def calculate(self, stock_code, date, short=12, long=26, mid=9):
dif = 0.0
dea = 0.0
macd = 0.0
max_time_period = max(short, long, mid)
data = su.get_basic_data(stock_code, date, max_time_period *
3).sort_index(ascending=False)
if data.shape[0] >= max_time_period * 3:
close = data['CLOSE'].values
ewma_short = pd.ewma(close, span=short)
ewma_long = pd.ewma(close, span=long)
difs = (ewma_short - ewma_long)
deas = pd.ewma(difs, span=mid)
macds = (difs - deas) * 2
dif = round(difs[-1], 3)
dea = round(deas[-1], 3)
macd = round(macds[-1], 3)
if np.isnan(dif) or np.isinf(dif) or np.isneginf(dif):
dif = 0.0
if np.isnan(dea) or np.isinf(dea) or np.isneginf(dea):
dea = 0.0
if np.isnan(macd) or np.isinf(macd) or np.isneginf(macd):
macd = 0.0
self.save_tech_data(stock_code, date, {
'MACD_DIF': dif,
'MACD_DEA': dea,
'MACD': macd
})
return dif, dea, macd
def calculate(self,
stock_code,
date,
time_period=20,
nbdev_up=2,
nbdev_down=2):
upper_brand = 0.0
middle_brand = 0.0
lower_brand = 0.0
data = su.get_basic_data(stock_code, date,
time_period + 1).sort_index(ascending=False)
if data.shape[0] >= time_period:
upper_brands, middle_brands, lower_brands = ta.BBANDS(
data['CLOSE'].as_matrix(), time_period, nbdev_up, nbdev_down)
upper_brand = round(upper_brands[-1], 3)
middle_brand = round(middle_brands[-1], 3)
lower_brand = round(lower_brands[-1], 3)
if np.isinf(upper_brand) or np.isnan(upper_brand) or np.isneginf(
upper_brand):
upper_brand = 0.0
if np.isinf(middle_brand) or np.isnan(middle_brand) or np.isneginf(
middle_brand):
middle_brand = 0.0
if np.isinf(lower_brand) or np.isnan(lower_brand) or np.isneginf(
lower_brand):
lower_brand = 0.0
self.save_tech_data(
stock_code, date, {
'BOLL_UP': upper_brand,
'BOLL_LOW': lower_brand,
'BOLL_MID': middle_brand
})
return upper_brand, middle_brand, lower_brand
def calculate(self, stock_code, date, time_period1=6, time_period2=12, time_period3=24):
bias1 = 0.0
bias2 = 0.0
bias3 = 0.0
max_time_period = max(time_period1, time_period2, time_period3)
#print max_time_period
data = su.get_basic_data(stock_code, date, max_time_period + 1).sort_index(ascending=False)
if data.shape[0] >= max_time_period + 1:
data['MA1'] = data['CLOSE'].rolling(window=time_period1).mean()
data['MA2'] = data['CLOSE'].rolling(window=time_period2).mean()
data['MA3'] = data['CLOSE'].rolling(window=time_period3).mean()
close = data['CLOSE'].as_matrix()[-1]
data['BIAS1'] = (close - data['MA1']) * 100 / data['MA1']
data['BIAS2'] = (close - data['MA2']) * 100 / data['MA2']
data['BIAS3'] = (close - data['MA3']) * 100 / data['MA3']
bias1 = round(data['BIAS1'].as_matrix()[-1], 3)
bias2 = round(data['BIAS2'].as_matrix()[-1], 3)
bias3 = round(data['BIAS3'].as_matrix()[-1], 3)
#print data
#print data
if np.isinf(bias1) or np.isnan(bias1) or np.isneginf(bias1):
bias1 = 0.0
if np.isinf(bias2) or np.isnan(bias2) or np.isneginf(bias2):
bias2 = 0.0
if np.isinf(bias3) or np.isnan(bias3) or np.isneginf(bias3):
