# 학습 데이터 분리
features_training_data = [
'open_lastclose_ratio', 'high_close_ratio', 'low_close_ratio',
'close_lastclose_ratio', 'volume_lastvolume_ratio', 'close_ma5_ratio',
'volume_ma5_ratio', 'close_ma10_ratio', 'volume_ma10_ratio',
'close_ma20_ratio', 'volume_ma20_ratio', 'close_ma60_ratio',
'volume_ma60_ratio', 'close_ma120_ratio', 'volume_ma120_ratio'
]
training_data = training_data[features_training_data]
# 강화학습 시작
policy_learner = PolicyLearner(stock_code=stock_code,
chart_data=chart_data,
training_data=training_data,
min_trading_unit=1,
max_trading_unit=2,
delayed_reward_threshold=.2,
lr=.001)
policy_learner.fit(balance=10000000,
num_epoches=1000,
discount_factor=0,
start_epsilon=.5)
# 정책 신경망을 파일로 저장
model_dir = os.path.join(settings.BASE_DIR, 'models/%s' % stock_code)
if not os.path.isdir(model_dir):
os.makedirs(model_dir)
model_path = os.path.join(model_dir, 'model_%s.h5' % timestr)
policy_learner.policy_network.save_model(model_path)
# 기간 필터링
training_data = training_data[(training_data['date'] >= '2018-01-01') &
(training_data['date'] <= '2018-01-31')]
training_data = training_data.dropna()
# 차트 데이터 분리
features_chart_data = ['date', 'open', 'high', 'low', 'close', 'volume']
chart_data = training_data[features_chart_data]
# 학습 데이터 분리
features_training_data = [
'open_lastclose_ratio', 'high_close_ratio', 'low_close_ratio',
'close_lastclose_ratio', 'volume_lastvolume_ratio',
'close_ma5_ratio', 'volume_ma5_ratio',
'close_ma10_ratio', 'volume_ma10_ratio',
'close_ma20_ratio', 'volume_ma20_ratio',
'close_ma60_ratio', 'volume_ma60_ratio',
'close_ma120_ratio', 'volume_ma120_ratio'
]
training_data = training_data[features_training_data]
# 비 학습 투자 시뮬레이션 시작
policy_learner = PolicyLearner(
stock_code=stock_code, chart_data=chart_data, training_data=training_data,
min_trading_unit=1, max_trading_unit=3)
policy_learner.trade(balance=10000000,
model_path=os.path.join(
settings.BASE_DIR,
'models/{}/model_{}.h5'.format(stock_code, model_ver)))
parser.add_argument('COIN', type=str, help="coin type?")
parser.add_argument('BALANCE', type=int, help="initial balance?")
args = parser.parse_args()
MODEL = args.MODEL
COIN = args.COIN
BALANCE = args.BALANCE
log_dir = os.path.join(settings.BASE_DIR, 'logs/%s' % COIN)
timestr = settings.get_time_str()
file_handler = logging.FileHandler(filename=os.path.join(
log_dir, "%s_%s.log" % (COIN, timestr)),
encoding='utf-8')
stream_handler = logging.StreamHandler()
file_handler.setLevel(logging.DEBUG)
stream_handler.setLevel(logging.INFO)
logging.basicConfig(format="%(message)s",
handlers=[file_handler, stream_handler],
level=logging.DEBUG)
policy_learner = PolicyLearner(coin_code=COIN,
coin_chart=None,
training_data=None,
min_trading_unit=1,
max_trading_unit=3)
policy_learner.trade(balance=BALANCE,
model_path=os.path.join(
settings.BASE_DIR,
'models/{}/model_{}.h5'.format(COIN, MODEL)))
'close_lastclose_ratio', 'volume_lastvolume_ratio', 'close_ma5_ratio',
'volume_ma5_ratio', 'close_ma10_ratio', 'volume_ma10_ratio',
'close_ma20_ratio', 'volume_ma20_ratio', 'close_ma60_ratio',
'volume_ma60_ratio', 'close_ma120_ratio', 'volume_ma120_ratio'
]
training_data = training_data[features_training_data]
print("coin_len", len(coin_chart))
print("train_len", len(training_data))
print(FROM, TO)
policy_learner = PolicyLearner(coin_code=COIN,
coin_chart=coin_chart,
training_data=training_data,
model_ver=model_ver,
min_trading_unit=0.001,
max_trading_unit=0.01,
delayed_reward_threshold=.3,
start_date=FROM,
end_date=TO,
lr=.09)
print("policy learner start")
policy_learner.fit(balance=1000000,
num_epoches=10,
discount_factor=0,
start_epsilon=.5)
model_dir = os.path.join(settings.BASE_DIR, 'models/%s' % COIN)
if not os.path.isdir(model_dir):
