug(f"{len(ticks)} ticks retrieved for {symbol}.")
138 |
139 | # Create dataframe from data and convert time in seconds to datetime format
140 | try:
141 | ticks_dataframe = pd.DataFrame(ticks)
142 | ticks_dataframe['time'] = pd.to_datetime(ticks_dataframe['time'], unit='s')
143 | except RecursionError:
144 | self.__log.warning("Error converting ticks to dataframe.")
145 |
146 | return ticks_dataframe
147 |
--------------------------------------------------------------------------------
/mt5_correlation/gui/mdi_child_diverged_symbols.py:
--------------------------------------------------------------------------------
1 | import logging
2 | import pandas as pd
3 | import wx
4 | import wx.grid
5 |
6 | import mt5_correlation.gui.mdi as mdi
7 |
8 | # Columns for diverged symbols table
9 | COLUMN_INDEX = 0
10 | COLUMN_SYMBOL = 1
11 | COLUMN_NUM_DIVERGENCES = 2
12 |
13 |
14 | class MDIChildDivergedSymbols(mdi.CorrelationMDIChild):
15 | """
16 | Shows the status of all correlations that are within the monitoring threshold
17 | """
18 |
19 | # The table and grid containing the symbols that form part of the diverged correlations. Defined at instance level
20 | # to enable refresh.
21 | __table = None
22 | __grid = None
23 |
24 | # Number of rows. Required for and updated by refresh method
25 | __rows = 0
26 |
27 | __log = None # The logger
28 |
29 | def __init__(self, parent):
30 | # Super
31 | wx.MDIChildFrame.__init__(self, parent=parent, id=wx.ID_ANY, title="Diverged Symbols",
32 | size=wx.Size(width=240, height=-1), style=wx.DEFAULT_FRAME_STYLE)
33 |
34 | # Create logger
35 | self.__log = logging.getLogger(__name__)
36 |
37 | # Panel and sizer for table
38 | panel = wx.Panel(self, wx.ID_ANY)
39 | sizer = wx.BoxSizer(wx.VERTICAL)
40 | panel.SetSizer(sizer)
41 |
42 | # Create the grid. This is a data table using pandas dataframe for underlying data. Add the
43 | # grid to the sizer.
44 | self.__table = _DataTable(columns=self.GetMDIParent().cor.diverged_symbols.columns)
45 | self.__grid = wx.grid.Grid(panel, wx.ID_ANY)
46 | self.__grid.SetTable(self.__table, takeOwnership=True)
47 | self.__grid.EnableEditing(False)
48 | self.__grid.EnableDragRowSize(False)
49 | self.__grid.EnableDragColSize(True)
50 | self.__grid.EnableDragGridSize(True)
51 | self.__grid.SetSelectionMode(wx.grid.Grid.SelectRows)
52 | self.__grid.SetRowLabelSize(0)
53 | self.__grid.SetColSize(COLUMN_INDEX, 0) # Index. Hide
54 | self.__grid.SetColSize(COLUMN_SYMBOL, 100) # Symbol
55 | self.__grid.SetColSize(COLUMN_NUM_DIVERGENCES, 100) # Num divergences
56 | self.__grid.SetMinSize((220, 100))
57 | sizer.Add(self.__grid, 1, wx.ALL | wx.EXPAND)
58 |
59 | # Bind row doubleclick
60 | self.Bind(wx.grid.EVT_GRID_CELL_LEFT_DCLICK, self.__on_doubleckick_row, self.__grid)
61 |
62 | # Refresh to populate
63 | self.refresh()
64 |
65 | def refresh(self):
66 | """
67 | Refreshes grid. Notifies if rows have been added or deleted.
68 | :return:
69 | """
70 | self.__log.debug(f"Refreshing grid.")
71 |
72 | # Update data
73 | self.__table.data = self.GetMDIParent().cor.diverged_symbols.copy()
74 |
75 | # Start refresh
76 | self.__grid.BeginBatch()
77 |
78 | # Check if num rows in dataframe has changed, and send appropriate APPEND or DELETE messages
79 | cur_rows = len(self.GetMDIParent().cor.diverged_symbols.index)
80 | if cur_rows < self.__rows:
81 | # Data has been deleted. Send message
82 | msg = wx.grid.GridTableMessage(self.__table, wx.grid.GRIDTABLE_NOTIFY_ROWS_DELETED,
83 | self.__rows - cur_rows, self.__rows - cur_rows)
84 | self.__grid.ProcessTableMessage(msg)
85 | elif cur_rows > self.__rows:
86 | # Data has been added. Send message
87 | msg = wx.grid.GridTableMessage(self.__table, wx.grid.GRIDTABLE_NOTIFY_ROWS_APPENDED,
88 | cur_rows - self.__rows) # how many
89 | self.__grid.ProcessTableMessage(msg)
90 |
91 | self.__grid.EndBatch()
92 |
93 | # Send updated message
94 | msg = wx.grid.GridTableMessage(self.__table, wx.grid.GRIDTABLE_REQUEST_VIEW_GET_VALUES)
95 | self.__grid.ProcessTableMessage(msg)
96 |
97 | # Update row count
98 | self.__rows = cur_rows
99 |
100 | def __on_doubleckick_row(self, evt):
101 | """
102 | Open the graphs when a row is doubleclicked.
103 | :param evt:
104 | :return:
105 | """
106 | row = evt.GetRow()
107 | symbol = self.__grid.GetCellValue(row, COLUMN_SYMBOL)
108 |
109 | mdi.FrameManager.open_frame(parent=self.GetMDIParent(),
110 | frame_module='mt5_correlation.gui.mdi_child_divergedgraph',
111 | frame_class='MDIChildDivergedGraph',
112 | raise_if_open=True,
113 | symbol=symbol)
114 |
115 |
116 | class _DataTable(wx.grid.GridTableBase):
117 | """
118 | A data table that holds data in a pandas dataframe.
119 | """
120 | data = None # The data for this table. A Pandas DataFrame
121 |
122 | def __init__(self, columns):
123 | wx.grid.GridTableBase.__init__(self)
124 | self.headerRows = 1
125 | self.data = pd.DataFrame(columns=columns)
126 |
127 | def GetNumberRows(self):
128 | return len(self.data)
129 |
130 | def GetNumberCols(self):
131 | return len(self.data.columns) + 1
132 |
133 | def GetValue(self, row, col):
134 | if row < self.RowsCount and col < self.ColsCount:
135 | return self.data.index[row] if col == 0 else self.data.iloc[row, col - 1]
136 | else:
137 | raise Exception(f"Trying to access row {row} and col {col} which does not exist.")
138 |
139 | def SetValue(self, row, col, value):
140 | self.data.iloc[row, col - 1] = value
141 |
142 | def GetColLabelValue(self, col):
143 | if col == 0:
144 | if self.data.index.name is None:
145 | return 'Index'
146 | else:
147 | return self.data.index.name
148 | return str(self.data.columns[col - 1])
149 |
150 | def GetTypeName(self, row, col):
151 | return wx.grid.GRID_VALUE_STRING
152 |
153 | def GetAttr(self, row, col, prop):
154 | attr = wx.grid.GridCellAttr()
155 |
156 | return attr
157 |
--------------------------------------------------------------------------------
/mt5_correlation/gui/mdi_child_status.py:
--------------------------------------------------------------------------------
1 | import logging
2 | import pandas as pd
3 | import wx
4 | import wx.grid
5 |
6 | from mt5_correlation import correlation as cor
7 | import mt5_correlation.gui.mdi as mdi
8 |
9 | # Columns for coefficient table
10 | COLUMN_INDEX = 0
11 | COLUMN_SYMBOL1 = 1
12 | COLUMN_SYMBOL2 = 2
13 | COLUMN_BASE_COEFFICIENT = 3
14 | COLUMN_DATE_FROM = 4
15 | COLUMN_DATE_TO = 5
16 | COLUMN_TIMEFRAME = 6
17 | COLUMN_LAST_CALCULATION = 7
18 | COLUMN_STATUS = 8
19 |
20 |
21 | class MDIChildStatus(mdi.CorrelationMDIChild):
22 | """
23 | Shows the status of all correlations that are within the monitoring threshold
24 | """
25 |
26 | # The table and grid containing the status of correlations. Defined at instance level to enable refresh.
27 | __table = None
28 | __grid = None
29 |
30 | # Number of rows. Required for and updated by refresh method
31 | __rows = 0
32 |
33 | __log = None # The logger
34 |
35 | def __init__(self, parent):
36 | # Super
37 | wx.MDIChildFrame.__init__(self, parent=parent, id=wx.ID_ANY, title="Correlation Status",
38 | size=wx.Size(width=440, height=-1), style=wx.DEFAULT_FRAME_STYLE)
39 |
40 | # Create logger
41 | self.__log = logging.getLogger(__name__)
42 |
43 | # Panel and sizer for table
44 | panel = wx.Panel(self, wx.ID_ANY)
45 | sizer = wx.BoxSizer(wx.VERTICAL)
46 | panel.SetSizer(sizer)
47 |
48 | # Create the correlations grid. This is a data table using pandas dataframe for underlying data. Add the
49 | # correlations_grid to the correlations sizer.
50 | self.__table = _DataTable(columns=self.GetMDIParent().cor.filtered_coefficient_data.columns)
51 | self.__grid = wx.grid.Grid(panel, wx.ID_ANY)
52 | self.__grid.SetTable(self.__table, takeOwnership=True)
53 | self.__grid.EnableEditing(False)
54 | self.__grid.EnableDragRowSize(False)
55 | self.__grid.EnableDragColSize(True)
56 | self.__grid.EnableDragGridSize(True)
57 | self.__grid.SetSelectionMode(wx.grid.Grid.SelectRows)
58 | self.__grid.SetRowLabelSize(0)
59 | self.__grid.SetColSize(COLUMN_INDEX, 0) # Index. Hide
60 | self.__grid.SetColSize(COLUMN_SYMBOL1, 100) # Symbol 1
61 | self.__grid.SetColSize(COLUMN_SYMBOL2, 100) # Symbol 2
62 | self.__grid.SetColSize(COLUMN_BASE_COEFFICIENT, 100) # Base Coefficient
63 | self.__grid.SetColSize(COLUMN_DATE_FROM, 0) # UTC Date From. Hide
64 | self.__grid.SetColSize(COLUMN_DATE_TO, 0) # UTC Date To. Hide
65 | self.__grid.SetColSize(COLUMN_TIMEFRAME, 0) # Timeframe. Hide.
66 | self.__grid.SetColSize(COLUMN_LAST_CALCULATION, 0) # Last Calculation. Hide
67 | self.__grid.SetColSize(COLUMN_STATUS, 100) # Status
68 | self.__grid.SetMinSize((420, 500))
69 | sizer.Add(self.__grid, 1, wx.ALL | wx.EXPAND)
70 |
71 | # Bind row doubleclick
72 | self.Bind(wx.grid.EVT_GRID_CELL_LEFT_DCLICK, self.__on_doubleckick_row, self.__grid)
73 |
74 | # Refresh to populate
75 | self.refresh()
76 |
77 | def refresh(self):
78 | """
79 | Refreshes grid. Notifies if rows have been added or deleted.
80 | :return:
81 | """
82 | self.__log.debug(f"Refreshing grid.")
83 |
84 | # Update data
85 | self.__table.data = self.GetMDIParent().cor.filtered_coefficient_data.copy()
86 |
87 | # Format
88 | self.__table.data.loc[:, 'Base Coefficient'] = self.__table.data['Base Coefficient'].map('{:.5f}'.format)
89 | self.__table.data.loc[:, 'Last Calculation'] = pd.to_datetime(self.__table.data['Last Calculation'], utc=True)
90 | self.__table.data.loc[:, 'Last Calculation'] = \
91 | self.__table.data['Last Calculation'].dt.strftime('%d-%m-%y %H:%M:%S')
92 |
93 | # Start refresh
94 | self.__grid.BeginBatch()
95 |
96 | # Check if num rows in dataframe has changed, and send appropriate APPEND or DELETE messages
97 | cur_rows = len(self.GetMDIParent().cor.filtered_coefficient_data.index)
98 | if cur_rows < self.__rows:
99 | # Data has been deleted. Send message
100 | msg = wx.grid.GridTableMessage(self.__table, wx.grid.GRIDTABLE_NOTIFY_ROWS_DELETED,
101 | self.__rows - cur_rows, self.__rows - cur_rows)
102 | self.__grid.ProcessTableMessage(msg)
103 | elif cur_rows > self.__rows:
104 | # Data has been added. Send message
105 | msg = wx.grid.GridTableMessage(self.__table, wx.grid.GRIDTABLE_NOTIFY_ROWS_APPENDED,
106 | cur_rows - self.__rows) # how many
107 | self.__grid.ProcessTableMessage(msg)
108 |
109 | self.__grid.EndBatch()
110 |
111 | # Send updated message
112 | msg = wx.grid.GridTableMessage(self.__table, wx.grid.GRIDTABLE_REQUEST_VIEW_GET_VALUES)
113 | self.__grid.ProcessTableMessage(msg)
114 |
115 | # Update row count
116 | self.__rows = cur_rows
117 |
118 | def __on_doubleckick_row(self, evt):
119 | """
120 | Open the graphs when a row is doubleclicked.
121 | :param evt:
122 | :return:
123 | """
124 | row = evt.GetRow()
125 | symbol1 = self.__grid.GetCellValue(row, COLUMN_SYMBOL1)
126 | symbol2 = self.__grid.GetCellValue(row, COLUMN_SYMBOL2)
127 |
128 | mdi.FrameManager.open_frame(parent=self.GetMDIParent(),
129 | frame_module='mt5_correlation.gui.mdi_child_correlationgraph',
130 | frame_class='MDIChildCorrelationGraph',
131 | raise_if_open=True,
132 | symbols=[symbol1, symbol2])
133 |
134 |
135 | class _DataTable(wx.grid.GridTableBase):
136 | """
137 | A data table that holds data in a pandas dataframe. Contains highlighting rules for status.
138 | """
139 | data = None # The data for this table. A Pandas DataFrame
140 |
141 | def __init__(self, columns):
142 | wx.grid.GridTableBase.__init__(self)
143 | self.headerRows = 1
144 | self.data = pd.DataFrame(columns=columns)
145 |
146 | def GetNumberRows(self):
147 | return len(self.data)
148 |
149 | def GetNumberCols(self):
150 | return len(self.data.columns) + 1
151 |
152 | def GetValue(self, row, col):
153 | if row < self.RowsCount and col < self.ColsCount:
154 | return self.data.index[row] if col == 0 else self.data.iloc[row, col - 1]
155 | else:
156 | raise Exception(f"Trying to access row {row} and col {col} which does not exist.")
