def get_SMA(self, p):
s = "SMA_{}".format(str(p))
if s not in self.ti_df.columns:
self.ti_df[s] = talib.SMA(self.history_df['Close'], p)
return self.ti_df[s]
def SMAWMA(df,count, acmroi, winnum, winfact,revenue_sum,avg_return,total_return_ratio):
#pd.options.mode.chained_assignment = None #
trade_record = pd.DataFrame(data=None, index=None,columns = ["stockid","bought_date","bought_time","bought_price","sold_date","sold_time","sold_price","QTY","cash_account","LS_type"])
highest= np.array(df['highest'], dtype=float)
close = np.array(df['close'], dtype=float)
vol =df['Quantity'].as_matrix().astype("float64")
df['volMA'] = talib.SMA(vol, timeperiod=20)
SMA = talib.SMA(highest,30) #close 代進SMA方法做計算
WMA = talib.WMA(close,5) #close 代進WMA方法做計算
df['SMA'] = SMA
df['WMA'] = WMA
# print(df)
#設定初始值
df["stockid"]= np.nan
df["bought_date"]= np.nan
df["bought_time"]= np.nan
df["bought_price"]= np.nan
df["sold_date"]= np.nan
df["sold_time"]= np.nan
df["sold_price"]= np.nan
df["QTY"]=1000
df["cash_account"]= np.nan
df["LS_type"]=0
df['XBuy'] = np.nan
df['YBuy'] = np.nan
df['XSell'] = np.nan
df['YSell'] = np.nan
row = len(df)
flag1 = False
flag = False
change = 0
buyprice=[]
sellprice=[]
win = 0
loss = 0
roi =0
revenue=0
total_return=0
revenue_sum=0
for i in range(row):
change =df['WMA'].iloc[i]/ df['SMA'].iloc[i]
if (flag1 == False) & (df['WMA'].iloc[i] <df['SMA'].iloc[i]) & (change <= 0.948)& (float(df['Quantity'].iloc[i]) >=(float(df['volMA'].iloc[i]))):
df['XSell'].iloc[i] = df['MATCH_TIME'].iloc[i]
df['YSell'].iloc[i] = df['close'].iloc[i]
sellprice= df['close'].iloc[i]
df["stockid"].iloc[i]=df['STOCK_SYMBOL'].iloc[i]
df["sold_date"].iloc[i]=df['date'].iloc[i]
df["sold_time"].iloc[i]=df['MATCH_TIME'].iloc[i]
df["sold_price"].iloc[i]=df['close'].iloc[i]
#df["cash_account"].iloc[i]= round(float(5000000+float(buyprice*1.001425-sellprice*(1-0.001425*0.6-0.001))*1000),0)
df["LS_type"].iloc[i]=2
print("S2賣出價"+str(sellprice))
flag1 = True
elif (flag == False) & (df['WMA'].iloc[i] <= df['SMA'].iloc[i]) &(change <= 0.995) & (float(df['Quantity'].iloc[i]) >=(float(df['volMA'].iloc[i]))):
df['XBuy'].iloc[i] = df['MATCH_TIME'].iloc[i]
df['YBuy'].iloc[i] = df['close'].iloc[i]
buyprice =df['close'].iloc[i]
df["stockid"].iloc[i]=df['STOCK_SYMBOL'].iloc[i]
df["bought_date"].iloc[i]=df['date'].iloc[i]
df["bought_time"].iloc[i]=df['MATCH_TIME'].iloc[i]
df["bought_price"].iloc[i]=df['close'].iloc[i]
#df["cash_account"].iloc[i]= round(float(5000000-float(buyprice)*1000*(1-0.001425*0.6-0.001)),0)
df["LS_type"].iloc[i]=1
flag = True
print("S1買進價"+str(buyprice))
elif(flag1 == True) & (df['WMA'].iloc[i] >= df['SMA'].iloc[i])& (change >= 1.009) & (float(df['Quantity'].iloc[i]) >=(float(df['volMA'].iloc[i]))):
try:
df['XBuy'].iloc[i] = df['MATCH_TIME'].iloc[i+1]
df['YBuy'].iloc[i] = df['close'].iloc[i+1]
buyprice =df['close'].iloc[i+1]
df["stockid"].iloc[i]=df['STOCK_SYMBOL'].iloc[i+1]
df["bought_date"].iloc[i]=df['date'].iloc[i+1]
df["bought_time"].iloc[i]=df['MATCH_TIME'].iloc[i+1]
df["bought_price"].iloc[i]=df['close'].iloc[i+1]
#df["cash_account"].iloc[i]= round(float(5000000-float(buyprice)*1000*(1-0.001425*0.6-0.001)),0)
df["LS_type"].iloc[i]=2
flag1 = False
print("S2回補價"+str(buyprice))
count += 1
revenue=((sellprice*0.995575)- (buyprice*1.001425))*1000
revenue_sum +=revenue
premium=revenue/((buyprice*1.001425)*1000)
total_return +=premium
avg_return=(total_return/count)
total_return_ratio=round((revenue_sum/5000000),3)*100
[roi, winnum] =roical(buyprice, sellprice, winnum)
acmroi += roi
[loss, win] = winfactor(buyprice, sellprice, loss, win)
except (ValueError):
continue
elif (flag == True) & (df['WMA'].iloc[i] >df['SMA'].iloc[i]) & (change >= 1.021) & (float(df['Quantity'].iloc[i]) >=(float(df['volMA'].iloc[i]))):
try:
df['XSell'].iloc[i] = df['MATCH_TIME'].iloc[i+2]
df['YSell'].iloc[i] = df['close'].iloc[i+2]
sellprice= df['close'].iloc[i+2]
count += 1
df["stockid"].iloc[i]=df['STOCK_SYMBOL'].iloc[i+2]
df["sold_date"].iloc[i]=df['date'].iloc[i+2]
df["sold_time"].iloc[i]=df['MATCH_TIME'].iloc[i+2]
df["sold_price"].iloc[i]=df['close'].iloc[i+2]
#df["cash_account"].iloc[i]= round(float(5000000+float(buyprice*1.001425-sellprice*(1-0.001425*0.6-0.001))*1000),0)
df["LS_type"].iloc[i]=1
flag = False
print("S1賣出價"+str(sellprice))
[roi, winnum] =roical(buyprice, sellprice, winnum)
acmroi += roi
[loss, win] = winfactor(buyprice, sellprice, loss, win)
revenue=((sellprice*0.995575)- (buyprice*1.001425))*1000
revenue_sum +=revenue
premium=revenue/((buyprice*1.001425)*1000)
total_return +=premium
avg_return=(total_return/count)
total_return_ratio=round((revenue_sum/5000000),3)*100
except (ValueError):
continue
if (flag == True & i==(row-1)):
df['XSell'].iloc[i] = df['MATCH_TIME'].iloc[i]
df['YSell'].iloc[i] = df['close'].iloc[i]
sellprice=df['close'].iloc[i]
df["stockid"].iloc[i]=df['STOCK_SYMBOL'].iloc[i]
df["sold_date"].iloc[i]=df['date'].iloc[i]
df["sold_time"].iloc[i]=df['MATCH_TIME'].iloc[i]
df["sold_price"].iloc[i]=df['close'].iloc[i]
df["QTY"].iloc[i]=1000
#df["cash_account"].iloc[i]= (5000000+(buyprice*1.001425*0.6-sellprice* (1-0.001425*0.6-0.001))*1000)
df["LS_type"].iloc[i]=1
count += 1
[roi, winnum] = roical(buyprice, sellprice, winnum)
acmroi += roi
[loss, win] = winfactor(buyprice, sellprice, loss, win)
print("S最後被迫賣價"+str(sellprice))
#revenue=((sellprice*0.995575)- (buyprice*1.001425))*1000
#revenue_sum +=revenue
#premium=revenue/((buyprice*1.001425)*1000)
#total_return +=premium
#avg_return=(total_return/count)
#total_return_ratio=round((revenue_sum/5000000),3)*100
elif (flag1 == True & i==(row-1)):
df['Xbuy'][i] = df['MATCH_TIME'].iloc[i]
df['Ybuy'][i] = df['close'].iloc[i]
buyprice=df['close'].iloc[i]