os.makedirs(model_dir)
if not os.path.exists('logs/%s' % STOCK_CODE):
os.makedirs('logs/%s' % STOCK_CODE)
file_handler = logging.FileHandler(filename=os.path.join(
log_dir, "%s_%s.log" % (STOCK_CODE, timestr)), encoding='utf-8')
stream_handler = logging.StreamHandler() # send logging output to stdout
file_handler.setLevel(logging.DEBUG)
stream_handler.setLevel(logging.INFO)
logging.basicConfig(format="%(message)s",
handlers=[file_handler, stream_handler], level=logging.DEBUG)
# 강화학습 시작
if LEARNING:
chart_data, data = prepare_data(STOCK_CODE, MARKET_CODE, TRAINING_START_DATE, TRAINING_END_DATE)
policy_learner = PolicyLearner(
stock_code=STOCK_CODE, chart_data=chart_data, training_data=data,
min_trading_unit=MIN_TRADING_UNIT, max_trading_unit=MAX_TRADING_UNIT,
delayed_reward_threshold=DELAYED_REWARD_THRESHOLD, lr=LEARNING_RATE)
policy_learner.fit(balance=INITIAL_BALANCE, num_epoches=NUM_EPOCHS,max_memory=MAX_MEMORY,
discount_factor=DISCOUNT_FACTOR, start_epsilon=START_EPSILON, learning=LEARNING_RATE)
# 정책 신경망을 파일로 저장
model_dir = os.path.join(settings.BASE_DIR, 'models/%s' % STOCK_CODE)
if not os.path.isdir(model_dir):
os.makedirs(model_dir)
#model_path = os.path.join(model_dir, 'model_%s.h5' % timestr)
model_path = os.path.join(model_dir, 'model_%s.h5' % STOCK_CODE)
policy_learner.policy_network.save_model(model_path)
if SIMULATION:
chart_data, data = prepare_data(STOCK_CODE, MARKET_CODE, SIMULATION_START_DATE, SIMULATION_END_DATE)
#'close_ma10_ratio', 'volume_ma10_ratio',
#'close_ma20_ratio', 'volume_ma20_ratio',
#'close_ma60_ratio', 'volume_ma60_ratio',
#'close_ma120_ratio', 'volume_ma120_ratio'
]
training_data = training_data[features_training_data]
# 강화학습 시작
policy_learner = PolicyLearner(
stock_code=stock_code,
chart_data=chart_data,
training_data=training_data,
policy_model_path=os.path.join(
settings.BASE_DIR,
'models/{}/model_{}.h5'.format(model_code, policy_model_ver)),
value_model_path=os.path.join(
settings.BASE_DIR,
'models/{}/model_{}.h5'.format(model_code, value_model_ver)),
lr=0.00000001,
discount_factor=0,
start_epsilon=0,
num_past_input=119,
load_weight_and_learn=False)
# 여기가 핵심
policy_learner.fit(balance=1, num_epoches=1000)
# 정책 신경망을 파일로 저장
model_dir = os.path.join(settings.BASE_DIR, 'models/%s' % model_code)
if not os.path.isdir(model_dir):
os.makedirs(model_dir)
features_chart_data = ['date', 'open', 'high', 'low', 'close', 'volume']
chart_data = training_data[features_chart_data]
# 학습 데이터 분리
features_training_data = [
'high_low_ratio', 'open_close_ratio',
'high_open_ratio', 'low_open_ratio',
'high_close_ratio', 'low_close_ratio',
'close_lastclose_ratio', 'volume_lastvolume_ratio'
]
training_data = training_data[features_training_data]
# 비 학습 투자 시뮬레이션 시작
if isFirst == True:
policy_learner = PolicyLearner(
stock_code=stock_code, chart_data=chart_data, training_data=training_data,
lr=0.00000001, discount_factor=0, start_epsilon=0, num_past_input=119)
isFirst = False
else:
policy_learner.stock_code = stock_code
policy_learner.chart_data = chart_data
policy_learner.training_data = training_data
policy_learner.trade(balance=20000,
model_path=os.path.join(
settings.BASE_DIR,
'models/{}/model_{}.h5'.format(model_code, model_ver)))
buy_probs.append(policy_learner.action[0])
buy_probs = np.array(buy_probs)
if np.max(buy_probs) > 0.5: # 매수하기
depth = client.get_order_book(symbol='{}BTC'.format(learned_coins[np.argmax(buy_probs)]))
def learnFunc(self):
if self.code is None or self.df is None:
return
self.change_value.emit(ZERO)
# 데이터 전처리
code = self.code
chart_data = self.df
prep_data = data_manager.preprocess(chart_data)
training_data = data_manager.build_training_data(prep_data)
training_data = training_data.dropna()