157 |
158 | def SetValue(self, row, col, value):
159 | self.data.iloc[row, col - 1] = value
160 |
161 | def GetColLabelValue(self, col):
162 | if col == 0:
163 | if self.data.index.name is None:
164 | return 'Index'
165 | else:
166 | return self.data.index.name
167 | return str(self.data.columns[col - 1])
168 |
169 | def GetTypeName(self, row, col):
170 | return wx.grid.GRID_VALUE_STRING
171 |
172 | def GetAttr(self, row, col, prop):
173 | attr = wx.grid.GridCellAttr()
174 |
175 | # Check that we are not out of bounds
176 | if row < self.RowsCount:
177 | # If column is status, check and highlight if diverging or converging.
178 | if col in [COLUMN_STATUS]:
179 | # Is status one of interest
180 | value = self.GetValue(row, col)
181 | if value != "":
182 | if value in [cor.STATUS_DIVERGING]:
183 | attr.SetBackgroundColour(wx.RED)
184 | elif value in [cor.STATUS_CONVERGING]:
185 | attr.SetBackgroundColour(wx.GREEN)
186 | else:
187 | attr.SetBackgroundColour(wx.WHITE)
188 |
189 | return attr
190 |
--------------------------------------------------------------------------------
/mt5_correlation/gui/mdi_child_divergedgraph.py:
--------------------------------------------------------------------------------
1 | import logging
2 | import matplotlib.dates
3 | import matplotlib.pyplot as plt
4 | import matplotlib.ticker as mticker
5 | import wx
6 | import wx.lib.scrolledpanel as scrolled
7 | import wxconfig as cfg
8 |
9 | from matplotlib.backends.backend_wxagg import FigureCanvasWxAgg as FigureCanvas
10 | from mpl_toolkits.axes_grid1 import host_subplot
11 | from mpl_toolkits import axisartist
12 |
13 | import mt5_correlation.gui.mdi as mdi
14 | from mt5_correlation import correlation as cor
15 |
16 |
17 | class MDIChildDivergedGraph(mdi.CorrelationMDIChild):
18 | """
19 | Shows the graphs for the specified correlation
20 | """
21 |
22 | symbol = None # Symbol to chart divergence for. Public as we use to check if window for the symbol is already open.
23 |
24 | # Logger
25 | __log = None
26 |
27 | # Date formats for graphs
28 | __tick_fmt_date = matplotlib.dates.DateFormatter('%d-%b')
29 | __tick_fmt_time = matplotlib.dates.DateFormatter('%H:%M:%S')
30 |
31 | # Graph Canvas
32 | __canvas = None
33 |
34 | # Colors for graph lines
35 | __colours = matplotlib.cm.get_cmap(cfg.Config().get("charts.colormap")).colors
36 |
37 | # We will store the other symbols last plotted. This will save us rebuilding teh figure if teh symbols haven't
38 | # changed.
39 | __other_symbols = None
40 |
41 | def __init__(self, parent, **kwargs):
42 | # Super
43 | wx.MDIChildFrame.__init__(self, parent=parent, id=wx.ID_ANY,
44 | title=f"Divergence Graph for {kwargs['symbol']}")
45 |
46 | # Create logger
47 | self.__log = logging.getLogger(__name__)
48 |
49 | # Store the symbol
50 | self.symbol = kwargs['symbol']
51 |
52 | # Create panel and sizer
53 | panel = scrolled.ScrolledPanel(self, wx.ID_ANY)
54 | sizer = wx.BoxSizer()
55 | panel.SetSizer(sizer)
56 |
57 | # Create figure and canvas. Add canvas to sizer
58 | self.__canvas = FigureCanvas(panel, wx.ID_ANY, plt.figure())
59 | panel.GetSizer().Add(self.__canvas, 1, wx.ALL | wx.EXPAND)
60 | panel.SetupScrolling()
61 |
62 | # Refresh to show content
63 | self.refresh()
64 |
65 | def refresh(self):
66 | """
67 | Refresh the graph
68 | :return:
69 | """
70 |
71 | # Get tick data for base symbol
72 | symbol_tick_data = self.GetMDIParent().cor.get_ticks(self.symbol, cache_only=True)
73 |
74 | # Get the other symbols and their tick data
75 | other_symbols_data = self.__get_other_symbols_data()
76 |
77 | # Delete all axes from the figure. They will need to be recreated as the symbols may be different
78 | for axes in self.__canvas.figure.axes:
79 | axes.remove()
80 |
81 | # Plot for all other symbols
82 | num_subplots = 1 if len(other_symbols_data.keys()) == 0 else len(other_symbols_data.keys())
83 | plotnum = 1
84 | axs = [] # Store the axes for sharing x axis
85 | for other_symbol in other_symbols_data:
86 | axs.append(self.__canvas.figure.add_subplot(num_subplots, 1, plotnum))
87 | self.__plot(axes=axs[-1], base_symbol=self.symbol, other_symbol=other_symbol,
88 | base_symbol_data=symbol_tick_data, other_symbol_data=other_symbols_data[other_symbol])
89 |
90 | # Next plot
91 | plotnum += 1
92 |
93 | # Share x axis of the last axes with all the others
94 | self.__share_xaxis(axs)
95 |
96 | # Redraw canvas
97 | self.__canvas.figure.tight_layout(pad=0.5)
98 | self.__canvas.draw()
99 |
100 | def __get_other_symbols_data(self):
101 | """
102 | Gets the data required for the graphs
103 | :param self:
104 | :return: dict of other symbols their tick data
105 | """
106 | # Get the symbols that this one has diverged against.
107 | data = self.GetMDIParent().cor.filtered_coefficient_data
108 | filtered_data = data.loc[(
109 | (data['Status'] == cor.STATUS_DIVERGED) |
110 | (data['Status'] == cor.STATUS_DIVERGING) |
111 | (data['Status'] == cor.STATUS_CONVERGING)
112 | ) &
113 | (
114 | (data['Symbol 1'] == self.symbol) |
115 | (data['Symbol 2'] == self.symbol)
116 | )]
117 |
118 | # Get all symbols and remove the base one. We will need to ensure that this is first in the list
119 | other_symbols = list(filtered_data['Symbol 1'].append(filtered_data['Symbol 2']).drop_duplicates())
120 | if self.symbol in other_symbols:
121 | other_symbols.remove(self.symbol)
122 |
123 | # Get the tick data other symbols and add to dict
124 | other_tick_data = {}
125 | for symbol in other_symbols:
126 | tick_data = self.GetMDIParent().cor.get_ticks(symbol, cache_only=True)
127 | other_tick_data[symbol] = tick_data
128 |
129 | return other_tick_data
130 |
131 | def __plot(self, axes, base_symbol, other_symbol, base_symbol_data, other_symbol_data):
132 | """
133 | Plots the data on the axes
134 | :param axes: The subplot to plot onto
135 | :param base_symbol
136 | :param other_symbol:
137 | """
138 | # Create the other axes. Will need an axes for the base symbol data and another for the other symbol data
139 | other_axes = axes.twinx()
140 |
141 | # Set plot title and axis labels
142 | axes.set_title(f"Tick Data for {base_symbol}:{other_symbol}")
143 | axes.set_ylabel(base_symbol, color=self.__colours[0], labelpad=10)
144 | other_axes.set_ylabel(other_symbol, color=self.__colours[1], labelpad=10)
145 |
146 | # Set the tick and axis colors
147 | self.__set_axes_color(axes, self.__colours[0], 'left')
148 | self.__set_axes_color(other_axes, self.__colours[1], 'right')
149 |
150 | # Plot both lines
151 | axes.plot(base_symbol_data['time'], base_symbol_data['ask'], color=self.__colours[0])
152 | other_axes.plot(other_symbol_data['time'], other_symbol_data['ask'], color=self.__colours[1])
153 |
154 | @staticmethod
155 | def __set_axes_color(axes, color, axis_loc='right'):
156 | """
157 | Set the color for the axes, including axis line, ticks and tick labels
158 | :param axes: The axes to set color for.
159 | :param color: The color to set to
160 | :param axis_loc: The location of the axis, label and ticks. Either left for base symbol or right for others
161 | :return:
162 | """
163 | # Change the color for the axis line, ticks and tick labels
164 | axes.spines[axis_loc].set_color(color)
165 | axes.tick_params(axis='y', colors=color)
166 |
167 | def __share_xaxis(self, axs):
168 | """
169 | Share the xaxis of the last axes with all other axes. Remove axis tick labels for all but the last. Format axis
170 | tick labels for the last.
171 | :param axs:
172 | :return:
173 | """
174 | if len(axs) > 0:
175 | last_ax = axs[-1]
176 | for ax in axs:
177 | # If we are not on the last one, share and hide tick labels. If we are on the last one, format tick
178 | # labels.
179 | if ax != last_ax:
180 | ax.sharex(last_ax)
181 | plt.setp(ax.xaxis.get_majorticklabels(), visible=False)
182 | else:
183 | # Ticks, labels and formats. Fixing xticks with FixedLocator but also using MaxNLocator to avoid
184 | # cramped x-labels
185 | ax.xaxis.set_major_locator(mticker.MaxNLocator(10))
186 | ticks_loc = ax.get_xticks().tolist()
187 | ax.xaxis.set_major_locator(mticker.FixedLocator(ticks_loc))
188 | ax.set_xticklabels(ticks_loc)
189 | ax.xaxis.set_major_formatter(self.__tick_fmt_time)
190 | plt.setp(ax.xaxis.get_majorticklabels(), rotation=45)
191 |
192 | def __del__(self):
193 | # Close all plots
194 | plt.close('all')
195 |
--------------------------------------------------------------------------------
/configmeta.yaml:
--------------------------------------------------------------------------------
1 | ---
2 | calculate:
3 | from:
4 | days:
5 | __label: Calculate from (days)
6 | __helptext: The number of days of data to be used to calculate the coefficient.
7 | timeframe:
8 | __label: Timeframe
9 | __helptext: The timeframe for price candles to use for the calculation. Possible values are 1=1 Minute candles; 2=2 Minute candles; 3=3 Minute candles; 4=4 Minute candles; 5=5 Minute candles; 6=6 Minute candles; 10=10 Minute candles; 15=15 Minute candles; 20=20 Minute candles; 30=30 Minute candles; 16385=1 Hour candles; 16386=2 Hour candles; 16387=3 Hour candles; 16388=4 Hour candles; 16390=6 Hour candles; 16392=8 Hour candles; 16396=12 Hour candles; 16408=1 Day candles; 32769=1 Week candles; or 49153=1 Month candles.
10 | min_prices:
11 | __label: Min Prices
12 | __helptext: The minimum number of candles required to calculate a coefficient from. If any of the symbols do not have at least this number of candles then the coefficient won't be calculated.
13 | max_set_size_diff_pct:
14 | __label: Min Set Size Difference %
15 | __helptext: For a meaningful coefficient calculation, the two sets of data should be of a similar size. This setting specifies the % difference allowed for a correlation to be calculated. The smallest set of candle data must be at least this values % of the largest set.
16 | overlap_pct:
17 | __label: Overlap %
18 | __helptext: The dates and times in the two sets of data must match. The coefficient will only be calculated against the dates that overlap. Any non overlapping dates will be discarded. This setting specifies the minimum size of the overlapping data when compared to the smallest set as a %. A coefficient will not be calculated if this threshold is not met.
19 | max_p_value:
20 | __label: Max Pearsonr P Value
21 | __helptext: The maximum P value for the coefficient to be considered valid. A full explanation on the correlation coefficient P value is available in the scipy pearsonr documentation.
22 | monitor:
23 | interval:
24 | __label: Monitoring Interval
25 | __helptext: The number of seconds between monitoring events.
26 | calculations:
27 | long:
28 | __label: Long
29 | __helptext: The settings for the correlation calculation using the longest timeframe.
30 | from:
31 | __label: From (Minutes)
32 | __helptext: The number of minutes of data to be used to calculate the long coefficient.
33 | min_prices:
34 | __label: Min Prices
35 | __helptext: Tick data will be converted to 1 second candles prior to calculation. This will enable data to be matched between symbols. This setting specifies the minimum number of price candles required to calculate a coefficient from. If any of the symbols do not have at least number of candles then the coefficient won't be calculated.
36 | max_set_size_diff_pct:
37 | __label: Max Set Size Difference %
38 | __helptext: For a meaningful coefficient calculation, the two sets of data should be of a similar size. This setting specifies the % difference allowed for a correlation to be calculated. The smallest set of price candles must be at least this values % of the largest set.
39 | overlap_pct:
40 | __label: Overlap %
41 | __helptext: The dates and times in the two sets of data must match. The ticks will be converted to 1 second price candles before calculation. The coefficient will only be calculated against the times from the candles that overlap. Any non overlapping times will be discarded. This setting specifies the minimum size of the overlapping data when compared to the smallest set as a %. A coefficient will not be calculated if this threshold is not met.
42 | max_p_value:
43 | __label: Max Pearsonr P Value
44 | __helptext: The maximum P value for the coefficient to be considered valid. A full explanation on the correlation coefficient P value is available in the scipy pearsonr documentation.
45 | medium:
46 | __label: Medium
47 | __helptext: The settings for the correlation calculation using a medium timeframe.
48 | from:
49 | __label: From (Minutes)
50 | __helptext: The number of minutes of data to be used to calculate the medium coefficient.
51 | min_prices:
52 | __label: Min Prices
53 | __helptext: Tick data will be converted to 1 second candles prior to calculation. This will enable data to be matched between symbols. This setting specifies the minimum number of price candles required to calculate a coefficient from. If any of the symbols do not have at least number of candles then the coefficient won't be calculated.
54 | max_set_size_diff_pct:
55 | __label: Max Set Size Difference %
56 | __helptext: For a meaningful coefficient calculation, the two sets of data should be of a similar size. This setting specifies the % difference allowed for a correlation to be calculated. The smallest set of price candles must be at least this values % of the largest set.
57 | overlap_pct:
58 | __label: Overlap %
59 | __helptext: The dates and times in the two sets of data must match. The ticks will be converted to 1 second price candles before calculation. The coefficient will only be calculated against the times from the candles that overlap. Any non overlapping times will be discarded. This setting specifies the minimum size of the overlapping data when compared to the smallest set as a %. A coefficient will not be calculated if this threshold is not met.