df["stockid"].iloc[i]=df['STOCK_SYMBOL'].iloc[i]
df["bought_date"].iloc[i]=df['date'].iloc[i]
df["bought_time"].iloc[i]=df['MATCH_TIME'].iloc[i]
df["bought_price"].iloc[i]=df['close'].iloc[i]
df["QTY"].iloc[i]=1000
#df["cash_account"].iloc[i]= (5000000+(buyprice*1.001425*0.6-sellprice* (1-0.001425*0.6-0.001))*1000)
df["LS_type"].iloc[i]=2
count += 1
[roi, winnum] = roical(buyprice, sellprice, winnum)
acmroi += roi
print("S2最後被迫回補價"+str(buyprice))
#revenue=((sellprice*0.995575)- (buyprice*1.001425))*1000
#revenue_sum +=round(revenue,2)
#premium=revenue/((buyprice*1.001425)*1000)
#total_return +=premium
#avg_return=(total_return/count)
#total_return_ratio=round((revenue_sum/5000000),3)*100
elif (flag1 == False or flag == False& i==(row-10)):
break
if (loss == 0):
loss = 1
winvar = win / loss
print(' win = ', win)
print('loss = ', loss)
print('winvar = ', winvar)
print('revenue_sum=',revenue_sum)
if (count == 0):
count = 0.01
#str1 = 'SMAWMA策略: ' +'交易次數 = '+ str(count) + ' 次; ' + '累計報酬率 = ' + str(round(acmroi*100, 2)) + '%;' + '勝率 = ' + str(round((winnum/count)*100,2)) + '%' + '; 獲利因子 = ' + str(round(winvar, 2) )+ '%' + '; 總收益 = ' + str(round(revenue_sum, 2)+ '; 平均報酬率 = ' + str(round(avg_return*100, 2))+ '%'+ '; 夏普指數 = ' + str(round(sharpe_ratio*100, 2))+ '%'+ '; MDD = ' + str(round(MDD, 2))+ '; 總資產報酬率 = ' + str(round(total_return_ratio*100, 2))+ '%')
#print(str1)
trade_record.index.names = ['trade_no']
trade_record["stockid"]= df["stockid"]
trade_record["bought_date"]=df["bought_date"]
trade_record["bought_time"]=df["bought_time"]
trade_record["bought_price"]=df["bought_price"]
trade_record["sold_date"]=df["sold_date"]
trade_record["sold_time"]= df["sold_time"]
trade_record["sold_price"]=df["sold_price"]
#trade_record["cash_account"]=df["cash_account"]
trade_record["LS_type"]=df["LS_type"]
trade_record["offset_date"] = trade_record[["bought_date","sold_date"]].max(axis=1) #抓出平倉日
#trade_record =trade_record.sort_values(by=["offset_date",'trade_no'])
#trade_record = trade_record.drop(["bought_date"],axis=1) #將多餘的紀錄刪除
#trade_record =trade_record.reset_index(drop=True) #重設index
#trade_record["Algorithm"] =2
#trade_record["QTY"]=1000
trade_record.to_csv(r"./SAM_trade_record.csv", sep = ",")
#print(str1)
#print(trade_record)
#print(df)
#print(trade_record)
return (count, acmroi, winnum, winvar,revenue_sum,avg_return,total_return_ratio )
def sma(px, lookback):
sig = ta.SMA(px, timeperiod=lookback)
return sig[-1]
def feature_engineering(train_data):
#SMA
train_data.insert(0, 'sma20',
talib.SMA(train_data['close'].values, timeperiod=20),
True)
#CCI
train_data.insert(
0, 'cci14',
talib.CCI(train_data['high'].values,
train_data['low'].values,
train_data['close'].values,
timeperiod=14), True)
#RSI
train_data.insert(0, 'rsi14',
talib.RSI(train_data['close'].values, timeperiod=14),
True)
#ADX
train_data.insert(
0, 'adx14',
talib.ADX(train_data['high'].values,
train_data['low'].values,
train_data['close'].values,
timeperiod=14), True)
#ATR
train_data.insert(
0, 'atr14',
talib.ATR(train_data['high'].values,
train_data['low'].values,
train_data['close'].values,
timeperiod=14), True)
#Bands
bb20Upperband, bb20Middleband, bb20Lowerband = talib.BBANDS(
train_data['close'].values,
timeperiod=20,
nbdevup=2,
nbdevdn=2,
matype=0)
train_data.insert(0, 'bb20Upperband', bb20Upperband, True)
train_data.insert(0, 'bb20Middleband', bb20Middleband, True)
train_data.insert(0, 'bb20Lowerband', bb20Lowerband, True)
bb50Upperband, bb50Middleband, bb50Lowerband = talib.BBANDS(
train_data['close'].values,
timeperiod=50,
nbdevup=2,
nbdevdn=2,
matype=0)
train_data.insert(0, 'bb50Upperband', bb50Upperband, True)
train_data.insert(0, 'bb50Middleband', bb50Middleband, True)
train_data.insert(0, 'bb50Lowerband', bb50Lowerband, True)
#MACD
macd1226, macdSignal1226, macdHist1226 = talib.MACD(
train_data['close'].values,
fastperiod=13,
slowperiod=26,
signalperiod=9)
train_data.insert(0, 'macd1226', macd1226, True)
train_data.insert(0, 'macdSignal1226', macdSignal1226, True)
train_data.insert(0, 'macdHist1226', macdHist1226, True)
#Stochastic
stochasticSlowK335, stochasticSlowD335 = talib.STOCH(
train_data['high'].values,
train_data['low'].values,
train_data['close'].values,
fastk_period=5,
slowk_period=3,
slowk_matype=0,
slowd_period=3,
slowd_matype=0)
train_data.insert(0, 'stochasticSlowK335', stochasticSlowK335, True)
train_data.insert(0, 'stochasticSlowD335', stochasticSlowD335, True)
return train_data
## get the data for all bitcoin pairs and tidy it
bin_products = client.get_products()
bin_data = bin_products['data']
btc_pairs = [pair['s'] for pair in bin_data if pair['s'].endswith("BTC")]
candle_data = [
client.get_historical_klines(pair, Client.KLINE_INTERVAL_1DAY,
str(active_window)) for pair in btc_pairs
]
close_data = np.zeros(
(len(btc_pairs), ndays)) #get close data and discard the rest
for token in range(len(candle_data)):
for day in range(len(candle_data[token])):
close_data[token][day] = candle_data[token][day][4]
## calculate short and long SMAs
SMA_long_data = np.zeros((len(btc_pairs), ndays))
SMA_short_data = np.zeros((len(btc_pairs), ndays))
for token in range(len(close_data)):
SMA_long_data[token] = talib.SMA(close_data[token], SMA_long)
SMA_short_data[token] = talib.SMA(close_data[token], SMA_short)
## run the analysis to find the crossings and add relevant coins to stonk list
stonks = []
for comp in range(len(btc_pairs)):
if SMA_short_data[comp][cross_window] < SMA_long_data[comp][cross_window]:
if SMA_short_data[comp][-1] > SMA_long_data[comp][-1]:
stonks.append(btc_pairs[comp])
def MA(timeperiod):
MA = tb.SMA(sp.Close, timeperiod)
return MA
def sma(self, n, array=False):
"""简单均线"""
