├── Data
    └── 000660.pkl
├── README.md
├── .gitignore
└── AFML-02-Financial-Data-Structure
    └── 3.1-Labeling.ipynb


/Data/000660.pkl:
--------------------------------------------------------------------------------
https://raw.githubusercontent.com/AndersonJo/finance-machine-learning/master/Data/000660.pkl


--------------------------------------------------------------------------------
/README.md:
--------------------------------------------------------------------------------
1 | # Finance Machine Learning 
2 | 
3 | 해당 repository는 다음의 책 내용의 코드를 구현 합니다. 
4 | 
5 | 1. Advances in Financial Machine Learning (AFML)
6 | 
7 | 
8 | 


--------------------------------------------------------------------------------
/.gitignore:
--------------------------------------------------------------------------------
 1 | # IDE
 2 | .idea
 3 | _site
 4 | 
 5 | 
 6 | # Python
 7 | *.pyc
 8 | .ipynb_checkpoints
 9 | 
10 | # Data
11 | _data
12 | *.gz
13 | *.zip
14 | *.alz
15 | *.data
16 | *.json
17 | *.npz
18 | cifar-10-batches-py
19 | *.h5
20 | joblib
21 | 
22 | # Specific Data
23 | imdb_full.pkl
24 | 
25 | # TensorFlow Model
26 | *.model
27 | *.tfmodel
28 | _network
29 | _network_backup
30 | _tfmodel
31 | 
32 | # R
33 | .Rhistory
34 | 
35 | 


--------------------------------------------------------------------------------
/AFML-02-Financial-Data-Structure/3.1-Labeling.ipynb:
--------------------------------------------------------------------------------
  1 | {
  2 |  "cells": [
  3 |   {
  4 |    "cell_type": "code",
  5 |    "execution_count": 2,
  6 |    "id": "6b8f737e-7c38-4051-8d2f-60915b41319c",
  7 |    "metadata": {},
  8 |    "outputs": [],
  9 |    "source": [
 10 |     "import pandas as pd\n",
 11 |     "import numpy as np\n",
 12 |     "import pickle\n",
 13 |     "import matplotlib.pylab as plt\n",
 14 |     "from zeta.train import ZetaTrainSQL\n"
 15 |    ]
 16 |   },
 17 |   {
 18 |    "cell_type": "markdown",
 19 |    "id": "69f8ca6c-2f2d-4483-a918-aa10da07b06b",
 20 |    "metadata": {},
 21 |    "source": [
 22 |     "# Data"
 23 |    ]
 24 |   },
 25 |   {
 26 |    "cell_type": "code",
 27 |    "execution_count": 6,
 28 |    "id": "9c170d88-7bb9-487d-9676-a06f98dbb8a1",
 29 |    "metadata": {},
 30 |    "outputs": [
 31 |     {
 32 |      "data": {
 33 |       "text/html": [
 34 |        "<div>\n",
 35 |        "<style scoped>\n",
 36 |        "    .dataframe tbody tr th:only-of-type {\n",
 37 |        "        vertical-align: middle;\n",
 38 |        "    }\n",
 39 |        "\n",
 40 |        "    .dataframe tbody tr th {\n",
 41 |        "        vertical-align: top;\n",
 42 |        "    }\n",
 43 |        "\n",
 44 |        "    .dataframe thead th {\n",
 45 |        "        text-align: right;\n",
 46 |        "    }\n",
 47 |        "</style>\n",
 48 |        "<table border=\"1\" class=\"dataframe\">\n",
 49 |        "  <thead>\n",
 50 |        "    <tr style=\"text-align: right;\">\n",
 51 |        "      <th></th>\n",
 52 |        "      <th>stock_id</th>\n",