60 | max_p_value:
61 | __label: Max Pearsonr P Value
62 | __helptext: The maximum P value for the coefficient to be considered valid. A full explanation on the correlation coefficient P value is available in the scipy pearsonr documentation.
63 | short:
64 | __label: Short
65 | __helptext: The settings for the correlation calculation using the shortest timeframe.
66 | from:
67 | __label: From (Minutes)
68 | __helptext: The number of minutes of data to be used to calculate the short coefficient.
69 | min_prices:
70 | __label: Min Prices
71 | __helptext: Tick data will be converted to 1 second candles prior to calculation. This will enable data to be matched between symbols. This setting specifies the minimum number of price candles required to calculate a coefficient from. If any of the symbols do not have at least number of candles then the coefficient won't be calculated.
72 | max_set_size_diff_pct:
73 | __label: Max Set Size Difference %
74 | __helptext: For a meaningful coefficient calculation, the two sets of data should be of a similar size. This setting specifies the % difference allowed for a correlation to be calculated. The smallest set of price candles must be at least this values % of the largest set.
75 | overlap_pct:
76 | __label: Overlap %
77 | __helptext: The dates and times in the two sets of data must match. The ticks will be converted to 1 second price candles before calculation. The coefficient will only be calculated against the times from the candles that overlap. Any non overlapping times will be discarded. This setting specifies the minimum size of the overlapping data when compared to the smallest set as a %. A coefficient will not be calculated if this threshold is not met.
78 | max_p_value:
79 | __label: Max Pearsonr P Value
80 | __helptext: The maximum P value for the coefficient to be considered valid. A full explanation on the correlation coefficient P value is available in the scipy pearsonr documentation.
81 | monitoring_threshold:
82 | __label: Monitoring Threshold
83 | __helptext: Only pairs with a coefficient over this threshold will be displayed and monitored.
84 | divergence_threshold:
85 | __label: Divergence Threshold
86 | __helptext: The application will consider a pair to have diverged if the correlation coefficient for all timeframes (long, medium and short) falls below this threshold.
87 | monitor_inverse:
88 | __label: Monitor Inverse
89 | __helptext: Monitor Inverse Correlations (uses negative scale with -1 being fully inversly correlated)
90 | tick_cache_time:
91 | __label: Tick Cache Time
92 | __helptext: Every calculation requires tick data for both symbols. Tick data will be cached for this number of seconds before being retrieved from MetaTrader. Some caching is recommended as a single monitoring run will request the same data for symbols that form multiple correlated pairs.
93 | autosave:
94 | __label: Auto Save
95 | __helptext: Whether to auto save after every monitoring event. If a file was opened or has been saved, then the data will be saved to this file, otherwise the data will be saved to a file named autosave.cpd.
96 | charts:
97 | colormap:
98 | __label: Color Map
99 | __helptext: The matplotlib color pallet to use for plotting graphs. A list of pallets is available at https://matplotlib.org/stable/tutorials/colors/colormaps.html
100 | developer:
101 | inspection:
102 | __label: Inspection
103 | __helptext: Provide GUI Inspection guidelines for developers modifying the GUI.
104 | ...
--------------------------------------------------------------------------------
/mt5_correlation/gui/mdi_child_correlationgraph.py:
--------------------------------------------------------------------------------
1 | import logging
2 | import matplotlib.dates
3 | import matplotlib.pyplot as plt
4 | import matplotlib.ticker as mticker
5 | import wx
6 | import wxconfig as cfg
7 | import wx.lib.scrolledpanel as scrolled
8 |
9 | from matplotlib.backends.backend_wxagg import FigureCanvasWxAgg as FigureCanvas
10 |
11 | import mt5_correlation.gui.mdi as mdi
12 |
13 |
14 | class MDIChildCorrelationGraph(mdi.CorrelationMDIChild):
15 | """
16 | Shows the graphs for the specified correlation
17 | """
18 |
19 | symbols = None # Symbols for correlation. Public as we use to check if window for the symbol pair is already open.
20 |
21 | # Date formats for graphs
22 | __tick_fmt_date = matplotlib.dates.DateFormatter('%d-%b')
23 | __tick_fmt_time = matplotlib.dates.DateFormatter('%H:%M:%S')
24 |
25 | # Colors for graph lines for symbol1 and symbol2. Will use first 2 colours in colormap
26 | __colours = matplotlib.cm.get_cmap(cfg.Config().get("charts.colormap")).colors
27 |
28 | # Fig, axes and canvas
29 | __fig = None
30 | __axs = None
31 | __canvas = None
32 |
33 | def __init__(self, parent, **kwargs):
34 | # Super
35 | wx.MDIChildFrame.__init__(self, parent=parent, id=wx.ID_ANY,
36 | title=f"Correlation Status for {kwargs['symbols'][0]}:{kwargs['symbols'][1]}")
37 |
38 | # Create logger
39 | self.__log = logging.getLogger(__name__)
40 |
41 | # Store the symbols
42 | self.symbols = kwargs['symbols']
43 |
44 | # We will freeze this frame and thaw once constructed to avoid flicker.
45 | self.Freeze()
46 |
47 | # Draw the empty graphs. We will populate with data in refresh. We will have 3 charts:
48 | # 1) Data used to calculate base coefficient for both symbols (2 lines on chart);
49 | # 2) Data used to calculate latest coefficient for both symbols (2 lines on chart); and
50 | # 3) Coefficient history and the divergence threshold lines
51 |
52 | # Create fig and 3 axes.
53 | self.__fig, self.__axs = plt.subplots(3)
54 |
55 | # Create additional axis for second line on charts 1 & 2
56 | self.__s2axs = [self.__axs[0].twinx(), self.__axs[1].twinx()]
57 |
58 | # Set titles
59 | self.__axs[0].set_title(f"Base Coefficient Price Data for {self.symbols[0]}:{self.symbols[1]}")
60 | self.__axs[1].set_title(f"Coefficient Tick Data for {self.symbols[0]}:{self.symbols[1]}")
61 | self.__axs[2].set_title(f"Coefficient History for {self.symbols[0]}:{self.symbols[1]}")
62 |
63 | # Set Y Labels and tick colours for charts 1 & 2. Left for symbol1, right for symbol2
64 | for i in range(0, 2):
65 | self.__axs[i].set_ylabel(f"{self.symbols[0]}", color=self.__colours[0], labelpad=10)
66 | self.__axs[i].tick_params(axis='y', labelcolor=self.__colours[0])
67 | self.__s2axs[i].set_ylabel(f"{self.symbols[1]}", color=self.__colours[1], labelpad=10)
68 | self.__s2axs[i].tick_params(axis='y', labelcolor=self.__colours[1])
69 |
70 | # Set Y label and limits for 3rd chart. Limits will be coefficients range from -1 to 1
71 | self.__axs[2].set_ylabel('Coefficient', labelpad=10)
72 | self.__axs[2].set_ylim([-1, 1])
73 |
74 | # Set X labels to ''. Workaround as matplotlib is not leaving space for ticks
75 | for ax in self.__axs:
76 | ax.set_xlabel(" ", labelpad=10)
77 |
78 | # Layout with padding between charts
79 | self.__fig.tight_layout(pad=0.5)
80 |
81 | # Create panel and sizer. This will provide scrollbar
82 | panel = scrolled.ScrolledPanel(self, wx.ID_ANY)
83 | sizer = wx.BoxSizer()
84 | panel.SetSizer(sizer)
85 |
86 | # Add fig to canvas and canvas to sizer. Thaw window to update
87 | self.__canvas = FigureCanvas(panel, wx.ID_ANY, self.__fig)
88 | sizer.Add(self.__canvas, 1, wx.ALL | wx.EXPAND)
89 | self.Thaw()
90 |
91 | # Setup scrolling
92 | panel.SetupScrolling()
93 |
94 | # Refresh to show content
95 | self.refresh()
96 |
97 | def refresh(self):
98 | """
99 | Refresh the graph
100 | :return:
101 | """
102 | # Get the price data for the base coefficient calculation, tick data that was used to calculate last
103 | # coefficient and and the coefficient history data
104 | price_data = [self.GetMDIParent().cor.get_price_data(self.symbols[0]),
105 | self.GetMDIParent().cor.get_price_data(self.symbols[1])]
106 |
107 | tick_data = [self.GetMDIParent().cor.get_ticks(self.symbols[0], cache_only=True),
108 | self.GetMDIParent().cor.get_ticks(self.symbols[1], cache_only=True)]
109 |
110 | history_data = []
111 | for timeframe in cfg.Config().get('monitor.calculations'):
112 | frm = cfg.Config().get(f'monitor.calculations.{timeframe}.from')
113 | history_data.append(self.GetMDIParent().cor.get_coefficient_history(
114 | {'Symbol 1': self.symbols[0], 'Symbol 2': self.symbols[1], 'Timeframe': frm}))
115 |
116 | # Check what data we have available
117 | price_data_available = price_data is not None and len(price_data) == 2 and price_data[0] is not None and \
118 | price_data[1] is not None and len(price_data[0]) > 0 and len(price_data[1]) > 0
119 |
120 | tick_data_available = tick_data is not None and len(tick_data) == 2 and tick_data[0] is not None and \
121 | tick_data[1] is not None and len(tick_data[0]) > 0 and len(tick_data[1]) > 0
122 |
123 | history_data_available = history_data is not None and len(history_data) > 0
124 |
125 | # Get all plots for coefficient history. History can contain multiple plots for different timeframes. They
126 | # will all be plotted on the same chart.
127 | times = []
128 | coefficients = []
129 | if history_data_available:
130 | for hist in history_data:
131 | times.append(hist['Date To'])
132 | coefficients.append(hist['Coefficient'])
133 |
134 | # Update graphs where we have data available
135 | if price_data_available:
136 | # Update range and ticks
137 | xrange = [min(min(price_data[0]['time']), min(price_data[1]['time'])),
138 | max(max(price_data[0]['time']), max(price_data[1]['time']))]
139 | self.__axs[0].set_xlim(xrange)
140 |
141 | # Plot both lines
142 | self.__axs[0].plot(price_data[0]['time'], price_data[0]['close'],
143 | color=self.__colours[0])
144 | self.__s2axs[0].plot(price_data[1]['time'], price_data[1]['close'],
145 | color=self.__colours[1])
146 |
147 | # Ticks, labels and formats. Fixing xticks with FixedLocator but also using MaxNLocator to avoid
148 | # cramped x-labels
149 | if len(price_data[0]['time']) > 0:
150 | self.__axs[0].xaxis.set_major_locator(mticker.MaxNLocator(10))
151 | ticks_loc = self.__axs[0].get_xticks().tolist()
152 | self.__axs[0].xaxis.set_major_locator(mticker.FixedLocator(ticks_loc))
153 | self.__axs[0].set_xticklabels(ticks_loc)
154 | self.__axs[0].xaxis.set_major_formatter(self.__tick_fmt_date)
155 | plt.setp(self.__axs[0].xaxis.get_majorticklabels(), rotation=45)
156 |
157 | if tick_data_available:
158 | # Update range and ticks
159 | xrange = [min(min(tick_data[0]['time']), min(tick_data[1]['time'])),
160 | max(max(tick_data[0]['time']), max(tick_data[1]['time']))]
161 | self.__axs[1].set_xlim(xrange)
162 |
163 | # Plot both lines
164 | self.__axs[1].plot(tick_data[0]['time'], tick_data[0]['ask'],
165 | color=self.__colours[0])
166 | self.__s2axs[1].plot(tick_data[1]['time'], tick_data[1]['ask'],
167 | color=self.__colours[1])
168 |
169 | if len(tick_data[0]['time']) > 0:
170 | self.__axs[1].xaxis.set_major_locator(mticker.MaxNLocator(10))
171 | ticks_loc = self.__axs[1].get_xticks().tolist()
172 | self.__axs[1].xaxis.set_major_locator(mticker.FixedLocator(ticks_loc))
173 | self.__axs[1].set_xticklabels(ticks_loc)
174 | self.__axs[1].xaxis.set_major_formatter(self.__tick_fmt_time)
175 | plt.setp(self.__axs[1].xaxis.get_majorticklabels(), rotation=45)
176 |
177 | if history_data_available:
178 | # Plot. There may be more than one set of data for chart. One for each coefficient date range. Convert
179 | # single data to list, then loop to plot
180 | xdata = times if isinstance(times, list) else [times, ]
181 | ydata = coefficients if isinstance(coefficients, list) else [coefficients, ]
182 |
183 | for i in range(0, len(xdata)):
184 | self.__axs[2].scatter(xdata[i], ydata[i], s=1)
185 |
186 | # Ticks, labels and formats. Fixing xticks with FixedLocator but also using MaxNLocator to avoid
187 | # cramped x-labels
188 | if len(times[0].array) > 0:
189 | self.__axs[2].xaxis.set_major_locator(mticker.MaxNLocator(10))
190 | ticks_loc = self.__axs[2].get_xticks().tolist()
191 | self.__axs[2].xaxis.set_major_locator(mticker.FixedLocator(ticks_loc))
192 | self.__axs[2].set_xticklabels(ticks_loc)
193 | self.__axs[2].xaxis.set_major_formatter(self.__tick_fmt_time)
194 | plt.setp(self.__axs[2].xaxis.get_majorticklabels(), rotation=45)
195 |
196 | # Legend
197 | self.__axs[2].legend([f"{cfg.Config().get('monitor.calculations.long.from')} Minutes",
198 | f"{cfg.Config().get('monitor.calculations.medium.from')} Minutes",
199 | f"{cfg.Config().get('monitor.calculations.short.from')} Minutes"])
200 |
201 | # Lines showing divergence threshold. 2 if we are monitoring inverse correlations.
202 | divergence_threshold = self.GetMDIParent().cor.divergence_threshold
203 | monitor_inverse = self.GetMDIParent().cor.monitor_inverse
204 |
205 | if divergence_threshold is not None:
206 | self.__axs[2].axhline(y=divergence_threshold, color="red", label='_nolegend_', linewidth=1)
207 | if monitor_inverse:
208 | self.__axs[2].axhline(y=divergence_threshold * -1, color="red", label='_nolegend_', linewidth=1)
209 |
210 | # Redraw canvas
211 | self.__canvas.draw()
212 |
213 | def __del__(self):
214 | # Close all plots
215 | plt.close('all')
216 |
--------------------------------------------------------------------------------
/mt5_correlation/gui/mdi.py:
--------------------------------------------------------------------------------
1 | import abc
2 | import importlib
3 | import logging
4 |
5 | import pytz
6 | import wx
7 | import wx.lib.inspection as ins
8 | import wxconfig
9 | import wxconfig as cfg
10 |
11 | from datetime import datetime, timedelta
12 |
13 | from mt5_correlation import correlation as cor
14 |
15 |
16 | class CorrelationMDIFrame(wx.MDIParentFrame):
17 | """
18 | The MDI Frame window for the correlation monitoring application
19 | """
20 | # The correlation instance that calculates coefficients and monitors for divergence. Needs to be accessible to
21 | # child frames.