# 차트데이터 분리
feature_chart_data = ['date', 'open', 'high', 'low', 'close', 'volume']
chart_data = training_data[feature_chart_data]
# emit
self.change_value.emit(TWENTY_FIVE)
# 학습데이터 분리
feature_chart_data = [
'open_lastclose_ratio',
'high_close_ratio',
'low_close_ratio',
'close_lastclose_ratio',
'volume_lastvolume_ratio',
'close_ma5_ratio',
'volume_ma5_ratio',
'close_ma10_ratio',
'volume_ma10_ratio',
'close_ma20_ratio',
'volume_ma20_ratio',
'close_ma60_ratio',
'volume_ma60_ratio',
'close_ma120_ratio',
'volume_ma120_ratio',
]
training_data = training_data[feature_chart_data]
# 정책 신경망을 파일로 저장
self.createFolder('model')
mdir = os.path.join(settings.BASE_DIR, 'model')
self.createFolder(os.path.join(mdir, code))
model_dir = os.path.join(mdir, code)
model_path = os.path.join(model_dir, 'model%s.h5' % code)
# model_path 경로가 없으면 학습모델을 해당 dir에 만들어서 학습
# model_path가 있으면 해당 모델 선택 후 예측
print(model_path)
# emit
self.change_value.emit(FIFTY)
if not os.path.isfile(model_path):
start_time = time.time()
policy_learner = PolicyLearner(stock_code=code,
chart_data=chart_data,
training_data=training_data,
fig=self.fig,
canvas=self.canvas,
min_trading_unit=1,
max_trading_unit=2,
delayed_reward_threshold=0.2,
lr=0.001)
policy_learner.fit(balance=10000000,
num_epoches=200,
discount_factor=0,
start_epsilon=0.5)
end_time = time.time()
policy_learner.policy_network.save_model(model_path)
print("LearningTime: {} sec".format(end_time - start_time))
else:
start_time = time.time()
policy_learner = PolicyLearner(stock_code=code,
chart_data=chart_data,
training_data=training_data,
fig=self.fig,
canvas=self.canvas,
min_trading_unit=1,
max_trading_unit=2)
end_time = time.time()
print("LearningTime: {} sec".format(end_time - start_time))
policy_learner.trade(balance=1000000,
model_path=os.path.join(
model_dir, 'model%s.h5' % (code)))
# emit
self.change_value.emit(A_HUNDRED)
training_data = data_manager.build_training_data(prep_data)
training_data = training_data[(training_data['date'] >= '2018-07-01')
& (training_data['date'] <= '2018-10-01')]
training_data = training_data.dropna()
features_chart_data = ['date', 'open', 'high', 'low', 'close', 'volume']
coin_chart = training_data[features_chart_data]
features_training_data = [
'open_lastclose_ratio', 'high_close_ratio', 'low_close_ratio',
'close_lastclose_ratio', 'volume_lastvolume_ratio', 'close_ma5_ratio',
'volume_ma5_ratio', 'close_ma10_ratio', 'volume_ma10_ratio',
'close_ma20_ratio', 'volume_ma20_ratio', 'close_ma60_ratio',
'volume_ma60_ratio', 'close_ma120_ratio', 'volume_ma120_ratio'
]
training_data = training_data[features_training_data]
policy_learner = PolicyLearner(coin_code=coin_code,
coin_chart=coin_chart,
training_data=training_data,
min_trading_unit=1,
max_trading_unit=3)
policy_learner.trade(balance=10000000,
model_path=os.path.join(
settings.BASE_DIR, 'models/{}/model_{}.h5'.format(
coin_code, model_ver)))
# exploration=True)
# Date range filtering
training_data = training_data[(training_data['date'] >= '2019-01-01')
& (training_data['date'] <= '2019-12-31')]
training_data = training_data.dropna()
# Chart Data Separation
features_chart_data = ['date', 'open', 'high', 'low', 'close', 'volume']
chart_data = training_data[features_chart_data]
# Training data separation
features_training_data = [
'open_lastclose_ratio', 'high_close_ratio', 'low_close_ratio',
'close_lastclose_ratio', 'volume_lastvolume_ratio', 'close_ma5_ratio',
'volume_ma5_ratio', 'close_ma10_ratio', 'volume_ma10_ratio',
'close_ma20_ratio', 'volume_ma20_ratio'
]
training_data = training_data[features_training_data]
# Start non-training investment simulation
policy_learner = PolicyLearner(stock_code=stock_code,
chart_data=chart_data,
training_data=training_data,