 53 |        "      <th>stock_code</th>\n",
 54 |        "      <th>Open</th>\n",
 55 |        "      <th>High</th>\n",
 56 |        "      <th>Low</th>\n",
 57 |        "      <th>Close</th>\n",
 58 |        "      <th>Volume</th>\n",
 59 |        "      <th>PriceVolume</th>\n",
 60 |        "      <th>adjusted_type</th>\n",
 61 |        "      <th>adjusted_ratio</th>\n",
 62 |        "    </tr>\n",
 63 |        "    <tr>\n",
 64 |        "      <th>date</th>\n",
 65 |        "      <th></th>\n",
 66 |        "      <th></th>\n",
 67 |        "      <th></th>\n",
 68 |        "      <th></th>\n",
 69 |        "      <th></th>\n",
 70 |        "      <th></th>\n",
 71 |        "      <th></th>\n",
 72 |        "      <th></th>\n",
 73 |        "      <th></th>\n",
 74 |        "      <th></th>\n",
 75 |        "    </tr>\n",
 76 |        "  </thead>\n",
 77 |        "  <tbody>\n",
 78 |        "    <tr>\n",
 79 |        "      <th>2021-08-30</th>\n",
 80 |        "      <td>1377</td>\n",
 81 |        "      <td>000660</td>\n",
 82 |        "      <td>105000.0</td>\n",
 83 |        "      <td>105500.0</td>\n",
 84 |        "      <td>103500.0</td>\n",
 85 |        "      <td>103500.0</td>\n",
 86 |        "      <td>1815760.0</td>\n",
 87 |        "      <td>188840.0</td>\n",
 88 |        "      <td>0.0</td>\n",
 89 |        "      <td>0.0</td>\n",
 90 |        "    </tr>\n",
 91 |        "    <tr>\n",
 92 |        "      <th>2021-08-31</th>\n",
 93 |        "      <td>1377</td>\n",
 94 |        "      <td>000660</td>\n",
 95 |        "      <td>103500.0</td>\n",
 96 |        "      <td>106500.0</td>\n",
 97 |        "      <td>102000.0</td>\n",
 98 |        "      <td>106500.0</td>\n",
 99 |        "      <td>4979280.0</td>\n",
100 |        "      <td>520812.0</td>\n",
101 |        "      <td>0.0</td>\n",
102 |        "      <td>0.0</td>\n",
103 |        "    </tr>\n",
104 |        "    <tr>\n",
105 |        "      <th>2021-09-01</th>\n",
106 |        "      <td>1377</td>\n",
107 |        "      <td>000660</td>\n",
108 |        "      <td>106000.0</td>\n",
109 |        "      <td>108000.0</td>\n",
110 |        "      <td>104500.0</td>\n",
111 |        "      <td>108000.0</td>\n",
112 |        "      <td>4736180.0</td>\n",
113 |        "      <td>504302.0</td>\n",
114 |        "      <td>0.0</td>\n",
115 |        "      <td>0.0</td>\n",
116 |        "    </tr>\n",
117 |        "    <tr>\n",
118 |        "      <th>2021-09-02</th>\n",
119 |        "      <td>1377</td>\n",
120 |        "      <td>000660</td>\n",
121 |        "      <td>108000.0</td>\n",
122 |        "      <td>108500.0</td>\n",
123 |        "      <td>106000.0</td>\n",
124 |        "      <td>106500.0</td>\n",
125 |        "      <td>2724560.0</td>\n",
126 |        "      <td>291367.0</td>\n",
127 |        "      <td>0.0</td>\n",
128 |        "      <td>0.0</td>\n",
129 |        "    </tr>\n",
130 |        "    <tr>\n",
131 |        "      <th>2021-09-03</th>\n",
132 |        "      <td>1377</td>\n",
133 |        "      <td>000660</td>\n",