22 | cor = None
23 |
24 | __opened_filename = None # So we can save to same file as we opened
25 | __log = None # The logger
26 | __menu_item_monitor = None # We need to store this menu item so that we can check if it is checked or not.
27 |
28 | def __init__(self):
29 | # Super
30 | wx.MDIParentFrame.__init__(self, parent=None, id=wx.ID_ANY, title="Divergence Monitor",
31 | pos=wx.Point(x=cfg.Config().get('window.x'), y=cfg.Config().get('window.y')),
32 | size=wx.Size(width=cfg.Config().get('window.width'),
33 | height=cfg.Config().get('window.height')),
34 | style=cfg.Config().get('window.style'))
35 |
36 | # Create logger
37 | self.__log = logging.getLogger(__name__)
38 |
39 | # Create correlation instance to maintain state of calculated coefficients. Set params from config
40 | self.cor = cor.Correlation(monitoring_threshold=cfg.Config().get("monitor.monitoring_threshold"),
41 | divergence_threshold=cfg.Config().get("monitor.divergence_threshold"),
42 | monitor_inverse=cfg.Config().get("monitor.monitor_inverse"))
43 |
44 | # Status bar. 2 fields, one for monitoring status and one for general status. On open, monitoring status is not
45 | # monitoring. SetBackgroundColour will change colour of both. Couldn't find a way to set on single field only.
46 | self.__statusbar = self.CreateStatusBar(2)
47 | self.__statusbar.SetStatusWidths([100, -1])
48 | self.SetStatusText("Not Monitoring", 0)
49 |
50 | # Create menu bar and bind menu items to methods
51 | self.menubar = wx.MenuBar()
52 |
53 | # File menu and items
54 | menu_file = wx.Menu()
55 | self.Bind(wx.EVT_MENU, self.__on_open_file, menu_file.Append(wx.ID_ANY, "&Open", "Open correlations file."))
56 | self.Bind(wx.EVT_MENU, self.__on_save_file, menu_file.Append(wx.ID_ANY, "Save", "Save correlations file."))
57 | self.Bind(wx.EVT_MENU, self.__on_save_file_as,
58 | menu_file.Append(wx.ID_ANY, "Save As", "Save correlations file."))
59 | menu_file.AppendSeparator()
60 | self.Bind(wx.EVT_MENU, self.__on_open_settings,
61 | menu_file.Append(wx.ID_ANY, "Settings", "Change application settings."))
62 | menu_file.AppendSeparator()
63 | self.Bind(wx.EVT_MENU, self.__on_exit, menu_file.Append(wx.ID_ANY, "Exit", "Close the application"))
64 | self.menubar.Append(menu_file, "&File")
65 |
66 | # Coefficient menu and items
67 | menu_coef = wx.Menu()
68 | self.Bind(wx.EVT_MENU, self.__on_calculate,
69 | menu_coef.Append(wx.ID_ANY, "Calculate", "Calculate base coefficients."))
70 | self.__menu_item_monitor = menu_coef.Append(wx.ID_ANY, "Monitor",
71 | "Monitor correlated pairs for changes to coefficient.",
72 | kind=wx.ITEM_CHECK)
73 | self.Bind(wx.EVT_MENU, self.__on_monitor, self.__menu_item_monitor)
74 | menu_coef.AppendSeparator()
75 | self.Bind(wx.EVT_MENU, self.__on_clear,
76 | menu_coef.Append(wx.ID_ANY, "Clear", "Clear coefficient and price history."))
77 | self.menubar.Append(menu_coef, "Coefficient")
78 |
79 | # View menu and items
80 | menu_view = wx.Menu()
81 | self.Bind(wx.EVT_MENU, self.__on_view_status, menu_view.Append(wx.ID_ANY, "Status",
82 | "View status of correlations."))
83 | self.Bind(wx.EVT_MENU, self.__on_view_diverged, menu_view.Append(wx.ID_ANY, "Diverged Symbols",
84 | "View diverged symbols."))
85 | self.menubar.Append(menu_view, "&View")
86 |
87 | # Help menu and items
88 | help_menu = wx.Menu()
89 | self.Bind(wx.EVT_MENU, self.__on_view_log, help_menu.Append(wx.ID_ANY, "View Log", "Show application log."))
90 | self.Bind(wx.EVT_MENU, self.__on_view_help, help_menu.Append(wx.ID_ANY, "Help",
91 | "Show application usage instructions."))
92 |
93 | self.menubar.Append(help_menu, "&Help")
94 |
95 | # Set menu bar
96 | self.SetMenuBar(self.menubar)
97 |
98 | # Set up timer to refresh
99 | self.timer = wx.Timer(self)
100 | self.Bind(wx.EVT_TIMER, self.__on_timer, self.timer)
101 |
102 | # Bind window close event
103 | self.Bind(wx.EVT_CLOSE, self.__on_close, self)
104 |
105 | def __on_close(self, event):
106 | """
107 | Window closing. Save coefficients and stop monitoring.
108 | :param event:
109 | :return:
110 | """
111 | # Save pos and size
112 | x, y = self.GetPosition()
113 | width, height = self.GetSize()
114 | cfg.Config().set('window.x', x)
115 | cfg.Config().set('window.y', y)
116 | cfg.Config().set('window.width', width)
117 | cfg.Config().set('window.height', height)
118 |
119 | # Style
120 | style = self.GetWindowStyle()
121 | cfg.Config().set('window.style', style)
122 |
123 | cfg.Config().save()
124 |
125 | # Stop monitoring
126 | self.cor.stop_monitor()
127 |
128 | # End
129 | event.Skip()
130 |
131 | def __on_open_file(self, evt):
132 | with wx.FileDialog(self, "Open Coefficients file", wildcard="cpd (*.cpd)|*.cpd",
133 | style=wx.FD_OPEN | wx.FD_FILE_MUST_EXIST) as fileDialog:
134 | if fileDialog.ShowModal() == wx.ID_CANCEL:
135 | return # the user changed their mind
136 |
137 | # Load the file chosen by the user.
138 | self.__opened_filename = fileDialog.GetPath()
139 |
140 | self.SetStatusText(f"Loading file {self.__opened_filename}.", 1)
141 | self.cor.load(self.__opened_filename)
142 |
143 | # Show calculated data and refresh all opened frames
144 | self.__on_view_status(evt)
145 | self.__refresh()
146 |
147 | self.SetStatusText(f"File {self.__opened_filename} loaded.", 1)
148 |
149 | def __on_save_file(self, evt):
150 | self.SetStatusText(f"Saving file as {self.__opened_filename}", 1)
151 |
152 | if self.__opened_filename is None:
153 | self.__on_save_file_as(evt)
154 | else:
155 | self.cor.save(self.__opened_filename)
156 |
157 | self.SetStatusText(f"File saved as {self.__opened_filename}", 1)
158 |
159 | def __on_save_file_as(self, evt):
160 | with wx.FileDialog(self, "Save Coefficients file", wildcard="cpd (*.cpd)|*.cpd",
161 | style=wx.FD_SAVE) as fileDialog:
162 | if fileDialog.ShowModal() == wx.ID_CANCEL:
163 | return # the user changed their mind
164 |
165 | # Save the file and price data file, changing opened filename so next save writes to new file
166 | self.SetStatusText(f"Saving file as {self.__opened_filename}", 1)
167 |
168 | self.__opened_filename = fileDialog.GetPath()
169 | self.cor.save(self.__opened_filename)
170 |
171 | self.SetStatusText(f"File saved as {self.__opened_filename}", 1)
172 |
173 | def __on_open_settings(self, evt):
174 | settings_dialog = cfg.SettingsDialog(parent=self, exclude=['window'])
175 | res = settings_dialog.ShowModal()
176 | if res == wx.ID_OK:
177 | # Stop the monitor
178 | self.cor.stop_monitor()
179 |
180 | # Build calculation params and restart the monitor
181 | calculation_params = [cfg.Config().get('monitor.calculations.long'),
182 | cfg.Config().get('monitor.calculations.medium'),
183 | cfg.Config().get('monitor.calculations.short')]
184 |
185 | self.cor.start_monitor(interval=cfg.Config().get('monitor.interval'),
186 | calculation_params=calculation_params,
187 | cache_time=cfg.Config().get('monitor.tick_cache_time'),
188 | autosave=cfg.Config().get('monitor.autosave'),
189 | filename=self.__opened_filename)
190 |
191 | # Refresh all open child frames
192 | self.__refresh()
193 |
194 | def __on_exit(self, evt):
195 | # Close
196 | self.Close()
197 |
198 | def __on_calculate(self, evt):
199 | # set time zone to UTC to avoid local offset issues, and get from and to dates (a week ago to today)
200 | timezone = pytz.timezone("Etc/UTC")
201 | utc_to = datetime.now(tz=timezone)
202 | utc_from = utc_to - timedelta(days=cfg.Config().get('calculate.from.days'))
203 |
204 | # Calculate
205 | self.SetStatusText("Calculating coefficients.", 1)
206 | self.cor.calculate(date_from=utc_from, date_to=utc_to,
207 | timeframe=cfg.Config().get('calculate.timeframe'),
208 | min_prices=cfg.Config().get('calculate.min_prices'),
209 | max_set_size_diff_pct=cfg.Config().get('calculate.max_set_size_diff_pct'),
210 | overlap_pct=cfg.Config().get('calculate.overlap_pct'),
211 | max_p_value=cfg.Config().get('calculate.max_p_value'))
212 | self.SetStatusText("", 1)
213 |
214 | # Show calculated data and refresh frames
215 | self.__on_view_status(evt)
216 | self.__refresh()
217 |
218 | def __on_monitor(self, evt):
219 | # Check state of toggle menu. If on, then start monitoring, else stop
220 | if self.__menu_item_monitor.IsChecked():
221 | self.__log.info("Starting monitoring for changes to coefficients.")
222 | self.SetStatusText("Monitoring", 0)
223 | self.__statusbar.SetBackgroundColour('green')
224 | self.__statusbar.Refresh()
225 |
226 | self.timer.Start(cfg.Config().get('monitor.interval') * 1000)
227 |
228 | # Autosave filename
229 | filename = self.__opened_filename if self.__opened_filename is not None else 'autosave.cpd'
230 |
231 | # Build calculation params and start monitor
232 | calculation_params = [cfg.Config().get('monitor.calculations.long'),
233 | cfg.Config().get('monitor.calculations.medium'),
234 | cfg.Config().get('monitor.calculations.short')]
235 |
236 | self.cor.start_monitor(interval=cfg.Config().get('monitor.interval'),
237 | calculation_params=calculation_params,
238 | cache_time=cfg.Config().get('monitor.tick_cache_time'),
239 | autosave=cfg.Config().get('monitor.autosave'),
240 | filename=filename)
241 | else:
242 | self.__log.info("Stopping monitoring.")
243 | self.SetStatusText("Not Monitoring", 0)
244 | self.__statusbar.SetBackgroundColour('lightgray')
245 | self.__statusbar.Refresh()
246 | self.timer.Stop()
247 | self.cor.stop_monitor()
248 |
249 | def __on_clear(self, evt):
250 | # Clear the history
251 | self.cor.clear_coefficient_history()
252 |
253 | # Refresh opened child frames
254 | self.__refresh()
255 |
256 | def __on_timer(self, evt):
257 | # Refresh opened child frames
258 | self.__refresh()
259 |
260 | # Set status message
261 | self.SetStatusText(f"Status updated at {self.cor.get_last_calculation():%d-%b %H:%M:%S}.", 1)
262 |
263 | def __on_view_status(self, evt):
264 | FrameManager.open_frame(parent=self, frame_module='mt5_correlation.gui.mdi_child_status',
265 | frame_class='MDIChildStatus',
266 | raise_if_open=True)
267 |
268 | def __on_view_diverged(self, evt):
269 | FrameManager.open_frame(parent=self, frame_module='mt5_correlation.gui.mdi_child_diverged_symbols',
270 | frame_class='MDIChildDivergedSymbols',
271 | raise_if_open=True)
272 |
273 | def __on_view_log(self, evt):
274 | FrameManager.open_frame(parent=self, frame_module='mt5_correlation.gui.mdi_child_log',
275 | frame_class='MDIChildLog',
276 | raise_if_open=True)
277 |
278 | def __on_view_help(self, evt):
279 | FrameManager.open_frame(parent=self, frame_module='mt5_correlation.gui.mdi_child_help',
280 | frame_class='MDIChildHelp',
281 | raise_if_open=True)
282 |
283 | def __refresh(self):
284 | """
285 | Refresh all open child frames
286 | :return:
287 | """
288 | children = self.GetChildren()
289 |
290 | for child in children:
291 | if isinstance(child, CorrelationMDIChild):
292 | child.refresh()
293 | elif isinstance(child, wx.StatusBar) or isinstance(child, ins.InspectionFrame) or \
294 | isinstance(child, wxconfig.SettingsDialog):
295 | # Ignore
296 | pass
297 | else:
298 | raise Exception(f"MDI Child for application must implement CorrelationMDIChild. MDI Child is "
299 | f"{type(child)}.")
300 |
301 |
302 | class CorrelationMDIChild(wx.MDIChildFrame):
303 | """
304 | Interface for all MDI Children supported by the MDIParent
305 | """
306 |
307 | @abc.abstractmethod
308 | def refresh(self):
309 | """
310 | Must be implemented. Refreshes the content. Called by MDIParents __refresh method
311 | :return:
312 | """
313 | raise NotImplementedError
314 |
315 |
316 | class FrameManager:
317 | """
318 | Manages the opening and raising of MDIChild frames
319 | """
320 | @staticmethod
321 | def open_frame(parent, frame_module, frame_class, raise_if_open=True, **kwargs):
322 | """
323 | Opens the frame specified by the frame class
324 | :param parent: The MDIParentFrame to open the child frame into
325 | :param frame_module: A string specifying the module containing the frame class to open or raise
326 | :param frame_class: A string specifying the frame class to open or raise
327 | :param raise_if_open: Whether the frame should raise rather than open if an instance is already open.