result = talib.SMA(self.close, n)
if array:
return result
return result[-1]
def compute_indicators(self):
df = deepcopy(self.df)
open = df.open
close = df.close
high = df.high
low = df.low
volume = df.volume
adjclose = close
nan_offset = 50
df['beta'] = talib.BETA(high, low, timeperiod=5)
df['rsi_14'] = talib.RSI(close, timeperiod=14)
df['rsi_14_thresh'] = df.rsi_14.apply(lambda c: 1 if c > 70 else (-1 if c < 30 else 0))
df['rsi_21'] = talib.RSI(close, timeperiod=21)
df['rsi_21_thresh'] = df.rsi_21.apply(lambda c: 1 if c > 70 else (-1 if c < 30 else 0))
df['rsi_60'] = talib.RSI(close, timeperiod=60)
df['rsi_60_thresh'] = df.rsi_60.apply(lambda c: 1 if c > 70 else (-1 if c < 30 else 0))
df['macd_12_26'], df['macdsignal_12_26'], df['macdhist_12_26'] = talib.MACD(close, fastperiod=12, slowperiod=26,
signalperiod=9)
df['macd_5_12'], df['macdsignal_5_12'], df['macdhist_5_12'] = talib.MACD(close, fastperiod=3, slowperiod=15,
signalperiod=5)
df['mom'] = talib.MOM(close, timeperiod=10)
df['roc_20'] = talib.ROC(close, timeperiod=21)
df['roc_125'] = talib.ROC(close, timeperiod=125)
df['roc_90'] = talib.ROC(close, timeperiod=90)
high = deepcopy(df.high)
low = deepcopy(df.low)
df['psar_005'] = self.get_sar(high, low, 0.005, 0.2)
df['psar_02'] = self.get_sar(high, low, 0.02, 0.2)
df['psar_1'] = self.get_sar(high, low, 0.1, 0.2)
try:
df['psar_005_dist'] = df.psar_005 - df.adjclose
df['psar_02_dist'] = df.psar_02 - df.adjclose
df['psar_1_dist'] = df.psar_1 - df.adjclose
except Exception as e:
print(e)
try:
df['psar_005_ind'] = (df.psar_005 < self.df.adjclose).astype(int)
df['psar_02_ind'] = (df.psar_02 < self.df.adjclose).astype(int)
df['psar_1_ind'] = (df.psar_1 < self.df.adjclose).astype(int)
except Exception as e:
print(e)
df['sma5'] = talib.SMA(close, timeperiod=5)
df['sma10'] = talib.SMA(close, timeperiod=10)
df['sma20'] = talib.SMA(close, timeperiod=20)
df['sma50'] = talib.SMA(close, timeperiod=50)
df['sma200'] = talib.SMA(close, timeperiod=200)
df['midprice_5'] = talib.MIDPRICE(high, low, timeperiod=5)
df['midprice_14'] = talib.MIDPRICE(high, low, timeperiod=14)
df['midprice_21'] = talib.MIDPRICE(high, low, timeperiod=21)
df['upperband'], df['middleband'], df['lowerband'] = talib.BBANDS(close, timeperiod=5, nbdevup=2., nbdevdn=2.,
matype=0)
df['instantaneous_trend'] = talib.HT_TRENDLINE(close)
df['adx_7'] = talib.ADX(high, low, close, timeperiod=7)
df['adx_7_pctchg'] = df.adx_7.pct_change()
df['adx_14'] = talib.ADX(high, low, close, timeperiod=14)
df['adx_14_pctchg'] = df.adx_14.pct_change()
df['adx_21'] = talib.ADX(high, low, close, timeperiod=21)
df['adx_21_pctchg'] = df.adx_21.pct_change()
df['adx_60'] = talib.ADX(high, low, close, timeperiod=61)
df['adx_60_pctchg'] = df.adx_60.pct_change()
df['cci'] = talib.CCI(high, low, close, timeperiod=14)
df['cci_chg'] = df.cci.pct_change()
df['cci_thresh'] = df.cci.apply(lambda c: 1 if c > 100 else (-1 if c < 20 else 0))
df['direction_movement_idx'] = talib.DX(high, low, close, timeperiod=14)
df['money_flow'] = talib.MFI(high, low, close, volume, timeperiod=14)
df['aroon_down'], df['aroon_up'] = talib.AROON(high, low)
df['ppo'] = talib.PPO(close, fastperiod=12, slowperiod=26, matype=0)
df['trix'] = talib.TRIX(close, timeperiod=30)
df['stoch_k'], df['stoch_d'] = talib.STOCH(high, low, close, fastk_period=5, slowk_period=3, slowk_matype=0,
slowd_period=3, slowd_matype=0)
df['stoch_rsi_k'], df['stoch_rsi_d'] = talib.STOCHRSI(close, timeperiod=14, fastk_period=5, fastd_period=3,
fastd_matype=0)
df['willR'] = talib.WILLR(high, low, close, timeperiod=14)
df['natr'] = talib.NATR(high, low, close, timeperiod=14)
df['trange'] = talib.TRANGE(high, low, close)
df['obv'] = talib.OBV(close, volume)
df['adosc'] = talib.ADOSC(high, low, close, volume, fastperiod=3, slowperiod=10)
df['ad'] = talib.AD(high, low, close, volume)
try:
df['log_return'] = df.ta.log_return()
df['percent_return'] = df.ta.percent_return()
except Exception as e:
print(e)
df['zscore'] = df.ta.zscore()
df['quantile'] = df.ta.quantile()
df['ht_dom_per'] = talib.HT_DCPERIOD(close)
df['ht_dom_cycle'] = talib.HT_DCPHASE(close)
df['ht_trendmode'] = talib.HT_TRENDMODE(close)
df['pvt'] = df.ta.pvt()
df['increasing'] = df.ta.increasing()
df['decreasing'] = df.ta.decreasing()
df['cross_sma5'] = df.ta.cross('close', 'sma5')
df['cross_sma10'] = df.ta.cross('close', 'sma10')
df['cross_sma20'] = df.ta.cross('close', 'sma20')
df['cross_sma50'] = df.ta.cross('close', 'sma50')
df['cross_sma200'] = df.ta.cross('close', 'sma200')
df['sma20_above'] = (df.sma20 > df.sma200).astype(int)
df['sma50_above'] = (df.sma50 > df.sma200).astype(int)
df['cross_psar_rsi14'] = df.ta.cross('psar_02', 'rsi_14')
df['cross_psar_005_close'] = df.ta.cross('psar_005', 'adjclose')
df['cross_psar_02_close'] = df.ta.cross('psar_02', 'adjclose')
df['cross_psar_1_close'] = df.ta.cross('psar_1', 'adjclose')
df['cross_adx14_psar_02'] = df.ta.cross('psar_02', 'adx_14')
df['cross_adx7_psar_02'] = df.ta.cross('psar_005', 'adx_7')
df['cross_adx14_psar_02'] = df.ta.cross('psar_005', 'adx_14')
df['cross_stoch'] = df.ta.cross('stoch_k', 'stoch_d')
df['cross_macd_12_26'] = df.ta.cross('macd_12_26', 'macdsignal_12_26')
df['cross_macd_5_12'] = df.ta.cross('macd_5_12', 'macdsignal_5_12')
df['roc_20_125'] = df.ta.cross('roc_20', 'roc_125')
df['roc_20_125_dist'] = df.roc_20 - df.roc_125
#df = df.iloc[nan_offset:, :].dropna(axis=1).copy()
self.df = df
self.indicator_matrix = df.iloc[:, 6:]
df = DataReader(stock, 'yahoo', start, end)
# Get Index Data
index = 'SPY'
spy = DataReader(index, 'yahoo', start, end)
spy['RSI'] = talib.RSI(spy['Adj Close'], timeperiod=5)
# Technical Indicators
df['upper_band'], df['middle_band'], df['lower_band'] = talib.BBANDS(