134 |        "      <td>106500.0</td>\n",
135 |        "      <td>107500.0</td>\n",
136 |        "      <td>106000.0</td>\n",
137 |        "      <td>107000.0</td>\n",
138 |        "      <td>2493050.0</td>\n",
139 |        "      <td>266006.0</td>\n",
140 |        "      <td>0.0</td>\n",
141 |        "      <td>0.0</td>\n",
142 |        "    </tr>\n",
143 |        "  </tbody>\n",
144 |        "</table>\n",
145 |        "</div>"
146 |       ],
147 |       "text/plain": [
148 |        "            stock_id stock_code      Open      High       Low     Close  \\\n",
149 |        "date                                                                      \n",
150 |        "2021-08-30      1377     000660  105000.0  105500.0  103500.0  103500.0   \n",
151 |        "2021-08-31      1377     000660  103500.0  106500.0  102000.0  106500.0   \n",
152 |        "2021-09-01      1377     000660  106000.0  108000.0  104500.0  108000.0   \n",
153 |        "2021-09-02      1377     000660  108000.0  108500.0  106000.0  106500.0   \n",
154 |        "2021-09-03      1377     000660  106500.0  107500.0  106000.0  107000.0   \n",
155 |        "\n",
156 |        "               Volume  PriceVolume  adjusted_type  adjusted_ratio  \n",
157 |        "date                                                               \n",
158 |        "2021-08-30  1815760.0     188840.0            0.0             0.0  \n",
159 |        "2021-08-31  4979280.0     520812.0            0.0             0.0  \n",
160 |        "2021-09-01  4736180.0     504302.0            0.0             0.0  \n",
161 |        "2021-09-02  2724560.0     291367.0            0.0             0.0  \n",
162 |        "2021-09-03  2493050.0     266006.0            0.0             0.0  "
163 |       ]
164 |      },
165 |      "execution_count": 6,
166 |      "metadata": {},
167 |      "output_type": "execute_result"
168 |     }
169 |    ],
170 |    "source": [
171 |     "train_sql = ZetaTrainSQL()\n",
172 |     "data_df = train_sql.get_ohlcv('000660')\n",
173 |     "data_df.tail()"
174 |    ]
175 |   },
176 |   {
177 |    "cell_type": "markdown",
178 |    "id": "de6f1fba-1873-4a0d-baf2-a6a0ff1aab61",
179 |    "metadata": {},
180 |    "source": [
181 |     "# Labeling\n",
182 |     "\n",
183 |     "ML에서 레이블링은 매우 중요합니다. <br>\n",
184 |     "예를 들어, image classification labeling에서 강아지를 고양이라고 하거나, 고양이를 사자라고 한다면 학습은 잘 안될 겁니다. <br>\n",
185 |     "마찬가지로 파이낸셜을 ML로 돌리기 위해서는 적절한 레이블링을 적용해주는 것이 매우 중요합니다. \n",
186 |     "\n",
187 |     "## Fixed-Time Horizon Method \n",
188 |     "\n",
189 |     "대부분의 시계열 방법론에서 h 이후의 고정된 시간에서 지표가 올랐는지 떨어졌는지로 레이블링을 합니다. <br>\n",
190 |     "하지만 단연컨데.. 이 방법은 주가 레이블링에서는 가장 멍청한 방법중 하나입니다. <br>\n",
191 |     "\n",
192 |     "1. 샘플링 이슈가 있습니다. \n",
193 |     "    - 장초반, 장 끝나기전 10분이 가장 변동성도 크며, 주문량도 큽니다. \n",
194 |     "    - 반면에 오후 12시쯤부터는 다들 밥먹으러 가고, 쉬기 때문에 체결량은 현저하게 떨어집니다. \n",
195 |     "    - 따라서.. 고정적인 시간단위로 샘플링을 하게 된다면 장초반, 끝나기전 10분은 under sampling이 되며, 오후 시간대는 over sampling이 됩니다. \n",
196 |     "2. 통계적 유의미성이 없습니다. \n",