bias3 = 0.0
self.save_tech_data(stock_code, date, {'BIAS_6': bias1, 'BIAS_12':bias2, 'BIAS_24':bias3})
return bias1, bias2, bias3
def calculate(self,
stock_code,
date,
time_period1=3,
time_period2=6,
time_period3=12,
time_period4=24):
bbi = 0.0
max_time_period = max(time_period1, time_period2, time_period3,
time_period4)
data = su.get_basic_data(stock_code, date, max_time_period +
1).sort_index(ascending=False)
if data.shape[0] >= max_time_period + 1:
data['MA1'] = data['CLOSE'].rolling(window=time_period1).mean()
data['MA2'] = data['CLOSE'].rolling(window=time_period2).mean()
data['MA3'] = data['CLOSE'].rolling(window=time_period3).mean()
data['MA4'] = data['CLOSE'].rolling(window=time_period4).mean()
data['BBI'] = (data['MA1'] + data['MA2'] + data['MA3'] +
data['MA4']) / 4
bbi = round(data['BBI'].as_matrix()[-1], 3)
if np.isnan(bbi) or np.isinf(bbi) or np.isneginf(bbi):
bbi = 0.0
self.save_tech_data(stock_code, date, {'BBI': bbi})
return bbi
def calculate(self,
stock_code,
date,
time_period1=5,
time_period2=10,
time_period3=20):
ma_1 = 0.0
ma_2 = 0.0
ma_3 = 0.0
max_time_period = max(time_period1, time_period2, time_period3)
data = su.get_basic_data(stock_code, date, max_time_period +
1).sort_index(ascending=False)
if data.shape[0] >= max_time_period + 1:
data['MA1'] = data['CLOSE'].rolling(window=time_period1).mean()
data['MA2'] = data['CLOSE'].rolling(window=time_period2).mean()
data['MA3'] = data['CLOSE'].rolling(window=time_period3).mean()
ma_1 = round(data['MA1'].as_matrix()[-1], 3)
ma_2 = round(data['MA2'].as_matrix()[-1], 3)
ma_3 = round(data['MA3'].as_matrix()[-1], 3)
if np.isnan(ma_1) or np.isinf(ma_1) or np.isneginf(ma_1):
ma_1 = 0.0
if np.isnan(ma_2) or np.isinf(ma_2) or np.isneginf(ma_2):
ma_2 = 0.0
if np.isnan(ma_3) or np.isinf(ma_3) or np.isneginf(ma_3):
ma_3 = 0.0
self.save_tech_data(stock_code, date, {
'MA_5': ma_1,
'MA_10': ma_2,
'MA_20': ma_3
})
return ma_1, ma_2, ma_3
def calculate(self, stock_code, date, time_period1=12, time_period2=6):
psy_12 = 0.0
psy_6 = 0.0
data = su.get_basic_data(stock_code, date,
time_period1 + 1).sort_index(ascending=False)
if data.shape[0] >= time_period1:
count = 0.0
data['P_CLOSE'] = data['CLOSE'].shift(1)
data['DIFF'] = data['CLOSE'] - data['P_CLOSE']
data['DIFF'].fillna(0, inplace=True)
#print data
for each in data[1:].itertuples():
if each.DIFF > 0:
count += 1.0
psy_12 = round((count / time_period1 * 100), 3)
count = 0.0
for each in data[-6:].itertuples():
if each.DIFF > 0:
count += 1.0
psy_6 = round((count / time_period2 * 100), 3)
#print data
if np.isnan(psy_12) or np.isinf(psy_12) or np.isneginf(psy_12):
psy_12 = 0.0
if np.isnan(psy_6) or np.isinf(psy_6) or np.isneginf(psy_6):
psy_6 = 0.0
self.save_tech_data(stock_code, date, {
'PSY_12': psy_12,
'PSY_6': psy_6
})
return psy_12, psy_6
def calculate(self,
stock_code,
date,
time_period1=6,
time_period2=14,
time_period3=24):
rsi_6 = 0.0
rsi_14 = 0.0
