File size: 8,700 Bytes
174772e |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 |
import gym
from gym import spaces
from gym.utils import seeding
import numpy as np
from enum import Enum
import matplotlib.pyplot as plt
class Actions(Enum):
Sell = 0
Buy = 1
Do_nothing = 2
class TradingEnv(gym.Env):
metadata = {'render.modes': ['human']}
def __init__(self, df, window_size, frame_bound):
assert df.ndim == 2
assert len(frame_bound) == 2
self.frame_bound = frame_bound
self.seed()
self.df = df
self.window_size = window_size
self.prices, self.signal_features = self._process_data()
self.shape = (window_size, self.signal_features.shape[1])
# spaces
self.action_space = spaces.Discrete(len(Actions))
self.observation_space = spaces.Box(low=-np.inf, high=np.inf, shape=self.shape, dtype=np.float64)
# episode
self._start_tick = self.window_size
self._end_tick = len(self.prices) - 1
self._done = None
self._current_tick = None
self._last_trade_tick = None
self._position = None
self._position_history = None
self._total_reward = None
self._total_profit = None
self._first_rendering = None
self.history = None
# fees
self.trade_fee_bid_percent = 0.0005 # unit
self.trade_fee_ask_percent = 0.0005 # unit
def seed(self, seed=None):
self.np_random, seed = seeding.np_random(seed)
return [seed]
def reset(self):
self._done = False
self._current_tick = self._start_tick
self._last_trade_tick = self._current_tick - 1
self._position = 0
self._position_history = (self.window_size * [None])
# self._position_history = (self.window_size * [None]) + [self._position]
self._total_reward = 0.
self._total_profit = 0.
self.history = {}
return self._get_observation()
def _calculate_reward(self, action):
step_reward = 0
current_price = self.prices[self._current_tick]
last_price = self.prices[self._current_tick - 1]
price_diff = current_price - last_price
# OPEN BUY - 1
if action == Actions.Buy.value and self._position == 0:
self._position = 1
step_reward += price_diff
self._last_trade_tick = self._current_tick - 1
self._position_history.append(1)
elif action == Actions.Buy.value and self._position > 0:
step_reward += 0
self._position_history.append(-1)
# CLOSE SELL - 4
elif action == Actions.Buy.value and self._position < 0:
self._position = 0
step_reward += -1 * (self.prices[self._current_tick -1] - self.prices[self._last_trade_tick])
self._total_profit += step_reward
self._position_history.append(4)
# OPEN SELL - 3
elif action == Actions.Sell.value and self._position == 0:
self._position = -1
step_reward += -1 * price_diff
self._last_trade_tick = self._current_tick - 1
self._position_history.append(3)
# CLOSE BUY - 2
elif action == Actions.Sell.value and self._position > 0:
self._position = 0
step_reward += self.prices[self._current_tick -1] - self.prices[self._last_trade_tick]
self._total_profit += step_reward
self._position_history.append(2)
elif action == Actions.Sell.value and self._position < 0:
step_reward += 0
self._position_history.append(-1)
# DO NOTHING - 0
elif action == Actions.Do_nothing.value and self._position > 0:
step_reward += price_diff
self._position_history.append(0)
elif action == Actions.Do_nothing.value and self._position < 0:
step_reward += -1 * price_diff
self._position_history.append(0)
elif action == Actions.Do_nothing.value and self._position == 0:
step_reward += -1 * abs(price_diff)
self._position_history.append(0)
return step_reward
def step(self, action):
self._done = False
self._current_tick += 1
if self._current_tick == self._end_tick:
self._done = True
step_reward = self._calculate_reward(action)
self._total_reward += step_reward
observation = self._get_observation()
info = dict(
total_reward = self._total_reward,
total_profit = self._total_profit,
position = self._position
)
self._update_history(info)
return observation, step_reward, self._done, info
def _get_observation(self):
return self.signal_features[(self._current_tick-self.window_size+1):self._current_tick+1]
def _update_history(self, info):
if not self.history:
self.history = {key: [] for key in info.keys()}
for key, value in info.items():
self.history[key].append(value)
def render(self, mode='human'):
window_ticks = np.arange(len(self._position_history))
plt.plot(self.prices)
open_buy = []
close_buy = []
open_sell = []
close_sell = []
do_nothing = []
for i, tick in enumerate(window_ticks):
if self._position_history[i] is None:
continue
if self._position_history[i] == 1:
open_buy.append(tick)
elif self._position_history[i] == 2 :
close_buy.append(tick)
elif self._position_history[i] == 3 :
open_sell.append(tick)
elif self._position_history[i] == 4 :
close_sell.append(tick)
elif self._position_history[i] == 0 :
do_nothing.append(tick)
plt.plot(open_buy, self.prices[open_buy], 'go', marker="^")
plt.plot(close_buy, self.prices[close_buy], 'go', marker="v")
plt.plot(open_sell, self.prices[open_sell], 'ro', marker="v")
plt.plot(close_sell, self.prices[close_sell], 'ro', marker="^")
plt.plot(do_nothing, self.prices[do_nothing], 'yo')
plt.suptitle(
"Total Reward: %.6f" % self._total_reward + ' ~ ' +
"Total Profit: %.6f" % self._total_profit
)
def close(self):
plt.close()
def save_rendering(self, filepath):
plt.savefig(filepath)
def pause_rendering(self):
plt.show()
def _process_data(self):
prices = self.df.loc[:, 'Close'].to_numpy()
prices[self.frame_bound[0] - self.window_size] # validate index (TODO: Improve validation)
prices = prices[self.frame_bound[0]-self.window_size:self.frame_bound[1]]
diff = np.insert(np.diff(prices), 0, 0)
signal_features = np.column_stack((prices, diff))
return prices, signal_features
def _update_profit(self, action):
trade = False
if ((action == Actions.Buy.value and self._position == Positions.Short) or
(action == Actions.Sell.value and self._position == Positions.Long)):
trade = True
if trade or self._done:
current_price = self.prices[self._current_tick]
last_trade_price = self.prices[self._last_trade_tick]
if self._position == Positions.Long:
shares = (self._total_profit * (1 - self.trade_fee_ask_percent)) / last_trade_price
self._total_profit = (shares * (1 - self.trade_fee_bid_percent)) * current_price
def max_possible_profit(self):
current_tick = self._start_tick
last_trade_tick = current_tick - 1
profit = 1.
while current_tick <= self._end_tick:
position = None
if self.prices[current_tick] < self.prices[current_tick - 1]:
while (current_tick <= self._end_tick and
self.prices[current_tick] < self.prices[current_tick - 1]):
current_tick += 1
position = Positions.Short
else:
while (current_tick <= self._end_tick and
self.prices[current_tick] >= self.prices[current_tick - 1]):
current_tick += 1
position = Positions.Long
if position == Positions.Long:
current_price = self.prices[current_tick - 1]
last_trade_price = self.prices[last_trade_tick]
shares = profit / last_trade_price
profit = shares * current_price
last_trade_tick = current_tick - 1
return profit
|