328 | :param kwargs: A dict of parameters to pass to frame constructor. These will also be checked in raise_if_open
329 | to determine uniqueness (i.e. If a frame of the same class is already open but its params are different,
330 | then the frame will be opened again with the new params instead of being raised.)
331 | :return:
332 | """
333 |
334 | # Load the module and class
335 | module = importlib.import_module(frame_module)
336 | clazz = getattr(module, frame_class)
337 |
338 | # Do we have an opened instance
339 | opened_instance = None
340 | for child in parent.GetChildren():
341 | if isinstance(child, clazz):
342 | # do the args match
343 | match = True
344 | for key in kwargs:
345 | if kwargs[key] != getattr(child, key):
346 | match = False
347 |
348 | # Only open existing instance if args matched
349 | if match:
350 | opened_instance = child
351 |
352 | # If we dont have an opened instance or raise_on_open is False then open new frame, otherwise raise it
353 | if opened_instance is None or raise_if_open is False:
354 | if len(kwargs) == 0:
355 | clazz(parent=parent).Show(True)
356 | else:
357 | clazz(parent=parent, **kwargs).Show(True)
358 | else:
359 | opened_instance.Raise()
360 |
--------------------------------------------------------------------------------
/test/test_correlation.py:
--------------------------------------------------------------------------------
1 | import unittest
2 | from unittest.mock import patch, PropertyMock
3 | import time
4 | import mt5_correlation.correlation as correlation
5 | import pandas as pd
6 | from datetime import datetime, timedelta
7 | from test_mt5 import Symbol
8 | import random
9 | import os
10 |
11 |
12 | class TestCorrelation(unittest.TestCase):
13 | # Mock symbols. 4 Symbols, 3 visible.
14 | mock_symbols = [Symbol(name='SYMBOL1', visible=True),
15 | Symbol(name='SYMBOL2', visible=True),
16 | Symbol(name='SYMBOL3', visible=False),
17 | Symbol(name='SYMBOL4', visible=True),
18 | Symbol(name='SYMBOL5', visible=True)]
19 |
20 | # Start and end date for price data and mock prices: base; correlated; and uncorrelated.
21 | start_date = None
22 | end_date = None
23 | price_columns = None
24 | mock_base_prices = None
25 | mock_correlated_prices = None
26 | mock_uncorrelated_prices = None
27 |
28 | def setUp(self):
29 | """
30 | Creates some price data fro use in tests
31 | :return:
32 | """
33 | # Start and end date for price data and mock price dataframes. One for: base; correlated; uncorrelated and
34 | # different dates.
35 | self.start_date = datetime(2021, 1, 1, 1, 5, 0)
36 | self.end_date = datetime(2021, 1, 1, 11, 30, 0)
37 | self.price_columns = ['time', 'close']
38 | self.mock_base_prices = pd.DataFrame(columns=self.price_columns)
39 | self.mock_correlated_prices = pd.DataFrame(columns=self.price_columns)
40 | self.mock_uncorrelated_prices = pd.DataFrame(columns=self.price_columns)
41 | self.mock_correlated_different_dates = pd.DataFrame(columns=self.price_columns)
42 | self.mock_inverse_correlated_prices = pd.DataFrame(columns=self.price_columns)
43 |
44 | # Build the price data for the test. One price every 5 minutes for 500 rows. Base will use min for price,
45 | # correlated will use min + 5 and uncorrelated will use random
46 | for date in (self.start_date + timedelta(minutes=m) for m in range(0, 500*5, 5)):
47 | self.mock_base_prices = self.mock_base_prices.append(pd.DataFrame(columns=self.price_columns,
48 | data=[[date, date.minute]]))
49 | self.mock_correlated_prices = \
50 | self.mock_correlated_prices.append(pd.DataFrame(columns=self.price_columns,
51 | data=[[date, date.minute + 5]]))
52 | self.mock_uncorrelated_prices = \
53 | self.mock_uncorrelated_prices.append(pd.DataFrame(columns=self.price_columns,
54 | data=[[date, random.randint(0, 1000000)]]))
55 |
56 | self.mock_correlated_different_dates = \
57 | self.mock_correlated_different_dates.append(pd.DataFrame(columns=self.price_columns,
58 | data=[[date + timedelta(minutes=100),
59 | date.minute + 5]]))
60 | self.mock_inverse_correlated_prices = \
61 | self.mock_inverse_correlated_prices.append(pd.DataFrame(columns=self.price_columns,
62 | data=[[date, (date.minute + 5) * -1]]))
63 |
64 | @patch('mt5_correlation.mt5.MetaTrader5')
65 | def test_calculate(self, mock):
66 | """
67 | Test the calculate method. Uses mock for MT5 symbols and prices.
68 | :param mock:
69 | :return:
70 | """
71 | # Mock symbol return values
72 | mock.symbols_get.return_value = self.mock_symbols
73 |
74 | # Correlation class
75 | cor = correlation.Correlation(monitoring_threshold=1, monitor_inverse=True)
76 |
77 | # Calculate for price data. We should have 100% matching dates in sets. Get prices should be called 4 times.
78 | # We don't have a SYMBOL3 as this is set as not visible. Correlations should be as follows:
79 | # SYMBOL1:SYMBOL2 should be fully correlated (1)
80 | # SYMBOL1:SYMBOL4 should be uncorrelated (0)
81 | # SYMBOL1:SYMBOL5 should be negatively correlated
82 | # SYMBOL2:SYMBOL5 should be negatively correlated
83 | # We will not use p_value as the last set uses random numbers so p value will not be useful.
84 | mock.copy_rates_range.side_effect = [self.mock_base_prices, self.mock_correlated_prices,
85 | self.mock_uncorrelated_prices, self.mock_inverse_correlated_prices]
86 | cor.calculate(date_from=self.start_date, date_to=self.end_date, timeframe=5, min_prices=100,
87 | max_set_size_diff_pct=100, overlap_pct=100, max_p_value=1)
88 |
89 | # Test the output. We should have 6 rows. S1:S2 c=1, S1:S4 c<1, S1:S5 c=-1, S2:S5 c=-1. We are not checking
90 | # S2:S4 or S4:S5
91 | self.assertEqual(len(cor.coefficient_data.index), 6, "There should be six correlations rows calculated.")
92 | self.assertEqual(cor.get_base_coefficient('SYMBOL1', 'SYMBOL2'), 1,
93 | "The correlation for SYMBOL1:SYMBOL2 should be 1.")
94 | self.assertTrue(cor.get_base_coefficient('SYMBOL1', 'SYMBOL4') < 1,
95 | "The correlation for SYMBOL1:SYMBOL4 should be <1.")
96 | self.assertEqual(cor.get_base_coefficient('SYMBOL1', 'SYMBOL5'), -1,
97 | "The correlation for SYMBOL1:SYMBOL5 should be -1.")
98 | self.assertEqual(cor.get_base_coefficient('SYMBOL2', 'SYMBOL5'), -1,
99 | "The correlation for SYMBOL2:SYMBOL5 should be -1.")
100 |
101 | # Monitoring threshold is 1 and we are monitoring inverse. Get filtered correlations. There should be 3 (S1:S2,
102 | # S1:S5 and S2:S5)
103 | self.assertEqual(len(cor.filtered_coefficient_data.index), 3,
104 | "There should be 3 rows in filtered coefficient data when we are monitoring inverse "
105 | "correlations.")
106 |
107 | # Now aren't monitoring inverse correlations. There should only be one correlation when filtered
108 | cor.monitor_inverse = False
109 | self.assertEqual(len(cor.filtered_coefficient_data.index), 1,
110 | "There should be only 1 rows in filtered coefficient data when we are not monitoring inverse "
111 | "correlations.")
112 |
113 | # Now were going to recalculate, but this time SYMBOL1:SYMBOL2 will have non overlapping dates and coefficient
114 | # should be None. There shouldn't be a row. We should have correlations for S1:S4, S1:S5 and S4:S5
115 | mock.copy_rates_range.side_effect = [self.mock_base_prices, self.mock_correlated_different_dates,
116 | self.mock_correlated_prices, self.mock_correlated_prices]
117 | cor.calculate(date_from=self.start_date, date_to=self.end_date, timeframe=5, min_prices=100,
118 | max_set_size_diff_pct=100, overlap_pct=100, max_p_value=1)
119 | self.assertEqual(len(cor.coefficient_data.index), 3, "There should be three correlations rows calculated.")
120 | self.assertEqual(cor.coefficient_data.iloc[0, 2], 1, "The correlation for SYMBOL1:SYMBOL4 should be 1.")
121 | self.assertEqual(cor.coefficient_data.iloc[1, 2], 1, "The correlation for SYMBOL1:SYMBOL5 should be 1.")
122 | self.assertEqual(cor.coefficient_data.iloc[2, 2], 1, "The correlation for SYMBOL4:SYMBOL5 should be 1.")
123 |
124 | # Get the price data used to calculate the coefficients for symbol 1. It should match mock_base_prices.
125 | price_data = cor.get_price_data('SYMBOL1')
126 | self.assertTrue(price_data.equals(self.mock_base_prices), "Price data returned post calculation should match "
127 | "mock price data.")
128 |
129 | def test_calculate_coefficient(self):
130 | """
131 | Tests the coefficient calculation.
132 | :return:
133 | """
134 | # Correlation class
135 | cor = correlation.Correlation()
136 |
137 | # Test 2 correlated sets
138 | coefficient = cor.calculate_coefficient(self.mock_base_prices, self.mock_correlated_prices)
139 | self.assertEqual(coefficient, 1, "Coefficient should be 1.")
140 |
141 | # Test 2 uncorrelated sets. Set p value to 1 to force correlation to be returned.
142 | coefficient = cor.calculate_coefficient(self.mock_base_prices, self.mock_uncorrelated_prices, max_p_value=1)
143 | self.assertTrue(coefficient < 1, "Coefficient should be < 1.")
144 |
145 | # Test 2 sets where prices dont overlap
146 | coefficient = cor.calculate_coefficient(self.mock_base_prices, self.mock_correlated_different_dates)
147 | self.assertTrue(coefficient < 1, "Coefficient should be None.")
148 |
149 | # Test 2 inversely correlated sets
150 | coefficient = cor.calculate_coefficient(self.mock_base_prices, self.mock_inverse_correlated_prices)
151 | self.assertEqual(coefficient, -1, "Coefficient should be -1.")
152 |
153 | @patch('mt5_correlation.mt5.MetaTrader5')
154 | def test_get_ticks(self, mock):
155 | """
156 | Test that caching works. For the purpose of this test, we can use price data rather than tick data.
157 | Mock 2 different sets of prices. Get three times. Base, One within cache threshold and one outside. Set 1
158 | should match set 2 but differ from set 3.
159 | :param mock:
160 | :return:
161 | """
162 |
163 | # Correlation class to test
164 | cor = correlation.Correlation()
165 |
166 | # Mock the tick data to contain 2 different sets. Then get twice. They should match as the data was cached.
167 | mock.copy_ticks_range.side_effect = [self.mock_base_prices, self.mock_correlated_prices]
168 |
169 | # We need to start and stop the monitor as this will set the cache time
170 | cor.start_monitor(interval=10, calculation_params={'from': 10, 'min_prices': 0, 'max_set_size_diff_pct': 0,
171 | 'overlap_pct': 0, 'max_p_value': 1}, cache_time=3)
172 | cor.stop_monitor()
173 |
174 | # Get the ticks within cache time and check that they match
175 | base_ticks = cor.get_ticks('SYMBOL1', None, None)
176 | cached_ticks = cor.get_ticks('SYMBOL1', None, None)
177 | self.assertTrue(base_ticks.equals(cached_ticks),
178 | "Both sets of tick data should match as set 2 came from cache.")
179 |
180 | # Wait 3 seconds
181 | time.sleep(3)
182 |
183 | # Retrieve again. This one should be different as the cache has expired.
184 | non_cached_ticks = cor.get_ticks('SYMBOL1', None, None)
185 | self.assertTrue(not base_ticks.equals(non_cached_ticks),
186 | "Both sets of tick data should differ as cached data had expired.")
187 |
188 | @patch('mt5_correlation.mt5.MetaTrader5')
189 | def test_start_monitor(self, mock):
190 | """
191 | Test that starting the monitor and running for 2 seconds produces two sets of coefficient history when using an
192 | interval of 1 second.
193 | :param mock:
194 | :return:
195 | """
196 | # Mock symbol return values
197 | mock.symbols_get.return_value = self.mock_symbols
198 |
199 | # Create correlation class. We will set a divergence threshold so that we can test status.
200 | cor = correlation.Correlation(divergence_threshold=0.8, monitor_inverse=True)
201 |
202 | # Calculate for price data. We should have 100% matching dates in sets. Get prices should be called 4 times.
203 | # We dont have a SYMBOL2 as this is set as not visible. All pairs should be correlated for the purpose of this
204 | # test.
205 | mock.copy_rates_range.side_effect = [self.mock_base_prices, self.mock_correlated_prices,
206 | self.mock_correlated_prices, self.mock_inverse_correlated_prices]
207 |
208 | cor.calculate(date_from=self.start_date, date_to=self.end_date, timeframe=5, min_prices=100,
209 | max_set_size_diff_pct=100, overlap_pct=100, max_p_value=1)
210 |
211 | # We will build some tick data for each symbol and patch it in. Tick data will be from 10 seconds ago to now.
212 | # We only need to patch in one set of tick data for each symbol as it will be cached.
213 | columns = ['time', 'ask']
214 | starttime = datetime.now() - timedelta(seconds=10)
215 | tick_data_s1 = pd.DataFrame(columns=columns)
216 | tick_data_s2 = pd.DataFrame(columns=columns)
217 | tick_data_s4 = pd.DataFrame(columns=columns)
218 | tick_data_s5 = pd.DataFrame(columns=columns)
219 |
220 | now = datetime.now()
221 | price_base = 1
222 | while starttime < now:
223 | tick_data_s1 = tick_data_s1.append(pd.DataFrame(columns=columns, data=[[starttime, price_base * 0.5]]))
224 | tick_data_s2 = tick_data_s1.append(pd.DataFrame(columns=columns, data=[[starttime, price_base * 0.1]]))
225 | tick_data_s4 = tick_data_s1.append(pd.DataFrame(columns=columns, data=[[starttime, price_base * 0.25]]))
226 | tick_data_s5 = tick_data_s1.append(pd.DataFrame(columns=columns, data=[[starttime, price_base * -0.25]]))
227 | starttime = starttime + timedelta(milliseconds=10*random.randint(0, 100))
228 | price_base += 1
229 |
230 | # Patch it in
231 | mock.copy_ticks_range.side_effect = [tick_data_s1, tick_data_s2, tick_data_s4, tick_data_s5]
232 |
233 | # Start the monitor. Run every second. Use ~10 and ~5 seconds of data. Were not testing the overlap and price
234 | # data quality metrics here as that is set elsewhere so these can be set to not take effect. Set cache level
235 | # high and don't use autosave. Timer runs in a separate thread so test can continue after it has started.