rsi_24 = 0.0
data = su.get_basic_data(stock_code, date,
time_period3 + 1).sort_index(ascending=False)
if data.shape[0] >= time_period3 + 1:
rsi_6 = round(
ta.RSI(data['CLOSE'].as_matrix(), time_period1)[-1], 3)
rsi_14 = round(
ta.RSI(data['CLOSE'].as_matrix(), time_period2)[-1], 3)
rsi_24 = round(
ta.RSI(data['CLOSE'].as_matrix(), time_period3)[-1], 3)
if np.isnan(rsi_6) or np.isinf(rsi_6) or np.isneginf(rsi_6):
rsi_6 = 0.0
if np.isnan(rsi_14) or np.isinf(rsi_14) or np.isneginf(rsi_14):
rsi_14 = 0.0
if np.isnan(rsi_24) or np.isinf(rsi_24) or np.isneginf(rsi_24):
rsi_24 = 0.0
self.save_tech_data(stock_code, date, {
'RSI_6': rsi_6,
'RSI_14': rsi_14,
'RSI_24': rsi_24
})
return rsi_6, rsi_14, rsi_24
def calculate(self,
stock_code,
date,
time_period1=10,
time_period2=50,
time_period3=6):
dma = 0.0
ama = 0.0
max_time_period = max(time_period1, time_period2, time_period3)
data = su.get_basic_data(stock_code, date, max_time_period +
time_period3).sort_index(ascending=False)
if data.shape[0] >= max_time_period + 1:
data['MA1'] = data['CLOSE'].rolling(window=time_period1).mean()
data['MA2'] = data['CLOSE'].rolling(window=time_period2).mean()
data['DMA'] = data['MA1'] - data['MA2']
data['AMA'] = data['DMA'].rolling(window=time_period3).mean()
dma = round(data['DMA'].as_matrix()[-1], 3)
ama = round(data['AMA'].as_matrix()[-1], 3)
if np.isnan(dma) or np.isinf(dma) or np.isneginf(dma):
dma = 0.0
if np.isnan(ama) or np.isinf(ama) or np.isneginf(ama):
ama = 0.0
self.save_tech_data(stock_code, date, {'DMA': dma, 'AMA': ama})
return dma, ama
def calculate(self, stock_code, date, time_period=14):
cci = 0.0
data = su.get_basic_data(stock_code, date, time_period + 1).sort_index(ascending=False)
if data.shape[0] >= time_period + 1:
cci = round(ta.CCI(data['HIGH'].as_matrix(), data['LOW'].as_matrix(), data['CLOSE'].as_matrix(), time_period)[-1], 3)
if np.isnan(cci) or np.isinf(cci) or np.isneginf(cci):
cci = 0.0
self.save_tech_data(stock_code, date, {'CCI':cci})
return cci
def calculate(self, stock_code, date, time_period1=14, time_period2=6):
roc = 0.0
maroc = 0.0
data = su.get_basic_data(stock_code, date,
time_period1 * 2).sort_index(ascending=False)
if data.shape[0] >= time_period1:
rocs = ta.ROC(data['CLOSE'].as_matrix(), time_period1)
roc = round(rocs[-1], 3)
maroc = round(rocs[-6:].sum() / float(time_period2), 3)
if np.isnan(roc) or np.isinf(roc) or np.isneginf(roc):
roc = 0.0
if np.isnan(maroc) or np.isinf(maroc) or np.isneginf(maroc):
maroc = 0.0
self.save_tech_data(stock_code, date, {'ROC': roc, 'MAROC': maroc})
return roc, maroc
def download_basic_data(self, stock_code, start_date, end_date):
data = su.get_hist_data(stock_code, start_date, end_date)
if data is None:
logger.warn(
"No basic data is downloaded from 163 for stock code " +
stock_code + ", try from tushare now...")
print("No basic data is downloaded from 163 for stock code " +
stock_code + ", try from tushare now...")