df['Adj Close'], timeperiod=7)
df['macd'], df['macdsignal'], df['macdhist'] = talib.MACD(
df['Adj Close'], fastperiod=12, slowperiod=26, signalperiod=9)
df['RSI'] = talib.RSI(df['Adj Close'], timeperiod=5)
df['Momentum'] = talib.MOM(df['Adj Close'], timeperiod=5)
df['Z-Score'] = zscore(df['Adj Close'])
df['SMA'] = talib.SMA(df['Adj Close'], timeperiod=7)
df['EMA'] = talib.EMA(df['Adj Close'], timeperiod=7)
df['OBV'] = talib.OBV(df['Adj Close'], df['Volume']) / 10**6
df['OBV'] = df['OBV'].diff()
df['CCI'] = ta.trend.cci(df['High'],
df['Low'],
df['Adj Close'],
n=7,
c=0.015)
# Set signal position columns
df['bbPos'] = None
df['macdPos'] = None
df['rsiPos'] = None
df['spyPos'] = None
df['zPos'] = None
def add_sma(self, df):
output = talib.SMA(df['Close'].values, timeperiod=25)
df['SMA'] = output
return df
def create_features(twii):
sma = talib.SMA(twii, timeperiod=120)
wma = talib.WMA(twii, timeperiod=120)
mom = talib.MOM(twii, timeperiod=120)
k, d = talib.STOCH(twii,
twii,
twii,
fastk_period=120,
slowk_period=60,
slowd_period=60)
k2, d2 = talib.STOCH(twii,
twii,
twii,
fastk_period=240,
slowk_period=120,
slowd_period=120)
k3, d3 = talib.STOCH(twii,
twii,
twii,
fastk_period=360,
slowk_period=180,
slowd_period=180)
k4, d4 = talib.STOCH(twii,
twii,
twii,
fastk_period=480,
slowk_period=240,
slowd_period=240)
k5, d5 = talib.STOCH(twii,
twii,
twii,
fastk_period=640,
slowk_period=320,
slowd_period=320)
k6, d6 = talib.STOCH(twii,
twii,
twii,
fastk_period=720,
slowk_period=360,
slowd_period=360)
k7, d7 = talib.STOCH(twii,
twii,
twii,
fastk_period=840,
slowk_period=420,
slowd_period=420)
k8, d8 = talib.STOCH(twii,
twii,
twii,
fastk_period=960,
slowk_period=480,
slowd_period=480)
rsi = talib.RSI(twii, timeperiod=120)
rsi2 = talib.RSI(twii, timeperiod=240)
rsi3 = talib.RSI(twii, timeperiod=480)
rsi4 = talib.RSI(twii, timeperiod=640)
rsi5 = talib.RSI(twii, timeperiod=720)
rsi6 = talib.RSI(twii, timeperiod=840)
macd1, macd2, macd3 = talib.MACD(twii,
fastperiod=120,
slowperiod=60,
signalperiod=60)
willr = talib.WILLR(twii, twii, twii, timeperiod=120)
cci = talib.CCI(twii, twii, twii, timeperiod=120)
return pd.DataFrame({
'RSIb': rsi / 50,
'RSIb2': rsi2 / 50,
'RSIb3': rsi3 / 50,
'RSIb4': rsi4 / 50,
'RSIb5': rsi5 / 50,
'RSIb6': rsi6 / 50,
'MOMb': mom - 0,
'KDb': k - d,
'KDb2': k2 - d2,
'KDb3': k3 - d3,
'KDb4': k4 - d4,
'KDb5': k5 - d5,
'KDb6': k6 - d6,
'KDb7': k7 - d7,
'KDb8': k8 - d8,
'a5': (twii.rolling(5).mean() / twii),
'a10': (twii.rolling(10).mean() / twii),
'a20': (twii.rolling(20).mean() / twii),
'a40': (twii.rolling(40).mean() / twii),
'a80': (twii.rolling(80).mean() / twii),
'a160': (twii.rolling(160).mean() / twii),
'a320': (twii.rolling(320).mean() / twii),
'a640': (twii.rolling(640).mean() / twii),
'a720': (twii.rolling(720).mean() / twii),
'a840': (twii.rolling(840).mean() / twii),
'a960': (twii.rolling(960).mean() / twii),
'a1024': (twii.rolling(1024).mean() / twii),
'b1': twii / twii.shift(50),
'b2': twii / twii.shift(100),
'b3': twii / twii.shift(150),
'b4': twii / twii.shift(200),
'b5': twii / twii.shift(250),
'b6': twii / twii.shift(300),
'b7': twii / twii.shift(350),
'LINEARREG_SLOPE0': talib.LINEARREG_SLOPE(twii, 60),
'LINEARREG_SLOPE1': talib.LINEARREG_SLOPE(twii, 120),
'ADXR0': talib.ADXR(twii, twii, twii, 60),
'ADXR1': talib.ADXR(twii, twii, twii, 120),
'ADXR2': talib.ADXR(twii, twii, twii, 240),
'ADXR3': talib.ADXR(twii, twii, twii, 360),
'ADXR4': talib.ADXR(twii, twii, twii, 480),
'ADXR5': talib.ADXR(twii, twii, twii, 640),
})
# input
symbol = str(input('Enter a ticker: '))
symbol = symbol.strip()
symbol = symbol.upper()
num_of_years = 1
start = dt.date.today() - dt.timedelta(days=int(365.25 * num_of_years))
end = dt.date.today()
# Read data
data = yf.download(symbol, start, end)
# ## SMA and EMA
#Simple Moving Average
data['SMA'] = talib.SMA(data['Adj Close'], timeperiod=20)
# Exponential Moving Average
data['EMA'] = talib.EMA(data['Adj Close'], timeperiod=20)
# Bollinger Bands
data['upper_band'], data['middle_band'], data['lower_band'] = talib.BBANDS(
data['Adj Close'], timeperiod=20)
# Plot
rcParams['figure.figsize'] = 15, 10
fig, (ax1, ax2) = plt.subplots(2)
ax1.plot(data[['Adj Close', 'SMA', 'EMA']])
ax1.legend(('Adj Close', 'SMA', 'EMA'))
ax1.set_ylabel('Close Price')
ax1.set_title(f'SMA vs. EMA for {symbol.upper()}')
def get_SMA_price(self, p):
s = "SMA_{}_price".format(str(p))
if s not in self.ti_df.columns:
self.ti_df[s] = talib.SMA(self.history_df['Low'], p)
return self.ti_df[s]
def get_SMA_turnover(self, p):
s = "SMA_{}_turnover".format(str(p))
if s not in self.ti_df.columns:
self.ti_df[s] = talib.SMA(self.history_df['Turnover'], p)
return self.ti_df[s]
def get_ma_ouptput(codes):
"""
get moving average day, week, month for output
:param codes:
:return:
codes = [("code", "name")]
"""
#codes = [('000636', '风华高科') , ('600066', '宇通客车'), ('000786','北新建材'), ('600817','ST宏盛'),
# ('000830', '鲁西化工'), ('510050', '50ETF')]
#codes2 = ['QQQ','SPY','IWM','DIA', 'GLD']
today = datetime.datetime.now()
start_date = today - datetime.timedelta(days = 200)
start_date = start_date.strftime("%Y-%m-%d")
end_date = today.strftime("%Y-%m-%d")
output_m =[]
output_w =[]
output_d =[]
coll_stock_day = DATABASE.stock_day
stock_list = DATABASE.stock_list
code_list = coll_stock_day.distinct('code')
index_day = DATABASE.index_day
index_list = index_day.distinct('code')
for (code, symbol) in codes:
''''
if code == '510050':
df = QA.QA_fetch_index_day_adv([code], start_date, end_date)
else:
df = QA.QA_fetch_stock_day_adv([code], start_date, end_date)
#df = ts.get_hist_data(code, ktype='D', start=start_date, end=end_date)
if code != '510050':
df = df.to_qfq()()
else:
df = df()
'''
if code in code_list:
df = QA.QA_fetch_stock_day_adv([code], start_date, end_date)
df = df.to_qfq()()
else:
df = QA.QA_fetch_index_day_adv([code], start_date, end_date)()
df_w = QA.QAData.data_resample.QA_data_day_resample(df, 'W')
df_m = QA.QAData.data_resample.QA_data_day_resample(df, 'M')
df['ma5'] = talib.SMA(df['close'].values, timeperiod=5)
df_w['ma5'] = talib.SMA(df_w['close'].values, timeperiod=5)
df_m['ma5'] = talib.SMA(df_m['close'].values, timeperiod=5)
df = df.reset_index().tail(1)
df_w = df_w.reset_index().tail(1)
df_m = df_m.reset_index().tail(1)
df['indicator'] = 'No'
df_w['indicator'] = 'No'
df_m['indicator'] = 'No'
df.loc[df.close > df.ma5,'indicator'] = "Yes"
df_w.loc[df_w.close > df_w.ma5,'indicator'] = "Yes"
df_m.loc[df_m.close > df_m.ma5,'indicator'] = "Yes"
df_m['symbol'] = symbol
#df_m['code'] = code
df_m['type'] = '月5MA'
#df['code'] = code
df['type'] = '日5MA'
df['symbol'] = symbol
#df_w['code'] = code
df_w['type'] = '周5MA'
df_w['symbol'] = symbol
output_m.append(df_m)
output_d.append(df)
output_w.append(df_w)
result_m = pd.concat(output_m)
result_d = pd.concat(output_d)
result_w = pd.concat(output_w)
cols = ['symbol','code','date','type','close','ma5','indicator']
result = [result_m[cols], result_d[cols], result_w[cols]]
combined = pd.concat(result)
combined = combined.sort_values(['code', 'type'])
combined['diff'] = round(combined.close - combined.ma5,2)
combined['close'] = round(combined.close, 2)
combined['ma5'] = round(combined.ma5, 2)
combined['date'] = combined.date.map(lambda x: x.strftime("%Y-%m-%d"))
combined['type'] = pd.Categorical(combined['type'], ['日5MA', '周5MA', '月5MA'])
combined = combined.sort_values(['symbol', 'type'])
return combined
def getSMA(self):
output = talib.SMA(self.close, 20)
a = pd.DataFrame(output)
a = a.rename(columns={0: "SMA"})
return a
def get_tech(filepath):
df = pd.read_csv(filepath, index_col='date')
# Simple Moving Average SMA 简单移动平均
df['SMA5'] = talib.MA(df['close'], timeperiod=5)
df['SMA10'] = talib.MA(df['close'], timeperiod=10)
df['SMA20'] = talib.MA(df['close'], timeperiod=20)
# Williams Overbought/Oversold Index WR 威廉指标
df['WR14'] = talib.WILLR(df['high'], df['low'], df['close'], timeperiod=14)
df['WR18'] = talib.WILLR(df['high'], df['low'], df['close'], timeperiod=18)
df['WR22'] = talib.WILLR(df['high'], df['low'], df['close'], timeperiod=22)
# Moving Average Convergence / Divergence MACD 指数平滑移动平均线
DIFF1, DEA1, df['MACD9'] = talib.MACD(np.array(df['close']),
fastperiod=12,
slowperiod=26,
signalperiod=9)
DIFF2, DEA2, df['MACD10'] = talib.MACD(np.array(df['close']),
fastperiod=14,
slowperiod=28,
signalperiod=10)
DIFF3, DEA3, df['MACD11'] = talib.MACD(np.array(df['close']),
fastperiod=16,
slowperiod=30,
signalperiod=11)
df['MACD9'] = df['MACD9'] * 2
df['MACD10'] = df['MACD10'] * 2
df['MACD11'] = df['MACD11'] * 2
# Relative Strength Index RSI 相对强弱指数
df['RSI15'] = talib.RSI(np.array(df['close']), timeperiod=15)
df['RSI20'] = talib.RSI(np.array(df['close']), timeperiod=20)
df['RSI25'] = talib.RSI(np.array(df['close']), timeperiod=25)
df['RSI30'] = talib.RSI(np.array(df['close']), timeperiod=30)
# Stochastic Oscillator Slow STOCH 常用的KDJ指标中的KD指标
df['STOCH'] = talib.STOCH(df['high'],
df['low'],
df['close'],
fastk_period=9,
slowk_period=3,
slowk_matype=0,
slowd_period=3,
slowd_matype=0)[1]
# On Balance Volume OBV 能量潮
df['OBV'] = talib.OBV(np.array(df['close']), df['volume'])
# Simple moving average SMA 简单移动平均
df['SMA15'] = talib.SMA(df['close'], timeperiod=15)
df['SMA20'] = talib.SMA(df['close'], timeperiod=20)
df['SMA25'] = talib.SMA(df['close'], timeperiod=25)
df['SMA30'] = talib.SMA(df['close'], timeperiod=30)
# Money Flow Index MFI MFI指标
df['MFI14'] = talib.MFI(df['high'],
df['low'],
df['close'],
df['volume'],
timeperiod=14)
df['MFI18'] = talib.MFI(df['high'],
df['low'],
df['close'],
df['volume'],
timeperiod=18)
df['MFI22'] = talib.MFI(df['high'],
df['low'],
df['close'],
df['volume'],
timeperiod=22)
# Ultimate Oscillator UO 终极指标
df['UO7'] = talib.ULTOSC(df['high'],
df['low'],
df['close'],
timeperiod1=7,
timeperiod2=14,
timeperiod3=28)
df['UO8'] = talib.ULTOSC(df['high'],
df['low'],
df['close'],
timeperiod1=8,
timeperiod2=16,
timeperiod3=22)
df['UO9'] = talib.ULTOSC(df['high'],
df['low'],
df['close'],
timeperiod1=9,
timeperiod2=18,
timeperiod3=26)
# Rate of change Percentage ROCP 价格变化率
df['ROCP'] = talib.ROCP(df['close'], timeperiod=10)
fig = plt.figure(figsize=(12, 6))
ax = fig.add_subplot(1, 1, 1)
ax.set_xticks(range(0, len(df_1110.index), 10))
ax.set_xticklabels(df_1110.index[::10])
mpf.candlestick2_ochl(ax,
df_1110['Open'],
df_1110['Close'],
df_1110['High'],
df_1110['Low'],
width=0.6,
colorup='r',
colordown='g',
alpha=0.75)
import talib
sma_10 = talib.SMA(np.array(df_1110['Close']), 10)
sma_30 = talib.SMA(np.array(df_1110['Close']), 30)
fig = plt.figure(figsize=(10, 4))
ax = fig.add_axes([0, 0.2, 1, 0.5])
ax2 = fig.add_axes([0, 0, 1, 0.2])
ax.set_xticks(range(0, len(df_1110.index), 10))
ax.set_xticklabels(df_1110.index[::10])
mpf.candlestick2_ochl(ax,
df_1110['Open'],
df_1110['Close'],
df_1110['High'],
df_1110['Low'],
width=0.6,
colorup='r',
from iexfinance import get_historical_data
from datetime import datetime
import pandas as pd
import talib as ta
import matplotlib.pyplot as plt
import seaborn as sns
plt.style.use('bmh')
start = datetime(2018, 1, 1)
end = datetime(2019, 1, 1)
data = get_historical_data('AAPL',
start=start,
end=end,
output_format='pandas')
data.index = pd.to_datetime(data.index)
data.head()
data['SMA'] = ta.SMA(data.close, timeperiod=20)
data['EMA'] = ta.EMA(data.close, timeperiod=20)
data[['close', 'SMA', 'EMA']].plot(figsize=(10, 5))
plt.show()
json.dump(account_info, dump_f)
print('connect ws error!')