min_trading_unit=1,
max_trading_unit=3,
delayed_reward_threshold=reward,
tax=tax)
policy_learner.trade(balance=bal,
model_path=os.path.join(
settings.BASE_DIR, 'models/{}/model_{}.h5'.format(
stock_code, model_ver)))
'close_lastclose_ratio', 'volume_lastvolume_ratio',
'close_ma5_ratio', 'volume_ma5_ratio',
'close_ma10_ratio', 'volume_ma10_ratio',
'close_ma20_ratio', 'volume_ma20_ratio',
'close_ma60_ratio', 'volume_ma60_ratio',
'close_ma120_ratio', 'volume_ma120_ratio',
'ema12','ema26','dn','mavg','up','pctB','macd','signal','cci'
]
training_data = training_data[features_training_data]
#print (training_data[:3])
training_data = training_data.dropna(axis=1)
chart_data=chart_data.dropna(axis=1)
chart_data = chart_data.loc[:'2018-06-30 23:00:00']
training_data = training_data.loc[:'2018-06-30 23:00:00']
delayed_reward_threshold=0.2
policy_learner = PolicyLearner(
stock_code=stock_code, chart_data=chart_data, training_data=training_data, delayed_reward_threshold=drt, lr=.1)
policy_learner.fit(balance=10000, num_epoches=10,discount_factor=0, start_epsilon=.5)
model_dir = os.path.join(settings.BASE_DIR, 'models/%s' % stock_code)
if not os.path.isdir(model_dir):
os.makedirs(model_dir)
model_path = os.path.join(model_dir, 'model_%s.h5' % timestr)
policy_learner.policy_network.save_model(model_path)
buycnt=0
sellcnt=0
confmat=[[0,0],[0,0]]
class TestStrategy(bt.Strategy):
features_chart_data = ['date', 'open', 'high', 'low', 'close', 'volume']
coin_chart = training_data[features_chart_data]
features_training_data = [
'open_lastclose_ratio', 'high_close_ratio', 'low_close_ratio',
'close_lastclose_ratio', 'volume_lastvolume_ratio', 'close_ma5_ratio',
'volume_ma5_ratio', 'close_ma10_ratio', 'volume_ma10_ratio',
'close_ma20_ratio', 'volume_ma20_ratio', 'close_ma60_ratio',
'volume_ma60_ratio', 'close_ma120_ratio', 'volume_ma120_ratio'
]
training_data = training_data[features_training_data]
policy_learner = PolicyLearner(coin_code=coin_code,
coin_chart=coin_chart,
training_data=training_data,
min_trading_unit=0.01,
max_trading_unit=3,
delayed_reward_threshold=.1,
lr=.001)
policy_learner.fit(balance=10000000,
num_epoches=5,
discount_factor=0,
start_epsilon=.5)
model_dir = os.path.join(settings.BASE_DIR, 'models/%s' % coin_code)
if not os.path.isdir(model_dir):
os.makedirs(model_dir)
model_path = os.path.join(model_dir, 'model_%s.h5' % timestr)
policy_learner.policy_network.save_model(model_path)
coin_chart = data_manager.load_chart_data(os.path.join(settings.BASE_DIR,'data/chart_data/{}.csv'.format(coin_code)))
prep_data = data_manager.preprocess_min(coin_chart)
training_data = data_manager.build_training_data(prep_data)
start_date = '2018-10-13 00:00:00'
end_date = '2018-10-14 00:00:00'
training_data = training_data[(training_data['date'] >= start_date)&(training_data['date'] < end_date)]
training_data = training_data.dropna()
features_chart_data = ['date', 'open', 'high', 'low', 'close', 'volume']
coin_chart = training_data[features_chart_data]
features_training_data = [
'open_lastclose_ratio', 'high_close_ratio', 'low_close_ratio',
'close_lastclose_ratio', 'volume_lastvolume_ratio',
'close_ma5_ratio', 'volume_ma5_ratio',
'close_ma10_ratio', 'volume_ma10_ratio',
'close_ma20_ratio', 'volume_ma20_ratio',
'close_ma60_ratio', 'volume_ma60_ratio',
'close_ma120_ratio', 'volume_ma120_ratio']
training_data = training_data[features_training_data]
policy_learner = PolicyLearner(
coin_code=COIN, coin_chart=coin_chart, training_data=training_data,
min_trading_unit=0.001, max_trading_unit=0.01,start_date=start_date, end_date=end_date)
policy_learner.trade(balance=BALANCE,
model_path=os.path.join(settings.BASE_DIR, 'models/{}/model_{}.h5'.format(COIN,MODEL)))