data = ts.get_hist_data(stock_code, start_date, end_date)
if not 'turnover' in data.columns:
logger.warn("Tushare data without turnover data")
data.insert(loc=2, column="turnover", value=0)
if data is not None:
data = data.sort_index(ascending=True)[1:]
data['code'] = pd.Series(stock_code, index=data.index)
return data
def calculate(self,
stock_code,
date,
time_period1=9,
time_period2=3,
time_period3=3):
kdj_k = 0.0
kdj_d = 0.0
kdj_j = 0.0
data = su.get_basic_data(
stock_code, date,
max(time_period1, time_period2, time_period3) * 5)
if data.shape[0] > time_period1:
data['LOW_N'] = data['LOW'].rolling(window=time_period1).min()
data['LOW_N'].fillna(value=data['LOW'].expanding().min(),
inplace=True)
data['HIGH_N'] = data['HIGH'].rolling(window=time_period1).max()
data['HIGH_N'].fillna(value=data['HIGH'].expanding().max(),
inplace=True)
data['RSV'] = (data['CLOSE'] - data['LOW_N']) / (
data['HIGH_N'] - data['LOW_N']) * 100
data.sort_index(ascending=False, inplace=True)
data['KDJ_K'] = data['RSV'].ewm(com=(time_period2 - 1)).mean()
data['KDJ_D'] = data['KDJ_K'].ewm(com=(time_period2 - 1)).mean()
data['KDJ_J'] = 3 * data['KDJ_K'] - 2 * data['KDJ_D']
kdj_k = round(data['KDJ_K'].as_matrix()[-1], 3)
kdj_d = round(data['KDJ_D'].as_matrix()[-1], 3)
kdj_j = round(data['KDJ_J'].as_matrix()[-1], 3)
if np.isnan(kdj_d) or np.isinf(kdj_d) or np.isneginf(kdj_d):
kdj_d = 0.0
if np.isnan(kdj_j) or np.isinf(kdj_j) or np.isneginf(kdj_j):
kdj_j = 0.0
if np.isnan(kdj_k) or np.isinf(kdj_k) or np.isneginf(kdj_k):
kdj_k = 0.0
self.save_tech_data(stock_code, date, {
'KDJ_K': kdj_k,
'KDJ_D': kdj_d,
'KDJ_J': kdj_j
})
return kdj_k, kdj_d, kdj_j
def calculate(self, stock_code, date, time_period1=14, time_period2=6):
wr_14 = 0.0
wr_6 = 0.0
data = su.get_basic_data(stock_code, date,
time_period1 + 1).sort_index(ascending=False)
if data.shape[0] >= time_period1 + 1:
wr_14 = round(
ta.WILLR(data['HIGH'].as_matrix(), data['LOW'].as_matrix(),
data['CLOSE'].as_matrix(), time_period1)[-1] * -1, 3)
wr_6 = round(
ta.WILLR(data['HIGH'].as_matrix(), data['LOW'].as_matrix(),
data['CLOSE'].as_matrix(), time_period2)[-1] * -1, 3)
if np.isnan(wr_14) or np.isinf(wr_14) or np.isneginf(wr_14):
wr_14 = 0.0
if np.isnan(wr_6) or np.isinf(wr_6) or np.isneginf(wr_6):
wr_6 = 0.0
self.save_tech_data(stock_code, date, {'WR_14': wr_14, 'WR_6': wr_6})
return wr_14, wr_6
def calculate(self, stock_code, date, time_period=26):
ar = 0.0
br = 0.0
data = su.get_basic_data(stock_code, date,
time_period + 1).sort_index(ascending=False)
if data.shape[0] >= time_period + 1:
ar = round((data['HIGH'][1:] - data['OPEN'][1:]).sum() /
(data['OPEN'][1:] - data['LOW'][1:]).sum() * 100, 3)
data['P_CLOSE'] = data['CLOSE'].shift(1)
data['BR_U'] = data['HIGH'][1:] - data['P_CLOSE'][1:]
data['BR_D'] = data['P_CLOSE'][1:] - data['LOW'][1:]
br = round(
data[data['BR_U'] > 0]['BR_U'].sum() /
data[data['BR_D'] > 0]['BR_D'].sum() * 100, 3)
if np.isnan(ar) or np.isinf(ar) or np.isneginf(ar):
ar = 0.0
if np.isnan(br) or np.isinf(br) or np.isneginf(br):