236 | cor.start_monitor(interval=1, calculation_params=[{'from': 0.66, 'min_prices': 0,
237 | 'max_set_size_diff_pct': 0, 'overlap_pct': 0,
238 | 'max_p_value': 1},
239 | {'from': 0.33, 'min_prices': 0,
240 | 'max_set_size_diff_pct': 0, 'overlap_pct': 0,
241 | 'max_p_value': 1}], cache_time=100, autosave=False)
242 |
243 | # Wait 2 seconds so timer runs twice
244 | time.sleep(2)
245 |
246 | # Stop the monitor
247 | cor.stop_monitor()
248 |
249 | # We should have 2 coefficients calculated for each symbol pair (6), for each date_from value (2),
250 | # for each run (2) so 24 in total.
251 | self.assertEqual(len(cor.coefficient_history.index), 24)
252 |
253 | # We should have 2 coefficients calculated for a single symbol pair and timeframe
254 | self.assertEqual(len(cor.get_coefficient_history({'Symbol 1': 'SYMBOL1', 'Symbol 2': 'SYMBOL2',
255 | 'Timeframe': 0.66})),
256 | 2, "We should have 2 history records for SYMBOL1:SYMBOL2 using the 0.66 min timeframe.")
257 |
258 | # The status should be DIVERGED for SYMBOL1:SYMBOL2 and CORRELATED for SYMBOL1:SYMBOL4 and SYMBOL2:SYMBOL4.
259 | self.assertTrue(cor.get_last_status('SYMBOL1', 'SYMBOL2') == correlation.STATUS_DIVERGED)
260 | self.assertTrue(cor.get_last_status('SYMBOL1', 'SYMBOL4') == correlation.STATUS_CORRELATED)
261 | self.assertTrue(cor.get_last_status('SYMBOL2', 'SYMBOL4') == correlation.STATUS_CORRELATED)
262 |
263 | # We are monitoring inverse correlations, status for SYMBOL1:SYMBOL5 should be DIVERGED
264 | self.assertTrue(cor.get_last_status('SYMBOL2', 'SYMBOL5') == correlation.STATUS_DIVERGED)
265 |
266 | @patch('mt5_correlation.mt5.MetaTrader5')
267 | def test_load_and_save(self, mock):
268 | """Calculate and run monitor for a few seconds. Store the data. Save it, load it then compare against stored
269 | data."""
270 |
271 | # Correlation class
272 | cor = correlation.Correlation()
273 |
274 | # Patch symbol and price data, then calculate
275 | mock.symbols_get.return_value = self.mock_symbols
276 | mock.copy_rates_range.side_effect = [self.mock_base_prices, self.mock_correlated_prices,
277 | self.mock_correlated_prices, self.mock_inverse_correlated_prices]
278 | cor.calculate(date_from=self.start_date, date_to=self.end_date, timeframe=5, min_prices=100,
279 | max_set_size_diff_pct=100, overlap_pct=100, max_p_value=1)
280 |
281 | # Patch the tick data
282 | columns = ['time', 'ask']
283 | starttime = datetime.now() - timedelta(seconds=10)
284 | tick_data_s1 = pd.DataFrame(columns=columns)
285 | tick_data_s3 = pd.DataFrame(columns=columns)
286 | tick_data_s4 = pd.DataFrame(columns=columns)
287 | now = datetime.now()
288 | price_base = 1
289 | while starttime < now:
290 | tick_data_s1 = tick_data_s1.append(pd.DataFrame(columns=columns, data=[[starttime, price_base * 0.5]]))
291 | tick_data_s3 = tick_data_s1.append(pd.DataFrame(columns=columns, data=[[starttime, price_base * 0.1]]))
292 | tick_data_s4 = tick_data_s1.append(pd.DataFrame(columns=columns, data=[[starttime, price_base * 0.25]]))
293 | starttime = starttime + timedelta(milliseconds=10 * random.randint(0, 100))
294 | price_base += 1
295 | mock.copy_ticks_range.side_effect = [tick_data_s1, tick_data_s3, tick_data_s4]
296 |
297 | # Start monitor and run for a seconds with a 1 second interval to produce some coefficient history. Then stop
298 | # the monitor
299 | cor.start_monitor(interval=1, calculation_params={'from': 0.66, 'min_prices': 0, 'max_set_size_diff_pct': 0,
300 | 'overlap_pct': 0, 'max_p_value': 1},
301 | cache_time=100, autosave=False)
302 | time.sleep(2)
303 | cor.stop_monitor()
304 |
305 | # Get copies of data that will be saved.
306 | cd_copy = cor.coefficient_data
307 | pd_copy = cor.get_price_data('SYMBOL1')
308 | mtd_copy = cor.get_ticks('SYMBOL1', cache_only=True)
309 | ch_copy = cor.coefficient_history
310 |
311 | # Save, reset data, then reload
312 | cor.save("unittest.cpd")
313 | cor.load("unittest.cpd")
314 |
315 | # Test that the reloaded data matches the original
316 | self.assertTrue(cd_copy.equals(cor.coefficient_data),
317 | "Saved and reloaded coefficient data should match original.")
318 | self.assertTrue(pd_copy.equals(cor.get_price_data('SYMBOL1')),
319 | "Saved and reloaded price data should match original.")
320 | self.assertTrue(mtd_copy.equals(cor.get_ticks('SYMBOL1', cache_only=True)),
321 | "Saved and reloaded tick data should match original.")
322 | self.assertTrue(ch_copy.equals(cor.coefficient_history),
323 | "Saved and reloaded coefficient history should match original.")
324 |
325 | # Cleanup. delete the file
326 | os.remove("unittest.cpd")
327 |
328 | @patch('mt5_correlation.correlation.Correlation.coefficient_data', new_callable=PropertyMock)
329 | def test_diverged_symbols(self, mock):
330 | """
331 | Test that diverged_symbols property correctly groups symbols and counts.
332 | :param mock:
333 | :return:
334 | """
335 | # Correlation class
336 | cor = correlation.Correlation()
337 |
338 | # Mock the correlation data. Symbol 1 has diverged 3 times; symbols 2 has diverged twice; symbol 3 has
339 | # diverged once; symbols 4 has diverged twice and symbol 5 has not diverged at all. Use all diverged status'
340 | # (diverged, diverging & converging). Also add a row for a non diverged pair.
341 | mock.return_value = pd.DataFrame(columns=['Symbol 1', 'Symbol 2', 'Status'], data=[
342 | ['SYMBOL1', 'SYMBOL2', correlation.STATUS_DIVERGED],
343 | ['SYMBOL1', 'SYMBOL3', correlation.STATUS_DIVERGING],
344 | ['SYMBOL1', 'SYMBOL4', correlation.STATUS_CONVERGING],
345 | ['SYMBOL2', 'SYMBOL4', correlation.STATUS_DIVERGED],
346 | ['SYMBOL2', 'SYMBOL3', correlation.STATUS_CORRELATED]])
347 |
348 | # Get the diverged_symbols data and check the counts
349 | diverged_symbols = cor.diverged_symbols
350 | self.assertEqual(diverged_symbols.loc[(diverged_symbols['Symbol'] == 'SYMBOL1')]['Count'].iloc[0], 3,
351 | "Symbol 1 has diverged three times.")
352 | self.assertEqual(diverged_symbols.loc[(diverged_symbols['Symbol'] == 'SYMBOL2')]['Count'].iloc[0], 2,
353 | "Symbol 2 has diverged twice.")
354 | self.assertEqual(diverged_symbols.loc[(diverged_symbols['Symbol'] == 'SYMBOL3')]['Count'].iloc[0], 1,
355 | "Symbol 3 has diverged once.")
356 | self.assertEqual(diverged_symbols.loc[(diverged_symbols['Symbol'] == 'SYMBOL4')]['Count'].iloc[0], 2,
357 | "Symbol 4 has diverged twice.")
358 | self.assertFalse('SYMBOL5' in diverged_symbols['Symbol'])
359 |
360 |
361 | if __name__ == '__main__':
362 | unittest.main()
363 |
--------------------------------------------------------------------------------
/mt5_correlation/correlation.py:
--------------------------------------------------------------------------------
1 | import math
2 | import logging
3 | import pandas as pd
4 | from datetime import datetime, timedelta
5 | import time
6 | import sched
7 | import threading
8 | import pytz
9 | from scipy.stats.stats import pearsonr
10 | import pickle
11 | import inspect
12 | import sys
13 |
14 | from mt5_correlation.mt5 import MT5
15 |
16 |
17 | class CorrelationStatus:
18 | """
19 | The status of the monitoring event for a symbol pair.
20 | """
21 |
22 | val = None
23 | text = None
24 | long_text = None
25 |
26 | def __init__(self, status_val, status_text, status_long_text=None):
27 | """
28 | Creates a status.
29 | :param status_val:
30 | :param status_text:
31 | :param status_long_text
32 | :return:
33 | """
34 | self.val = status_val
35 | self.text = status_text
36 | self.long_text = status_long_text
37 |
38 | def __eq__(self, other):
39 | """
40 | Compare the status val. We can compare against other CorrelationStatus instances or against int.
41 | :param other:
42 | :return:
43 | """
44 | if isinstance(other, self.__class__):
45 | return self.val == other.val
46 | elif isinstance(other, int):
47 | return self.val == other
48 | else:
49 | return False
50 |
51 | def __str__(self):
52 | """
53 | str is the text for the status.
54 | :return:
55 | """
56 | return self.text
57 |
58 |
59 | # All status's for symbol pair from monitoring. Status set from assessing coefficient for all timeframes from last run.
60 | STATUS_NOT_CALCULATED = CorrelationStatus(-1, 'NOT CALC', 'Coefficient could not be calculated')
61 | STATUS_CORRELATED = CorrelationStatus(1, 'CORRELATED', 'Coefficients for all timeframes are equal to or above the '
62 | 'divergence threshold')
63 | STATUS_DIVERGED = CorrelationStatus(2, 'DIVERGED', 'Coefficients for all timeframes are below the divergence threshold')
64 | STATUS_INCONSISTENT = CorrelationStatus(3, 'INCONSISTENT', 'Coefficients not consistently above or below divergence '
65 | 'threshold and are neither trending towards divergence or '
66 | 'convergence')
67 | STATUS_DIVERGING = CorrelationStatus(4, 'DIVERGING', 'Coefficients, when ordered by timeframe, are trending '
68 | 'towards convergence. The shortest timeframe is below the '
69 | 'divergence threshold and the longest timeframe is above the '
70 | 'divergence threshold')
71 | STATUS_CONVERGING = CorrelationStatus(5, 'CONVERGING', 'Coefficients, when ordered by timeframe, are trending '
72 | 'towards divergence. The shortest timeframe is above the '
73 | 'divergence threshold and the longest timeframe is below the '
74 | 'divergence threshold')
75 |
76 |
77 | class Correlation:
78 | """
79 | A class to maintain the state of the calculated correlation coefficients.
80 | """
81 |
82 | # Connection to MetaTrader5
83 | __mt5 = None
84 |
85 | # Minimum base coefficient for monitoring. Symbol pairs with a lower correlation
86 | # coefficient than ths won't be monitored.
87 | monitoring_threshold = 0.9
88 |
89 | # Threshold for divergence. Correlation coefficients that were previously above the monitoring_threshold and fall
90 | # below this threshold will be considered as having diverged
91 | divergence_threshold = 0.8
92 |
93 | # Flag to determine we monitor and report on inverse correlations
94 | monitor_inverse = False
95 |
96 | # Toggle on whether we are monitoring or not. Set through start_monitor and stop_monitor
97 | __monitoring = False
98 |
99 | # Monitoring calculation params, interval, cache_time, autosave and filename. Passed to start_monitor
100 | __monitoring_params = []
101 | __interval = None
102 | __cache_time = None
103 | __autosave = None
104 | __filename = None
105 |
106 | # First run of scheduler
107 | __first_run = True
108 |
109 | # The price data used to calculate the correlations
110 | __price_data = None
111 |
112 | # Coefficient data and history. Will be created in init call to __reset_coefficient_data
113 | coefficient_data = None
114 | coefficient_history = None
115 |
116 | # Stores tick data used to calculate coefficient during Monitor.
117 | # Dict: {Symbol: [retrieved datetime, ticks dataframe]}
118 | __monitor_tick_data = {}
119 |
120 | def __init__(self, monitoring_threshold=0.9, divergence_threshold=0.8, monitor_inverse=False):
121 | """
122 | Initialises the Correlation class.
123 | :param monitoring_threshold: Only correlations that are greater than or equal to this threshold will be
124 | monitored.
125 | :param divergence_threshold: Correlations that are being monitored and fall below this threshold are considered
126 | to have diverged.
127 | :param monitor_inverse: Whether we will monitor and report on negative / inverse correlations.
128 | """
129 | # Logger
130 | self.__log = logging.getLogger(__name__)
131 |
132 | # Connection to MetaTrader5
133 | self.__mt5 = MT5()
134 |
135 | # Create dataframe for coefficient data
136 | self.__reset_coefficient_data()
137 |
138 | # Create timer for continuous monitoring
139 | self.__scheduler = sched.scheduler(time.time, time.sleep)
140 |
141 | # Set thresholds and flags
142 | self.monitoring_threshold = monitoring_threshold
143 | self.divergence_threshold = divergence_threshold
144 | self.monitor_inverse = monitor_inverse
145 |
146 | @property
147 | def filtered_coefficient_data(self):
148 | """
149 | :return: Coefficient data filtered so that all base coefficients >= monitoring_threshold
150 | """
151 | filtered_data = None
152 | if self.coefficient_data is not None:
153 | if self.monitor_inverse:
154 | filtered_data = self.coefficient_data \
155 | .loc[(self.coefficient_data['Base Coefficient'] >= self.monitoring_threshold) |
156 | (self.coefficient_data['Base Coefficient'] <= self.monitoring_threshold * -1)]
157 |
158 | else:
159 | filtered_data = self.coefficient_data.loc[self.coefficient_data['Base Coefficient'] >=
160 | self.monitoring_threshold]
161 |
162 | return filtered_data
163 |
164 | @property
165 | def diverged_symbols(self):
166 | """
167 | :return: dataframe containing all diverged, diverging or converging symbols and count of number of
168 | divergences for those symbols.