197 |     "    - 시계열간의 상관관계나, 정규성이 있어야 하는데. 주가는 그런 통계적 유의미성이 없습니다. \n",
198 |     "    - 참고로.. 이런 문제를 해결하기 위해서 Volume Bar, Dollar Bar같은 방법론들이 등장했습니다.\n",
199 |     "3. 변동성에 취약합니다. \n",
200 |     "    - 아래 공식에서 써놨듯이 특정 델타값 이상 또는 이하일 경우를 찾는데.. 종목마다 모두 다른 변동성을 갖고 있으며, 이벤트가 있을때마다 모두 다릅니다.\n",
201 |     "    - 따라서 고정된 델타값 또한 문제가 있습니다. \n",
202 |     "\n",
203 |     "참고로.. Fixed-Time Horizon Method의 공식은 다음과 같이 적용할 수 있습니다.\n",
204 |     "\n",
205 |     "$$ \\begin{align} r &= \\frac{p_t + p_{t+h}}{p_t} - 1  \\\\\n",
206 |     "y_i &= \n",
207 |     "\\begin{cases}\n",
208 |     "-1 & \\text{if } r < - \\delta \\\\\n",
209 |     "0 &  \\text{if } | r | \\le \\delta \\\\\n",
210 |     "1 &  \\text{if } r > \\delta\n",
211 |     "\\end{cases}\n",
212 |     "\\end{align} $$\n",
213 |     "\n",
214 |     "p는 가격이고, 그냥 percentage로 변환한 다음 어떤 특정 델타값 이상, 이하면 증감으로 레이블링을 한다는 것 입니다. "
215 |    ]
216 |   },
217 |   {
218 |    "cell_type": "markdown",
219 |    "id": "17be0d04-9970-4cf3-aed0-7ac2b090df20",
220 |    "metadata": {},
221 |    "source": [
222 |     "\n",
223 |     "## Volatility Threshold\n",
224 |     "\n",
225 |     "Fixed-time horizon method의 문제중 하나는 변동성을 포함하고 있지 않는다는 것 입니다.<br>\n",
226 |     "이부분이 중요한 이유는 실제 알고리즘 트레이딩에서는 profit taking 그리고 stop loss를 반드시 설정하기 때문입니다. <br>\n",
227 |     "아래는 하나의 예 일 뿐입니다. \n",
228 |     "\n",
229 |     "변동성을 확인하기 위해서.. 전날 종가와 비교해서 exponential weighted standard deviation을 구하기도 합니다. <br>\n",
230 |     "또는 여기서 더 나아가 다음날 종가를 머신러닝 모델을 태워서 예측 하기도 합니다. <br>\n",
231 |     "어려운 내용이 아니기 때문에 개념만 설명하고 넘어갑니다. "
232 |    ]
233 |   },
234 |   {
235 |    "cell_type": "code",
236 |    "execution_count": 27,
237 |    "id": "7ddb71ff-7fda-4062-ac1c-b65b55aa9f64",
238 |    "metadata": {},
239 |    "outputs": [
240 |     {
241 |      "data": {
242 |       "text/plain": [
243 |        "<matplotlib.collections.PolyCollection at 0x7f81b63d4580>"
244 |       ]
245 |      },
246 |      "execution_count": 27,
247 |      "metadata": {},
248 |      "output_type": "execute_result"
249 |     },
250 |     {
251 |      "data": {
252 |       "image/png": "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\n",
253 |       "text/plain": [
254 |        "<Figure size 720x288 with 1 Axes>"
255 |       ]
256 |      },
257 |      "metadata": {
258 |       "needs_background": "light"
259 |      },
260 |      "output_type": "display_data"
261 |     }
262 |    ],
263 |    "source": [
264 |     "pct_change = data_df.Close/ data_df.Close.shift() - 1 # np.minimum(data_df.Low, data_df.Close.shift()) - 1\n",
265 |     "pct_change = pct_change[(pct_change.quantile(0.01) < pct_change) & \n",
266 |     "                        (pct_change < pct_change.quantile(0.99))]\n",
267 |     "\n",
268 |     "volatility = pct_change.ewm(7).std().iloc[7:]\n",
269 |     "pct_change = pct_change.iloc[7:]\n",
270 |     "\n",
271 |     "fig, ax = plt.subplots(1, figsize=(10, 4))\n",
272 |     "ax.fill_between(pct_change.index, 0, pct_change, color='orange')\n",
273 |     "ax.fill_between(volatility.index, -volatility, volatility, color='red', alpha=0.6)"