br = 0.0
self.save_tech_data(stock_code, date, {'BRAR_BR': br, 'BRAR_AR': ar})
return br, ar
def calculate(self, stock_code, date, time_period1=12, time_period2=6):
max_time_period = max(time_period1, time_period2)
mtm = 0.0
mamtm = 0.0
data = su.get_basic_data(stock_code, date, max_time_period *
2).sort_index(ascending=False)
if data.shape[0] >= max_time_period * 2:
data['N_CLOSE'] = data['CLOSE'].shift(time_period1)
data['MTM'] = data['CLOSE'] - data['N_CLOSE']
mtm = round(data['MTM'].as_matrix()[-1], 3)
mamtm = round(
data['MTM'].as_matrix()[-6:].sum() / float(time_period2), 3)
if np.isnan(mtm) or np.isinf(mtm) or np.isneginf(mtm):
mtm = 0.0
if np.isnan(mamtm) or np.isinf(mamtm) or np.isneginf(mamtm):
mamtm = 0.0
self.save_tech_data(stock_code, date, {'MTM': mtm, 'MAMTM': mamtm})
return mtm, mamtm
def calculate(self, stock_code, date, time_period=24):
vr = 0.0
data = su.get_basic_data(stock_code, date, time_period + 1).sort_index(ascending=False)
if data.shape[0] >= time_period:
p_volume = 0.0
n_volume = 0.0
data['P_CLOSE'] = data['CLOSE'].shift(1)
data['DIFF'] = data['CLOSE']-data['P_CLOSE']
data['DIFF'].fillna(0, inplace=True)
for each in data[1:].itertuples():
if each.DIFF >= 0:
p_volume += each.VOLUME
else:
n_volume += each.VOLUME
vr = round(p_volume / n_volume * 100, 3)
if np.isnan(vr) or np.isinf(vr) or np.isneginf(vr):
vr = 0.0
self.save_tech_data(stock_code, date, {'VR': vr})
return vr
se.export()
sc = StockCleaner(stock_code, end_date)
sc.clean()
else:
logger.info("Only the first day of the year will do export and clean")
else:
logger.warn("No data found for stock code: " + stock_code)
if __name__ == '__main__':
try:
now = datetime.now()
today = du.convertDateToString(now, '%Y-%m-%d')
logger.debug("Today is: " + today)
# today = '2017-09-29'
if su.is_working_day(today):
sbd = StockBasicData()
stock_list = sbd.get_stock_list(today)
pool = mp.Pool(processes=1)
for stock in stock_list:
stock_code = stock[0]
last_updated_date = du.convertDateToString(stock[1], '%Y-%m-%d')
pool.apply_async(create_process, (stock_code, last_updated_date, today))
pool.close()
pool.join()
logger.info("All stocks information is downloaded")
else:
logging.info("Today is holiday, program will terminate...")
except Exception, e:
import os
from util import StockUtil, MysqlUtil
import shutil
import sys
import time
def move_directory(source, destination):
for file_name in os.listdir(source):
if file_name.endswith(".csv"):
shutil.move(os.path.join(source, file_name), destination)
if __name__ == "__main__":
stock_util = StockUtil()
mysql_util = MysqlUtil(stock_util.conn, stock_util.cursor)
stock_util.test = 0
# stock_util.stagnate = 1 if len(sys.argv) >= 2 and sys.argv[1] == '--stagnate' else 0
stock_util.stagnate = 0
stock_util.get_new_stocks = 0
if stock_util.get_new_stocks:
if os.path.exists('GICS/backup'):
shutil.rmtree('GICS/backup')
os.makedirs('GICS/backup')
move_directory('GICS', 'GICS/backup')
stock_util.renew_categories_index_volume()
shutil.rmtree('GICS/finished')
sys.exit(0)