break
continue
retryCount=0
kline = (pd.DataFrame.from_dict(kline_1min))[['id', 'close', 'high', 'low', 'open', 'amount']]
id = kline['id'].values
id = (id[-1] / 60)
closed = kline['close'].values
opened = kline['open'].values
highed = kline['high'].values
lowed = kline['low'].values
amounted = kline['amount'].values
ma5 = talib.SMA(closed, timeperiod=5)
ma20 = talib.SMA(closed, timeperiod=20)
ma30 = talib.SMA(closed, timeperiod=30)
ma60 = talib.SMA(closed, timeperiod=60)
rsi = talib.RSI(closed, timeperiod=14)
macd, signal, hist = talib.MACD(closed, fastperiod=12, slowperiod=26, signalperiod=9)
if (account_info['margin_available'] + account_info['margin_frozen'] + (account_info['price'] - highed[-1]) *
account_info['volume']) <= 0:
account_info['margin_available'] = 0
account_info['margin_frozen'] = 0
account_info['cost_price'] = 0
account_info['volume'] =0
account_info['margin_available'] += (account_info['price'] - closed[-1]) * account_info['volume']
account_info['price'] = closed[-1]
def handle_bar(context, bar_dict):
# 开始编写你的主要的算法逻辑
# bar_dict[order_book_id] 可以拿到某个证券的bar信息
# context.portfolio 可以拿到现在的投资组合信息
# 使用order_shares(id_or_ins, amount)方法进行落单
# TODO: 开始编写你的算法吧！
# 因为策略需要用到均线，所以需要读取历史数据
#做成交量低迷排名
ranking_list = pd.DataFrame(columns=[
"code", "volume_division"
]) #columes: "code",“volume_division”， rows:每个股票占一行
for stock in context.fundamental_df.columns.values:
volumes = history_bars(stock, context.LONGPERIOD + 1, '1d', 'volume')
volumes = [float(f) for f in volumes]
volumes = np.array(volumes)
# 使用talib计算长短两根均线，均线以array的格式表达
short_avg = talib.SMA(volumes, context.SHORTPERIOD)
long_avg = talib.SMA(volumes, context.LONGPERIOD)
plot("short avg", short_avg[-1])
plot("long avg", long_avg[-1])
volume_division = long_avg[-1] / short_avg[-1] #越大越无人问津
ranking_list = ranking_list.append(
{
"code": stock,
"volume_division": volume_division
},
ignore_index=True)
ranking_list = ranking_list.sort(["volume_division"], ascending=False)
logger.info(ranking_list)
for i in range(0, 10):
stock = ranking_list.iloc[i, 0]
# 计算现在portfolio中股票的仓位
cur_position = context.portfolio.positions[stock].quantity
# 首次买入条件,应该添加只买业绩好的
if (cur_position == 0 and ranking_list.iloc[i, 1] > 2):
# 买入
# 计算现在portfolio中的现金可以购买多少股票
if (context.portfolio.cash > 10000.0):
shares = 10000 / bar_dict[stock].close
else:
shares = context.portfolio.cash / bar_dict[stock].close
order_shares(stock, shares)
for stock in context.portfolio.positions:
volumes = history_bars(stock, context.LONGPERIOD + 1, '1d', 'volume')
volumes = [float(f) for f in volumes]
volumes = np.array(volumes)
# 使用talib计算长短两根均线，均线以array的格式表达
short_avg = talib.SMA(volumes, context.SHORTPERIOD)
long_avg = talib.SMA(volumes, context.LONGPERIOD)
plot("short avg", short_avg[-1])
plot("long avg", long_avg[-1])
# 计算现在portfolio中股票的仓位
cur_position = context.portfolio.positions[stock].quantity
# 清仓条件：
# 成交量太多或者达到止盈条件
try:
if ((2 * long_avg[-1] < short_avg[-1]
or context.portfolio.positions[stock].market_value /
context.portfolio.positions[stock].quantity /
context.portfolio.positions[stock].avg_price > 1.06)
and cur_position > 0):
# 进行清仓
order_target_value(stock, 0)
except:
pass
# 持仓时买入条件,做T
try:
if (cur_position > 0
and context.portfolio.positions[stock].market_value < 20000
and context.portfolio.positions[stock].market_value /
context.portfolio.positions[stock].quantity /
context.portfolio.positions[stock].avg_price < 0.92):
# 买入
if (context.portfolio.cash > 5000.0):
shares = 5000 / bar_dict[stock].close
else:
shares = context.portfolio.cash / bar_dict[stock].close
order_shares(stock, shares)
except:
pass
# 1.先测试单进程 cpu == 1
strat_time = np.array([1, 9, 20])
end_time = np.array([27, 16, 20])
a = 1
b = 10
c = 10
if True == a:
df_time_list = [['2019-10-10 09:15:00', '2019-12-10 16:25:00']]
s_time, e_time = df_time_list[0]
df = get_local_hsi_csv(s_time, e_time, datapath) # 获取本地数据
df000 = df.copy()
time0 = time.process_time()
df000 = df.copy()
df000['ma'] = talib.SMA(df000['close'], 5000)
# df000['ma2']=talib.MA(df000['close'],5000)
#[44.000 140.000 30.000 50.000 140.000]
df = zsy_bollin_01(df000, para=[90, 2])
print(df.tail())
huice_hsi(df,
bzj=1 / 5,
leverage_rate=1,
c_rate=20,
hycs=50,
slip=1,
min_margin_rate=(0.8),
is_print=True)
exit()
res0, resdf = duojincheng_backtesting(
import numpy as np import talib close = np.random.random(100) output = talib.SMA(close)
def backtest_KD_pick(mainParament):
userInfo = get_user_info.data_user_info(mainParament.money_start,mainParament.date_start,mainParament.date_end)
Temp_result_pick = pd.DataFrame(columns=['date','選股數量'])
Temp_table = get_stock_history.get_stock_history(2330,"2005-01-01",reGetInfo = False,UpdateInfo = False)
All_stock_signal = dict()
ROE_record_day = datetime.strptime("2000-01-01","%Y-%m-%d")
buy_data = pd.DataFrame(columns = ['Date','code']).set_index('Date')
sell_data = pd.DataFrame(columns = ['Date','code']).set_index('Date')
ROE_data = {}
add_one_day = userInfo.add_one_day
changeDateMonth = tools.changeDateMonth
get_ROE_range = get_stock_history.get_ROE_range
MixDataFrames = tools.MixDataFrames
sell_stock = userInfo.sell_stock
buy_all_stock = userInfo.buy_all_stock
for index,row in Temp_table.iterrows():
if index < userInfo.now_day:
continue
while userInfo.now_day != index:
if add_one_day() == False:
break
has_trade = False
#ROE篩選
if ROE_record_day <= index:
ROE_data = {}
ROE_record_day = changeDateMonth(index,1)
ROE_data["ROE_data_1"] = pd.DataFrame(get_ROE_range(changeDateMonth(index,-3),0,999))
ROE_data["ROE_data_1"].rename(columns={'ROE':'ROE_data_1'},inplace=True)
ROE_data["ROE_data_2"] = pd.DataFrame(get_ROE_range(changeDateMonth(index,-6),0,999))
ROE_data["ROE_data_2"].rename(columns={'ROE':'ROE_data_2'},inplace=True)
ROE_data["ROE_data_3"] = pd.DataFrame(get_ROE_range(changeDateMonth(index,-9),0,999))
ROE_data["ROE_data_3"].rename(columns={'ROE':'ROE_data_3'},inplace=True)
ROE_data["ROE_data_4"] = pd.DataFrame(get_ROE_range(changeDateMonth(index,-12),0,999))
ROE_data["ROE_data_4"].rename(columns={'ROE':'ROE_data_4'},inplace=True)
mask = MixDataFrames(ROE_data)
ROE_data_result =(mask["ROE_data_1"]+mask["ROE_data_2"]+mask["ROE_data_3"]+mask["ROE_data_4"])/4
ROE_data_result = ROE_data_result.dropna()
ROE_data_mask = mask["ROE_data_1"]> ROE_data_result
ROE_data = ROE_data_mask[ROE_data_mask]
for key,value in ROE_data.iteritems():#先算出股票的買賣訊號
if get_stock_history.check_no_use_stock(key) == True:
print('get_stock_price: ' + str(key) + ' in no use')
continue
if All_stock_signal.__contains__(key):
continue
table = get_stock_history.get_stock_history(key,"2005-01-01",reGetInfo = False,UpdateInfo = False)
table_K,table_D = talib.STOCH(table['High'],table['Low'],table['Close'],fastk_period=50, slowk_period=20, slowk_matype=0, slowd_period=20, slowd_matype=0)