274 |    ]
275 |   },
276 |   {
277 |    "cell_type": "markdown",
278 |    "id": "c18314b6-d15f-49db-9265-b8779a72b613",
279 |    "metadata": {},
280 |    "source": [
281 |     "## Triple-Barrier Method\n",
282 |     "\n",
283 |     "Volatility threshold 방법은 변동성에 따라서 상승, 하락 (또는 profit taking, stop loss) 를 다이나믹하게 잡아주는데 좋은 방법입니다.<br>\n",
284 |     "문제는 .. 얼마나 오랫동안 들고 있어야 하는지 입니다. <br>\n",
285 |     "무한정 기다려서 upper bound 또는 under bound를 주가가 넘어갈때를 기다릴수는 없습니다. \n",
286 |     "\n",
287 |     "이 문제를 해결하기 위해서, 하나의 barrier를 더해 줍니다. \n",
288 |     "\n",
289 |     "| Name             | Label   | Description                                                      |\n",
290 |     "|:-----------------|:--------|:-----------------------------------------------------------------|\n",
291 |     "| Upper Barrier    | 1       | 윗쪽 barrier에 주가가 처음 닿을때                                     |  \n",
292 |     "| Vertical Barrier | 0       | 일정 시점이후까지도 upper 또는 under barrier에 닿지 않고 주가가 횡보할 경우 | \n",
293 |     "| Under Barrier    | -1      | 아래쪽 barrier에 주가가 처음 닿을때                                    |\n",
294 |     "\n",
295 |     "위의 내용은 일종의 기본형이고, 응용버젼은 매우 많습니다.<br>\n",
296 |     "더 간단하게 만들기 위해서 Vertical Barrier 는 학습에서 제외시키고, binary classification으로 만든다던가.. <br>\n",
297 |     "또는 메타 데이터 또는 메타 레이블을 넣어서 좀 더 복잡하게도 만들수 있습니다. <br>\n",
298 |     "중요한건.. 기본형은 이렇다라는 것 입니다 ."
299 |    ]
300 |   },
301 |   {
302 |    "cell_type": "markdown",
303 |    "id": "687296bb-0d44-46df-ae1c-5a81eeb1ae9a",
304 |    "metadata": {},
305 |    "source": [
306 |     "## Meta Labeling \n",
307 |     "\n",
308 |     "Triple-barrier 기반으로 만들어진 모델을 primary model 이라고 하겠습니다. <br>\n",
309 |     "primary model의 문제점은 기술적으로 다음과 같습니다. \n",
310 |     "\n",
311 |     " - 얼마만큼의 베팅을 해야 하는가? -> 모름\n",
312 |     " - binary classification의 문제 -> high true positive rate(Recall) 증가하면 false positive 또한 증가. 또는 그 반대\n",
313 |     "\n",
314 |     "메타 레이블링은.. primary model이 상승한다고 예측한 데이터만 골라서, <br>\n",
315 |     "다시 tripple-barrier label을 통한 second model을 생성 -> 이 경우 TP 와 FP안에서 accuracy를 높입니다. <br>\n",
316 |     "이때 tripple-barrier가 primary model과 동일할 필요는 없으며, 또는 input에 들어가는 데이터를 다르게 해주기도 합니다. \n",
317 |     "\n",
318 |     "\n",
319 |     "베팅 사이즈는 second model에서 나온 확률값으로, 해당 확률만큼 베팅 사이즈를 설정 합니다. <br>\n",
320 |     "실전에서는 primary model은 성능을 높이기 위해서 작은 모델로, 그리고 second 모델은 좀 더 복잡한 모델을 사용 합니다. "
321 |    ]
322 |   }
323 |  ],
324 |  "metadata": {
325 |   "kernelspec": {
326 |    "display_name": "Python 3 (ipykernel)",
327 |    "language": "python",
328 |    "name": "python3"
329 |   },
330 |   "language_info": {
331 |    "codemirror_mode": {
332 |     "name": "ipython",
333 |     "version": 3
334 |    },
335 |    "file_extension": ".py",
336 |    "mimetype": "text/x-python",
337 |    "name": "python",
338 |    "nbconvert_exporter": "python",
339 |    "pygments_lexer": "ipython3",
340 |    "version": "3.8.10"
341 |   }
342 |  },
343 |  "nbformat": 4,
344 |  "nbformat_minor": 5
345 | }
346 | 


--------------------------------------------------------------------------------