169 | """
170 | filtered_data = None
171 |
172 | if self.coefficient_data is not None:
173 | # Only rows where we have a divergence
174 | filtered_data = self.coefficient_data \
175 | .loc[(self.coefficient_data['Status'] == STATUS_DIVERGED) |
176 | (self.coefficient_data['Status'] == STATUS_DIVERGING) |
177 | (self.coefficient_data['Status'] == STATUS_CONVERGING)]
178 |
179 | # We only need the symbols
180 | all_symbols = pd.DataFrame(columns=['Symbol', 'Count'],
181 | data={'Symbol': filtered_data['Symbol 1'].append(filtered_data['Symbol 2']),
182 | 'Count': 1})
183 |
184 | # Group and count. Reset index so that we have named SYMBOL column.
185 | filtered_data = all_symbols.groupby(by='Symbol').count().reset_index()
186 |
187 | # Sort
188 | filtered_data = filtered_data.sort_values('Count', ascending=False)
189 |
190 | return filtered_data
191 |
192 | def load(self, filename):
193 | """
194 | Loads calculated coefficients, price data used to calculate them and tick data used during monitoring.
195 | coefficients
196 | :param filename: The filename for the coefficient data to load.
197 | :return:
198 | """
199 | # Load data
200 | with open(filename, 'rb') as file:
201 | loaded_dict = pickle.load(file)
202 |
203 | # Get data from loaded dict and save
204 | self.coefficient_data = loaded_dict["coefficient_data"]
205 | self.__price_data = loaded_dict["price_data"]
206 | self.__monitor_tick_data = loaded_dict["monitor_tick_data"]
207 | self.coefficient_history = loaded_dict["coefficient_history"]
208 |
209 | def save(self, filename):
210 | """
211 | Saves the calculated coefficients, the price data used to calculate and the tick data for monitoring to a file.
212 | :param filename: The filename to save the data to.
213 | :return:
214 | """
215 | # Add data to dict then use pickle to save
216 | save_dict = {"coefficient_data": self.coefficient_data, "price_data": self.__price_data,
217 | "monitor_tick_data": self.__monitor_tick_data, "coefficient_history": self.coefficient_history}
218 | with open(filename, 'wb') as file:
219 | pickle.dump(save_dict, file, protocol=pickle.HIGHEST_PROTOCOL)
220 |
221 | def calculate(self, date_from, date_to, timeframe, min_prices=100, max_set_size_diff_pct=90, overlap_pct=90,
222 | max_p_value=0.05):
223 | """
224 | Calculates correlation coefficient between all symbols in MetaTrader5 Market Watch. Updates coefficient data.
225 |
226 | :param date_from: From date for price data from which to calculate correlation coefficients
227 | :param date_to: To date for price data from which to calculate correlation coefficients
228 | :param timeframe: Timeframe for price data from which to calculate correlation coefficients
229 | :param min_prices: The minimum number of prices that should be used to calculate coefficient. If this threshold
230 | is not met then returned coefficient will be None
231 | :param max_set_size_diff_pct: Correlations will only be calculated if the sizes of the two price data sets are
232 | within this pct of each other
233 | :param overlap_pct:The dates and times in the two sets of data must match. The coefficient will only be
234 | calculated against the dates that overlap. Any non overlapping dates will be discarded. This setting
235 | specifies the minimum size of the overlapping data when compared to the smallest set as a %. A coefficient
236 | will not be calculated if this threshold is not met.
237 | :param max_p_value: The maximum p value for the correlation to be meaningful
238 |
239 | :return:
240 | """
241 |
242 | coefficient = None
243 |
244 | # If we are monitoring, stop. We will need to restart later
245 | was_monitoring = self.__monitoring
246 | if self.__monitoring:
247 | self.stop_monitor()
248 |
249 | # Clear the existing correlations
250 | self.__reset_coefficient_data()
251 |
252 | # Get all visible symbols
253 | symbols = self.__mt5.get_symbols()
254 |
255 | # Get price data for selected symbols. 1 week of 15 min OHLC data for each symbol. Add to dict.
256 | self.__price_data = {}
257 | for symbol in symbols:
258 | self.__price_data[symbol] = self.__mt5.get_prices(symbol=symbol, from_date=date_from, to_date=date_to,
259 | timeframe=timeframe)
260 |
261 | # Loop through all symbol pair combinations and calculate coefficient. Make sure you don't double count pairs
262 | # eg. (USD/GBP AUD/USD vs AUD/USD USD/GBP). Use grid of all symbols with i and j axis. j starts at i + 1 to
263 | # avoid duplicating. We will store all coefficients in a dataframe.
264 | index = 0
265 | # There will be (x^2 - x) / 2 pairs where x is number of symbols
266 | num_pair_combinations = int((len(symbols) ** 2 - len(symbols)) / 2)
267 |
268 | for i in range(0, len(symbols)):
269 | symbol1 = symbols[i]
270 |
271 | for j in range(i + 1, len(symbols)):
272 | symbol2 = symbols[j]
273 | index += 1
274 |
275 | # Get price data for both symbols
276 | symbol1_price_data = self.__price_data[symbol1]
277 | symbol2_price_data = self.__price_data[symbol2]
278 |
279 | # Get coefficient
280 | if symbol1_price_data is not None and symbol2_price_data is not None:
281 | coefficient = self.calculate_coefficient(symbol1_prices=symbol1_price_data,
282 | symbol2_prices=symbol2_price_data,
283 | min_prices=min_prices,
284 | max_set_size_diff_pct=max_set_size_diff_pct,
285 | overlap_pct=overlap_pct, max_p_value=max_p_value)
286 |
287 | # Store if valid
288 | if coefficient is not None:
289 |
290 | self.coefficient_data = \
291 | self.coefficient_data.append({'Symbol 1': symbol1, 'Symbol 2': symbol2,
292 | 'Base Coefficient': coefficient, 'UTC Date From': date_from,
293 | 'UTC Date To': date_to, 'Timeframe': timeframe, 'Status': ''},
294 | ignore_index=True)
295 |
296 | self.__log.debug(f"Pair {index} of {num_pair_combinations}: {symbol1}:{symbol2} has a "
297 | f"coefficient of {coefficient}.")
298 | else:
299 | self.__log.debug(f"Coefficient for pair {index} of {num_pair_combinations}: {symbol1}:"
300 | f"{symbol2} could no be calculated.")
301 |
302 | # Sort, highest correlated first
303 | self.coefficient_data = self.coefficient_data.sort_values('Base Coefficient', ascending=False)
304 |
305 | # If we were monitoring, we stopped, so start again.
306 | if was_monitoring:
307 | self.start_monitor(interval=self.__interval, calculation_params=self.__monitoring_params,
308 | cache_time=self.__cache_time, autosave=self.__autosave, filename=self.__filename)
309 |
310 | def get_price_data(self, symbol):
311 | """
312 | Returns the price data used to calculate the base coefficients for the specified symbol
313 | :param symbol: Symbol to get price data for.
314 | :return: price data
315 | """
316 | price_data = None
317 | if symbol in self.__price_data:
318 | price_data = self.__price_data[symbol]
319 |
320 | return price_data
321 |
322 | def start_monitor(self, interval, calculation_params, cache_time=10, autosave=False, filename='autosave.cpd'):
323 | """
324 | Starts monitor to continuously update the coefficient for all symbol pairs in that meet the min_coefficient
325 | threshold.
326 |
327 | :param interval: How often to check in seconds
328 | :param calculation_params: A single dict or list of dicts containing the parameters for the coefficient
329 | calculations. On every iteration, a coefficient will be calculated for every set of params in list. Params
330 | contain the following values:
331 | from: The number of minutes of tick data to use for calculation. This can be a single value or
332 | a list. If a list, then calculations will be performed for every from date in list.
333 | min_prices: The minimum number of prices that should be used to calculate coefficient. If this threshold
334 | is not met then returned coefficient will be None
335 | max_set_size_diff_pct: Correlations will only be calculated if the sizes of the two price data sets are
336 | within this pct of each other
337 | overlap_pct: The dates and times in the two sets of data must match. The coefficient will only be
338 | calculated against the dates that overlap. Any non overlapping dates will be discarded. This
339 | setting specifies the minimum size of the overlapping data when compared to the smallest set as a %.
340 | A coefficient will not be calculated if this threshold is not met.
341 | max_p_value: The maximum p value for the correlation to be meaningful
342 | :param cache_time: Tick data is cached so that we can check coefficients for multiple symbol pairs and reuse
343 | the tick data. Number of seconds to cache tick data for before it becomes stale.
344 | :param autosave: Whether to autosave after every monitor run. If there is no filename specified then will
345 | create one named autosave.cpd
346 | :param filename: Filename for autosave. Default is autosave.cpd.
347 |
348 | :return: correlation coefficient, or None if coefficient could not be calculated.
349 | """
350 |
351 | if self.__monitoring:
352 | self.__log.debug(f"Request to start monitor when monitor is already running. Monitor will be stopped and"
353 | f"restarted with new parameters.")
354 | self.stop_monitor()
355 |
356 | self.__log.debug(f"Starting monitor.")
357 | self.__monitoring = True
358 |
359 | # Store the calculation params. If it isn't a list, convert to list of one to make code simpler later on.
360 | self.__monitoring_params = calculation_params if isinstance(calculation_params, list) \
361 | else [calculation_params, ]
362 |
363 | # Store the other params. We will need these later if monitor is stopped and needs to be restarted. This
364 | # happens in calculate.
365 | self.__interval = interval
366 | self.__cache_time = cache_time
367 | self.__autosave = autosave
368 | self.__filename = filename
369 |
370 | # Create thread to run monitoring This will call private __monitor method that will run the calculation and
371 | # keep scheduling itself while self.monitoring is True.
372 | thread = threading.Thread(target=self.__monitor)
373 | thread.start()
374 |
375 | def stop_monitor(self):
376 | """
377 | Stops monitoring symbol pairs for correlation.
378 | :return:
379 | """
380 | if self.__monitoring:
381 | self.__log.debug(f"Stopping monitor.")
382 | self.__monitoring = False
383 | else:
384 | self.__log.debug(f"Request to stop monitor when it is not running. No action taken.")
385 |
386 | def calculate_coefficient(self, symbol1_prices, symbol2_prices, min_prices: int = 100,
387 | max_set_size_diff_pct: int = 90, overlap_pct: int = 90,
388 | max_p_value: float = 0.05):
389 | """
390 | Calculates the correlation coefficient between two sets of price data. Uses close price.
391 |
392 | :param symbol1_prices: Pandas dataframe containing prices for symbol 1
393 | :param symbol2_prices: Pandas dataframe containing prices for symbol 2
394 | :param min_prices: The minimum number of prices that should be used to calculate coefficient. If this threshold
395 | is not met then returned coefficient will be None
396 | :param max_set_size_diff_pct: Correlations will only be calculated if the sizes of the two price data sets are
397 | within this pct of each other
398 | :param overlap_pct:
399 | :param max_p_value: The maximum p value for the correlation to be meaningful
400 |
401 | :return: correlation coefficient, or None if coefficient could not be calculated.
402 | :rtype: float or None
403 | """
404 |
405 | assert symbol1_prices is not None and symbol2_prices is not None
406 |
407 | # Calculate size of intersection and determine if prices for symbols have enough overlapping timestamps for
408 | # correlation coefficient calculation to be meaningful. Is the smallest set at least max_set_size_diff_pct % of
409 | # the size of the largest set and is the overlap set size at least overlap_pct % the size of the smallest set?
410 | coefficient = None
411 |
412 | intersect_dates = (set(symbol1_prices['time']) & set(symbol2_prices['time']))
413 | len_smallest_set = int(min([len(symbol1_prices.index), len(symbol2_prices.index)]))
414 | len_largest_set = int(max([len(symbol1_prices.index), len(symbol2_prices.index)]))
415 | similar_size = len_largest_set * (max_set_size_diff_pct / 100) <= len_smallest_set
416 | enough_overlap = len(intersect_dates) >= len_smallest_set * (overlap_pct / 100)
417 | enough_prices = len_smallest_set >= min_prices
418 | suitable = similar_size and enough_overlap and enough_prices
419 |
420 | if suitable:
421 | # Calculate coefficient on close prices
422 |
423 | # First filter prices to only include those that intersect
424 | symbol1_prices_filtered = symbol1_prices[symbol1_prices['time'].isin(intersect_dates)]
425 | symbol2_prices_filtered = symbol2_prices[symbol2_prices['time'].isin(intersect_dates)]
426 |
427 | # Calculate coefficient. Only use if p value is < max_p_value (highly likely that coefficient is valid
428 | # and null hypothesis is false).
429 | coefficient_with_p_value = pearsonr(symbol1_prices_filtered['close'], symbol2_prices_filtered['close'])
430 | coefficient = None if coefficient_with_p_value[1] > max_p_value else coefficient_with_p_value[0]
431 |
432 | # If NaN, change to None
433 | if coefficient is not None and math.isnan(coefficient):
434 | coefficient = None
435 |
436 | self.__log.debug(f"Calculate coefficient returning {coefficient}. "
437 | f"Symbol 1 Prices: {len(symbol1_prices)} Symbol 2 Prices: {len(symbol2_prices)} "
438 | f"Overlap Prices: {len(intersect_dates)} Similar size: {similar_size} "
439 | f"Enough overlap: {enough_overlap} Enough prices: {enough_prices} Suitable: {suitable}.")
440 |
441 | return coefficient
442 |
443 | def get_coefficient_history(self, filters=None):
444 | """
445 | Returns the coefficient history that matches the supplied filter.
446 | :param filters: Dict of all filters to apply. Possible values in dict are:
447 | Symbol 1
448 | Symbol 2
449 | Coefficient
450 | Timeframe
451 | Date From
452 | Date To
453 |
454 | If filter is not supplied, then all history is returned.
455 |
456 | :return: dataframe containing history of coefficient data.
457 | """
458 | history = self.coefficient_history
459 |
460 | # Apply filters
461 | if filters is not None:
462 | for key in filters:
463 | if key in history.columns:
464 | history = history[history[key] == filters[key]]
465 | else:
466 | self.__log.warning(f"Invalid column name provided for filter. Filter column: {key} "
467 | f"Valid columns: {history.columns}")
468 |
469 | return history
470 |
471 | def clear_coefficient_history(self):
472 | """
473 | Clears the coefficient history for all symbol pairs
474 | :return:
475 | """
476 | # Create dataframes for coefficient history.