table_sma10 = talib.SMA(np.array(table['Close']), 10)
table_sma240 = talib.SMA(np.array(table['Close']), 240)
signal_buy = (table_K > table_D)
signal_sell = (table_K < table_D)
signal_sma10 = table.Close < table_sma10
signal_sma240 = table.Close > table_sma240
signal = signal_buy.copy()
signal = (signal_sma10 & signal_sma240 & signal_buy)
signal[signal_sell] = -1
All_stock_signal[key] = signal
#找出買入訊號跟賣出訊號-------------------------
buy_numbers = []
sell_numbers = []
for i,value in All_stock_signal.items():
try:
if value[index] == True:
if ROE_data.index.__contains__(int(i)):
if ROE_data[int(i)] == True:
buy_numbers.append(i)
elif value[index] < 0 :
sell_numbers.append(i)
except:
#print('error:' + str(index) + ' at ' + str(i))
continue
#出場訊號篩選--------------------------------------
if len(userInfo.handle_stock) > 0:
Temp_data = userInfo.handle_stock
for key,value in list(Temp_data.items()):
if sell_numbers.__contains__(int(key)):
sell_stock(key,value.amount)
has_trade = True
#入場訊號篩選--------------------------------------
if len(buy_numbers) > 0 :
Temp_buy = pd.DataFrame(columns={'code','volume'})
for number in buy_numbers:
volume = get_stock_history.get_stock_price(number,tools.DateTime2String(userInfo.now_day),get_stock_history.stock_data_kind.Volume)
Temp_buy = Temp_buy.append({'code':str(number),'volume':volume},ignore_index = True)
Temp_buy = Temp_buy.sort_values(by='volume', ascending=False).set_index('code')
buy_all_stock(Temp_buy)
has_trade = True
#更新資訊--------------------------------------
if has_trade:
if len(buy_numbers) != 0:
buy_data = buy_data.append({'Date':index,'code':buy_numbers},ignore_index = True)
if len(sell_numbers) != 0:
sell_data = sell_data.append({'Date':index,'code':sell_numbers},ignore_index = True)
userInfo.Record_userInfo()
userInfo.Recod_tradeInfo()
Temp_result_pick = Temp_result_pick.append({'date':userInfo.now_day,
'選股數量':len(buy_numbers)},ignore_index = True)
buy_data = buy_data.set_index('Date')
buy_data.to_csv('buy.csv')
sell_data = sell_data.set_index('Date')
sell_data.to_csv('sell.csv')
#最後總結算----------------------------
Temp_result_pick.set_index('date',inplace=True)
Temp_alldata = tools.MixDataFrames({'draw':userInfo.Temp_result_draw,'pick':Temp_result_pick},'date')
Temp_alldata = tools.MixDataFrames({'all':Temp_alldata,'userinfo':userInfo.Temp_result_All},'date')
userInfo.Temp_result_draw.set_index('date').to_csv('backtestdata.csv')
userInfo.Temp_trade_info.set_index('date').to_csv('backtesttrade.csv')
Temp_alldata.set_index('date').to_csv('backtestAll.csv')
return userInfo.Temp_result_draw.set_index('date')
DataDict = {}
for file in filelist:
key = file.split('.')[0]
DataDict[key] = pd.read_csv(raw_data_dir + file, index_col=0)
format = lambda x: '%.4f' % x
#计算指标
for key in DataDict:
data = DataDict[key]
data.sort_index(inplace=True) #根据日期进行排序
close = np.array(data['close'])
high = np.array(data['high'])
low = np.array(data['low'])
volume = np.array(data['volume'])
data['SMA5'] = ta.SMA(close, timeperiod=5) #5日均线
data['SMA10'] = ta.SMA(close, timeperiod=10) #10日均线
data['SMA30'] = ta.SMA(close, timeperiod=30) #30日均线
data['SMA60'] = ta.SMA(close, timeperiod=60) #60日均线
#upper, middle, lower = ta.BBANDS(close, matype=ta.MA_Type.T3)
data['DEMA5'] = ta.DEMA(close, timeperiod=5) #5日双指数平均
data['DEMA10'] = ta.DEMA(close, timeperiod=10) #10日双指数平均
data['DEMA30'] = ta.DEMA(close, timeperiod=30) #30日双指数平均
data['DEMA60'] = ta.DEMA(close, timeperiod=60) #60日双指数平均
data['SAR'] = ta.SAR(high, low)
data['WMA'] = ta.WMA(close, timeperiod=10)
data['MIDPOINT'] = ta.MIDPOINT(close, timeperiod=10)
data['MIDPRICE'] = ta.MIDPRICE(high, low, timeperiod=10)
data['MFI'] = ta.MFI(high, low, close, volume, timeperiod=10)
data['MOM10'] = ta.MOM(close, timeperiod=10) #动量指标
data['MOM5'] = ta.MOM(close, timeperiod=5) #动量指标
def find_ma_crossovers(serverSite):
crossovers = []
bt_tickers = read_from_database("SELECT DISTINCT ticker FROM backtestdata;",serverSite).ticker.tolist()
for ticker in bt_tickers:
bt_data = read_from_database("SELECT ticker, timestamp, open, high ,low, close, volume FROM backtestdata WHERE ticker ='"+ticker+"';",serverSite)
#Talib need the oldest data to be first
bt_data = bt_data.iloc[::-1]
bt_data["EMA"+str(9)] = talib.EMA(bt_data.close, timeperiod = 9)
bt_data["EMA"+str(20)] = talib.EMA(bt_data.close, timeperiod = 20)
bt_data["EMA"+str(50)] = talib.SMA(bt_data.close, timeperiod = 50)
bt_data["EMA"+str(100)] = talib.SMA(bt_data.close, timeperiod = 100)
bt_data["EMA"+str(200)] = talib.SMA(bt_data.close, timeperiod = 200)
bt_data["MACD"],bt_data["MACDsignal"], bt_data["MACDhist"] = talib.MACD(bt_data.close) #Using default params
bt_data["RSI"] = talib.RSI(bt_data.close) #Using default params
bt_data["ATR"] = talib.ATR(bt_data.high, bt_data.low,bt_data.close)
bt_data["TATR"] = talib.TRANGE(bt_data.high, bt_data.low,bt_data.close)
bt_data["BBUpper"],bt_data["BBmiddle"], bt_data["BBlower"] = talib.BBANDS(bt_data.close, timeperiod=20, nbdevup=2, nbdevdn=2, matype=0) #Using default params
bt_data["dojiDH"] = talib.CDLDRAGONFLYDOJI(bt_data.open, bt_data.high, bt_data.low, bt_data.close)
bt_data["hammer"] = talib.CDLHAMMER(bt_data.open, bt_data.high, bt_data.low, bt_data.close)
bt_data["CDLDOJI"] = talib.CDLDOJI(bt_data.open, bt_data.high, bt_data.low, bt_data.close)
#Dropping NAs
bt_data.dropna(inplace = True)
#Find the crossovers
#Extremely bad solution. Im sure there are better ways
crossAbove = []
crossBelow = []
for key, row in enumerate(bt_data.iterrows()):
if(bt_data.iloc[key,5] > bt_data.iloc[key,8] and bt_data.iloc[key-1,5] < bt_data.iloc[key-1,8]):
crossAbove.append(True)
else:
crossAbove.append(False)
for key, row in enumerate(bt_data.iterrows()):
if(bt_data.iloc[key,5] < bt_data.iloc[key,8] and bt_data.iloc[key-1,5] > bt_data.iloc[key-1,8]):
crossBelow.append(True)
else:
crossBelow.append(False)
bt_data["crossBelow"] = crossBelow
bt_data["crossAbove"] = crossAbove
class Trade:
def __init__(self,ticker, targetEntry, stoploss, crossrow, active = True):
self.ticker = ticker
self.targetEntry = targetEntry
self.actualEntry = 0
self.stoploss = stoploss
self.target = 0
self.crossrow = crossrow
self.searching_for_entry = True
self.active = active
def tradeFoundLong(self,actualEntry):
self.actualEntry = actualEntry
self.target = self.actualEntry + (self.actualEntry- self.stoploss) * 2
self.searching_for_entry = False
#print(self.entry, self.target, self.stoploss)
def tradeFoundShort(self,actualEntry):
self.actualEntry = actualEntry
self.target = self.actualEntry - (self.stoploss - self.actualEntry) * 2
self.searching_for_entry = False
def deactivateTrade(self):
self.active = False
#Looking at the longs!
