├── introduction
├── Intro_to_RL_for_Trading_harmonic_functions.ipynb
└── readme.md
└── odsc_workshop
├── ODSC_RL_1.ipynb
├── ODSC_RL_2.ipynb
├── ODSC_RL_3.ipynb
├── RL in Finance.pdf
└── readme.md
/introduction/Intro_to_RL_for_Trading_harmonic_functions.ipynb:
--------------------------------------------------------------------------------
1 | {
2 | "nbformat": 4,
3 | "nbformat_minor": 0,
4 | "metadata": {
5 | "colab": {
6 | "name": "Intro to RL for Trading: harmonic functions.ipynb",
7 | "provenance": [],
8 | "collapsed_sections": []
9 | },
10 | "kernelspec": {
11 | "name": "python3",
12 | "display_name": "Python 3"
13 | },
14 | "language_info": {
15 | "name": "python"
16 | },
17 | "accelerator": "GPU"
18 | },
19 | "cells": [
20 | {
21 | "cell_type": "code",
22 | "metadata": {
23 | "id": "HjUmgGL28VkP"
24 | },
25 | "source": [
26 | "import numpy as np\n",
27 | "import pandas as pd\n",
28 | "from matplotlib import pyplot as plt"
29 | ],
30 | "execution_count": 2,
31 | "outputs": []
32 | },
33 | {
34 | "cell_type": "code",
35 | "metadata": {
36 | "id": "PgYTJF7FOeyI"
37 | },
38 | "source": [
39 | "import tensorflow as tf\n",
40 | "import tensorflow.keras as tfk"
41 | ],
42 | "execution_count": 3,
43 | "outputs": []
44 | },
45 | {
46 | "cell_type": "markdown",
47 | "metadata": {
48 | "id": "IhO6h_Uc__Wi"
49 | },
50 | "source": [
51 | "# Environment"
52 | ]
53 | },
54 | {
55 | "cell_type": "code",
56 | "metadata": {
57 | "id": "sN_kNTX-AZ2A"
58 | },
59 | "source": [
60 | "class Environment:\n",
61 | " '''\n",
62 | " Simple environment\n",
63 | " '''\n",
64 | " def __init__(self, length = 100, normalize = True, noise = True, data = []):\n",
65 | " self.length = length\n",
66 | "\n",
67 | " if len(data) == 0:\n",
68 | " # self.data = -pd.DataFrame(np.arange(self.length))\n",
69 | " self.data = pd.DataFrame(np.sin(np.arange(length)/30.0))\n",
70 | " else:\n",
71 | " self.data = data\n",
72 | "\n",
73 | " if noise:\n",
74 | " self.data += pd.DataFrame(np.random.normal(0, 0.1, size=(length, )))\n",
75 | "\n",
76 | " if normalize:\n",
77 | " self.data = (self.data - self.data.min()) / (self.data.max() - self.data.min())\n",
78 | "\n",
79 | " def get_state(self, time, lookback, diff = True):\n",
80 | " window = self.data.iloc[time-lookback:time]\n",
81 | " if diff: window = window.diff().fillna(0.0)\n",
82 | " return window\n",
83 | "\n",
84 | " def get_reward(self, action, action_time, reward_time, coef = 100):\n",
85 | " # 0 => long; 1 => hold, 2 => short\n",
86 | " if action == 0:\n",
87 | " action = 1\n",
88 | " elif action == 1:\n",
89 | " action = 0\n",
90 | " else:\n",
91 | " action = -1\n",
92 | " price_now = self.data.iloc[action_time]\n",
93 | " price_reward = self.data.iloc[reward_time]\n",
94 | " price_diff = (price_reward - price_now) / price_now\n",
95 | " reward = np.sign(price_diff) * action * coef\n",
96 | "\n",
97 | " return reward.values.tolist()[0]"
98 | ],
99 | "execution_count": 5,
100 | "outputs": []
101 | },
102 | {
103 | "cell_type": "code",
104 | "metadata": {
105 | "id": "zrbu-JaeKu1V"
106 | },
107 | "source": [
108 | "lin_env = Environment()"
109 | ],
110 | "execution_count": 6,
111 | "outputs": []
112 | },
113 | {
114 | "cell_type": "code",
115 | "metadata": {
116 | "colab": {
117 | "base_uri": "https://localhost:8080/",
118 | "height": 204
119 | },
120 | "id": "3tOp5fKwKxOW",
121 | "outputId": "a3bb3a15-f9f1-40b6-da1f-4240ed8c89b9"
122 | },
123 | "source": [
124 | "lin_env.get_state(10, 5, True)"
125 | ],
126 | "execution_count": 7,
127 | "outputs": [
128 | {
129 | "output_type": "execute_result",
130 | "data": {
131 | "text/html": [
132 | "\n",
133 | "\n",
146 | "
\n",
147 | " \n",
148 | " \n",
149 | " | \n",
150 | " 0 | \n",
151 | "
\n",
152 | " \n",
153 | " \n",
154 | " \n",
155 | " | 5 | \n",
156 | " 0.000000 | \n",
157 | "
\n",
158 | " \n",
159 | " | 6 | \n",
160 | " -0.053434 | \n",
161 | "
\n",
162 | " \n",
163 | " | 7 | \n",
164 | " 0.159560 | \n",
165 | "
\n",
166 | " \n",
167 | " | 8 | \n",
168 | " -0.069024 | \n",
169 | "
\n",
170 | " \n",
171 | " | 9 | \n",
172 | " 0.098430 | \n",
173 | "
\n",
174 | " \n",
175 | "
\n",
176 | "