477 | coefficient_history_columns = ['Symbol 1', 'Symbol 2', 'Coefficient', 'Timeframe', 'Date To']
478 | self.coefficient_history = pd.DataFrame(columns=coefficient_history_columns)
479 |
480 | # Clear tick data
481 | self.__monitor_tick_data = {}
482 |
483 | # Clear status from coefficient data
484 | self.coefficient_data['Status'] = ''
485 |
486 | def get_ticks(self, symbol, date_from=None, date_to=None, cache_only=False):
487 | """
488 | Returns the ticks for the specified symbol. Get's from cache if available and not older than cache_timeframe.
489 |
490 | :param symbol: Name of symbol to get ticks for.
491 | :param date_from: Date to get ticks from. Can only be None if getting from cache (cache_only=True)
492 | :param date_to:Date to get ticks to. Can only be None if getting from cache (cache_only=True)
493 | :param cache_only: Only retrieve from cache. cache_time is ignored. Returns None if symbol is not available in
494 | cache.
495 |
496 | :return:
497 | """
498 |
499 | timezone = pytz.timezone("Etc/UTC")
500 | utc_now = datetime.now(tz=timezone)
501 |
502 | ticks = None
503 |
504 | # Cache only
505 | if cache_only:
506 | if symbol in self.__monitor_tick_data:
507 | ticks = self.__monitor_tick_data[symbol][1]
508 | # Check if we have a cache time defined, if we already have the tick data and it is not stale
509 | elif self.__cache_time is not None and symbol in self.__monitor_tick_data and utc_now < \
510 | self.__monitor_tick_data[symbol][0] + timedelta(seconds=self.__cache_time):
511 | # Cached ticks are not stale. Get them
512 | ticks = self.__monitor_tick_data[symbol][1]
513 | self.__log.debug(f"Ticks for {symbol} retrieved from cache.")
514 | else:
515 | # Data does not exist in cache or cached data is stale. Retrieve from source and cache.
516 | ticks = self.__mt5.get_ticks(symbol=symbol, from_date=date_from, to_date=date_to)
517 | self.__monitor_tick_data[symbol] = [utc_now, ticks]
518 | self.__log.debug(f"Ticks for {symbol} retrieved from source and cached.")
519 | return ticks
520 |
521 | def get_last_status(self, symbol1, symbol2):
522 | """
523 | Get the last status for the specified symbol pair.
524 | :param symbol1
525 | :param symbol2
526 |
527 | :return: CorrelationStatus instance for symbol pair
528 | """
529 | status_col = self.coefficient_data.loc[(self.coefficient_data['Symbol 1'] == symbol1) &
530 | (self.coefficient_data['Symbol 2'] == symbol2), 'Status']
531 | status = status_col.values[0]
532 | return status
533 |
534 | def get_last_calculation(self, symbol1=None, symbol2=None):
535 | """
536 | Get the last calculation time the specified symbol pair. If no symbols are specified, then gets the last
537 | calculation time across all pairs
538 | :param symbol1
539 | :param symbol2
540 |
541 | :return: last calculation time
542 | """
543 | last_calc = None
544 | if self.coefficient_data is not None and len(self.coefficient_data.index) > 0:
545 | data = self.coefficient_data.copy()
546 |
547 | # Filter by symbols if specified
548 | data = data.loc[data['Symbol 1'] == symbol1] if symbol1 is not None else data
549 | data = data.loc[data['Symbol 2'] == symbol2] if symbol2 is not None else data
550 |
551 | # Filter to remove blank dates
552 | data = data.dropna(subset=['Last Calculation'])
553 |
554 | # Get the column
555 | col = data['Last Calculation']
556 |
557 | # Get max date from column
558 | if col is not None and len(col) > 0:
559 | last_calc = max(col.values)
560 |
561 | return last_calc
562 |
563 | def get_base_coefficient(self, symbol1, symbol2):
564 | """
565 | Returns the base coefficient for the specified symbol pair
566 | :param symbol1:
567 | :param symbol2:
568 | :return:
569 | """
570 | base_coefficient = None
571 | if self.coefficient_data is not None:
572 | row = self.coefficient_data[(self.coefficient_data['Symbol 1'] == symbol1) &
573 | (self.coefficient_data['Symbol 2'] == symbol2)]
574 |
575 | if row is not None and len(row) == 1:
576 | base_coefficient = row.iloc[0]['Base Coefficient']
577 |
578 | return base_coefficient
579 |
580 | def __monitor(self):
581 | """
582 | The actual monitor method. Private. This should not be called outside of this class. Use start_monitoring and
583 | stop_monitoring.
584 |
585 | :return: correlation coefficient, or None if coefficient could not be calculated.
586 | """
587 | self.__log.debug(f"In monitor event. Monitoring: {self.__monitoring}.")
588 |
589 | # Only run if monitor is not stopped
590 | if self.__monitoring:
591 | # Update all coefficients
592 | self.__update_all_coefficients()
593 |
594 | # Autosave
595 | if self.__autosave:
596 | self.save(filename=self.__filename)
597 |
598 | # Schedule the timer to run again
599 | self.__scheduler.enter(delay=self.__interval, priority=1, action=self.__monitor)
600 |
601 | # Log the stack. Debug stack overflow
602 | self.__log.debug(f"Current stack size: {len(inspect.stack())} Recursion limit: {sys.getrecursionlimit()}")
603 |
604 | # Run
605 | if self.__first_run:
606 | self.__first_run = False
607 | self.__scheduler.run()
608 |
609 | def __update_coefficients(self, symbol1, symbol2):
610 | """
611 | Updates the coefficients for the specified symbol pair
612 | :param symbol1: Name of symbol to calculate coefficient for.
613 | :param symbol2: Name of symbol to calculate coefficient for.
614 | :return: correlation coefficient, or None if coefficient could not be calculated.
615 | """
616 |
617 | # Get the largest value of from in monitoring_params. This will be used to retrieve the data. We will only
618 | # retrieve once and use for every set of params by getting subset of the data.
619 | max_from = None
620 | for params in self.__monitoring_params:
621 | if max_from is None:
622 | max_from = params['from']
623 | else:
624 | max_from = max(max_from, params['from'])
625 |
626 | # Date range for data
627 | timezone = pytz.timezone("Etc/UTC")
628 | date_to = datetime.now(tz=timezone)
629 | date_from = date_to - timedelta(minutes=max_from)
630 |
631 | # Get the tick data for the longest timeframe calculation.
632 | symbol1ticks = self.get_ticks(symbol=symbol1, date_from=date_from, date_to=date_to)
633 | symbol2ticks = self.get_ticks(symbol=symbol2, date_from=date_from, date_to=date_to)
634 |
635 | # Resample to 1 sec OHLC, this will help with coefficient calculation ensuring that we dont have more than
636 | # one tick per second and ensuring that times can match. We will need to set the index to time for the
637 | # resample then revert back to a 'time' column. We will then need to remove rows with nan in 'close' price
638 | s1_prices = None
639 | s2_prices = None
640 | if symbol1ticks is not None and symbol2ticks is not None and len(symbol1ticks.index) > 0 and \
641 | len(symbol2ticks.index) > 0:
642 |
643 | try:
644 | symbol1ticks = symbol1ticks.set_index('time')
645 | symbol2ticks = symbol2ticks.set_index('time')
646 | s1_prices = symbol1ticks['ask'].resample('1S').ohlc()
647 | s2_prices = symbol2ticks['ask'].resample('1S').ohlc()
648 | except RecursionError:
649 | self.__log.warning(f"Coefficient could not be calculated for {symbol1}:{symbol2}. prices could not "
650 | f"be resampled.")
651 | else:
652 | s1_prices.reset_index(inplace=True)
653 | s2_prices.reset_index(inplace=True)
654 | s1_prices = s1_prices[s1_prices['close'].notna()]
655 | s2_prices = s2_prices[s2_prices['close'].notna()]
656 |
657 | # Calculate for all sets of monitoring_params
658 | if s1_prices is not None and s2_prices is not None:
659 | coefficients = {}
660 | for params in self.__monitoring_params:
661 | # Get the from date as a datetime64
662 | date_from_subset = pd.Timestamp(date_to - timedelta(minutes=params['from'])).to_datetime64()
663 |
664 | # Get subset of the price data
665 | s1_prices_subset = s1_prices[(s1_prices['time'] >= date_from_subset)]
666 | s2_prices_subset = s2_prices[(s2_prices['time'] >= date_from_subset)]
667 |
668 | # Calculate the coefficient
669 | coefficient = \
670 | self.calculate_coefficient(symbol1_prices=s1_prices_subset, symbol2_prices=s2_prices_subset,
671 | min_prices=params['min_prices'],
672 | max_set_size_diff_pct=params['max_set_size_diff_pct'],
673 | overlap_pct=params['overlap_pct'], max_p_value=params['max_p_value'])
674 |
675 | self.__log.debug(f"Symbol pair {symbol1}:{symbol2} has a coefficient of {coefficient} for last "
676 | f"{params['from']} minutes.")
677 |
678 | # Add the coefficient to a dict {timeframe: coefficient}. We will update together for all for
679 | # symbol pair and time
680 | coefficients[params['from']] = coefficient
681 |
682 | # Update coefficient data for all coefficients for all timeframes for this run and symbol pair.
683 | self.__update_coefficient_data(symbol1=symbol1, symbol2=symbol2, coefficients=coefficients,
684 | date_to=date_to)
685 |
686 | def __update_all_coefficients(self):
687 | """
688 | Updates the coefficient for all symbol pairs in that meet the min_coefficient threshold. Symbol pairs that meet
689 | the threshold can be accessed through the filtered_coefficient_data property.
690 | """
691 | # Update latest coefficient for every pair
692 | for index, row in self.filtered_coefficient_data.iterrows():
693 | symbol1 = row['Symbol 1']
694 | symbol2 = row['Symbol 2']
695 | self.__update_coefficients(symbol1=symbol1, symbol2=symbol2)
696 |
697 | def __reset_coefficient_data(self):
698 | """
699 | Clears coefficient data and history.
700 | :return:
701 | """
702 | # Create dataframes for coefficient data.
703 | coefficient_data_columns = ['Symbol 1', 'Symbol 2', 'Base Coefficient', 'UTC Date From', 'UTC Date To',
704 | 'Timeframe', 'Last Calculation', 'Status']
705 | self.coefficient_data = pd.DataFrame(columns=coefficient_data_columns)
706 |
707 | # Clear coefficient history
708 | self.clear_coefficient_history()
709 |
710 | # Clear price data
711 | self.__price_data = None
712 |
713 | def __update_coefficient_data(self, symbol1, symbol2, coefficients, date_to):
714 | """
715 | Updates the coefficient data with the latest coefficient and adds to coefficient history.
716 | :param symbol1:
717 | :param symbol2:
718 | :param coefficients: Dict of all coefficients calculated for this run and symbol pair. {timeframe: coefficient}
719 | :param date_to: The date from for which the coefficient was calculated
720 | :return:
721 | """
722 |
723 | timezone = pytz.timezone("Etc/UTC")
724 | now = datetime.now(tz=timezone)
725 |
726 | # Update data if we have a coefficient and add to history
727 | if coefficients is not None:
728 | # Update the coefficient data table with the Last Calculation time.
729 | self.coefficient_data.loc[(self.coefficient_data['Symbol 1'] == symbol1) &
730 | (self.coefficient_data['Symbol 2'] == symbol2),
731 | 'Last Calculation'] = now
732 |
733 | # Are we an inverse correlation
734 | inverse = self.get_base_coefficient(symbol1, symbol2) <= self.monitoring_threshold * -1
735 |
736 | # Calculate status and update
737 | status = self.__calculate_status(coefficients=coefficients, inverse=inverse)
738 | self.coefficient_data.loc[(self.coefficient_data['Symbol 1'] == symbol1) &
739 | (self.coefficient_data['Symbol 2'] == symbol2),
740 | 'Status'] = status
741 |
742 | # Update history data
743 | for key in coefficients:
744 | row = pd.DataFrame(columns=self.coefficient_history.columns,
745 | data=[[symbol1, symbol2, coefficients[key], key, date_to]])
746 | self.coefficient_history = self.coefficient_history.append(row)
747 |
748 | def __calculate_status(self, coefficients, inverse):
749 | """
750 | Calculates the status from the supplied set of coefficients
751 | :param coefficients: Dict of timeframes and coefficients {timeframe: coefficient} to calculate status from
752 | :param: Whether we are calculating status based on normal or inverse correlation
753 | :return: status
754 | """
755 | status = STATUS_NOT_CALCULATED
756 |
757 | # Only continue if we have calculated all coefficients, otherwise we will return STATUS_NOT_CALCULATED
758 | if None not in coefficients.values():
759 | # Get the values ordered by timeframe descending
760 | ordered_values = []
761 | for key in sorted(coefficients, reverse=True):
762 | ordered_values.append(coefficients[key])
763 |
764 | if self.monitor_inverse and inverse:
765 | # Calculation for inverse calculations
766 | if all(i <= self.divergence_threshold * -1 for i in ordered_values):
767 | status = STATUS_CORRELATED
768 | elif all(i > self.divergence_threshold * -1 for i in ordered_values):
769 | status = STATUS_DIVERGED
770 | elif all(ordered_values[i] <= ordered_values[i+1] for i in range(0, len(ordered_values)-1, 1)):
771 | status = STATUS_CONVERGING
772 | elif all(ordered_values[i] > ordered_values[i+1] for i in range(0, len(ordered_values)-1, 1)):
773 | status = STATUS_DIVERGING
774 | else:
775 | status = STATUS_INCONSISTENT
776 | else:
777 | # Calculation for standard correlations
778 | if all(i >= self.divergence_threshold for i in ordered_values):
779 | status = STATUS_CORRELATED
780 | elif all(i < self.divergence_threshold for i in ordered_values):
781 | status = STATUS_DIVERGED
782 | elif all(ordered_values[i] <= ordered_values[i+1] for i in range(0, len(ordered_values)-1, 1)):
783 | status = STATUS_DIVERGING
784 | elif all(ordered_values[i] > ordered_values[i+1] for i in range(0, len(ordered_values)-1, 1)):
785 | status = STATUS_CONVERGING
786 | else:
787 | status = STATUS_INCONSISTENT
788 |
789 | return status
790 |
--------------------------------------------------------------------------------