#Check if hit stoploss or target
count_target = 0
count_stop = 0
count_started_trade = 0
active_trade = Trade("Ticker",1,1,1, active = False) #Dummy trade, not important for result but excecution needs once to start with.
for key, row in enumerate(bt_data.iterrows()):
if (row[1].crossAbove):
active_trade = Trade(ticker, targetEntry = row[1].high, stoploss = row[1].EMA20 , crossrow = row[0])
print("Started search at row",row[0])
#Enter at open if there has been a big gap durin pre and post market, othervise assume we enter at the highs
if (active_trade.targetEntry < row[1].high and active_trade.searching_for_entry and active_trade.active):
if (row[1].open > active_trade.targetEntry):
active_trade.tradeFoundLong(actualEntry = row[1].open)
else:
(active_trade.tradeFoundLong(actualEntry = active_trade.targetEntry))
print("Started trade at row", row[0], "from", active_trade.crossrow)
count_started_trade += 1
#Assume if it the candle hits out target we get out
if (active_trade.target < row[1].high and active_trade.searching_for_entry == False and active_trade.active):
active_trade.deactivateTrade()
crossovers.append([active_trade.ticker, active_trade.crossrow,"Target"])
print("Target met at row", row[0], " from row ", active_trade.crossrow)
count_target += 1
#If the candle closes above our stop, stop out
if (active_trade.stoploss > row[1].close and active_trade.searching_for_entry == False and active_trade.active):
active_trade.deactivateTrade()
crossovers.append([active_trade.ticker, active_trade.crossrow,"Stoploss"])
print("Stoploss met at row", row[0], " from row ", active_trade.crossrow)
count_stop += 1
print("Trades started:",count_started_trade ,"Stops:", count_stop, "Targets:",count_target,"from total", sum(bt_data.crossAbove))
#Shorts!!
count_target = 0
count_stop = 0
count_started_trade = 0
active_trade = Trade("Ticker",1,1,1, active = False) #Dummy trade, not important for result but excecution needs once to start with.
for key, row in enumerate(bt_data.iterrows()):
if (row[1].crossBelow):
active_trade = Trade(ticker,targetEntry = row[1].low, stoploss = row[1].EMA20 , crossrow = row[0])
print("Started search at row",row[0])
#Enter at open if there has been a big gap durin pre and post market, othervise assume we enter at the lows
if (active_trade.targetEntry < row[1].low and active_trade.searching_for_entry and active_trade.active):
if (row[1].open < active_trade.targetEntry):
active_trade.tradeFoundShort(actualEntry = row[1].open)
else:
(active_trade.tradeFoundShort(actualEntry = active_trade.targetEntry))
print("Started trade at row", row[0], "from", active_trade.crossrow)
count_started_trade += 1
#Assume if it the candle hits out target we get out
if (active_trade.target < row[1].low and active_trade.searching_for_entry == False and active_trade.active):
active_trade.deactivateTrade()
crossovers.append([active_trade.ticker, active_trade.crossrow,"Target"])
print("Target met at row", row[0], " from row ", active_trade.crossrow)
count_target += 1
#If the candle closes above our stop, stop out
if (active_trade.stoploss < row[1].close and active_trade.searching_for_entry == False and active_trade.active):
active_trade.deactivateTrade()
crossovers.append([active_trade.ticker, active_trade.crossrow,"Stoploss"])
print("Stoploss met at row", row[0], " from row ", active_trade.crossrow)
count_stop += 1
print("Trades started:",count_started_trade ,"Stops:", count_stop, "Targets:",count_target,"from total", sum(bt_data.crossAbove))
return crossovers
def BBandMA(df, count, acmroi, winnum, winfact,revenue_sum,avg_return,total_return_ratio, MDD,sharpe_ratio):
#設定初始值
#for stk_no in range(df):
#total_trade_record=pd.DataFrame(data=None, index=None,columns = ["stockid","bought_date","bought_time","bought_price","sold_date","sold_time","sold_price","QTY","cash_account","LS_type"])
trade_record = pd.DataFrame(data=None, index=None,columns = ["stockid","bought_date","bought_time","bought_price","sold_date","sold_time","sold_price","QTY","cash_account","LS_type"])
df["stockid"]= np.nan
df["bought_date"]= np.nan
df["bought_time"]= np.nan
df["bought_price"]= np.nan
df["sold_date"]= np.nan
df["sold_time"]= np.nan
df["sold_price"]= np.nan
df["QTY"]=1000
df["cash_account"]= np.nan
df["LS_type"]=0
df['BBXBuy'] = np.nan
df['BBYBuy'] = np.nan
df['BBXSell'] = np.nan
df['BBYSell'] = np.nan
#計算BBand上, 中, 下線
close = np.array(df['close'], dtype=float)
upper, middle, lower = talib.BBANDS(close, timeperiod=10, nbdevup=0.03, nbdevdn=2, matype=MA_Type.T3)
df['BBupper'] = upper
df['BBlower'] = lower
vol =df['Quantity'].as_matrix().astype("float64")
df['volMA'] = talib.SMA(vol, timeperiod=30)
flag = False
buyprice=[]
sellprice=[]
win = 0
loss = 0
roi = 0
revenue=0
total_return=0
revenue_sum=0
for i in range(len(df)):
if (flag == False) & (float(df['highest'].iloc[i])> float(df['BBupper'].iloc[i])*1.075) :
#if (flag == False) & (float(df['highest'].iloc[i]) > float(df['BBupper'].iloc[i])) & (float(df['Quantity'].iloc[i]) > (float(df['volMA'].iloc[i]))) & (float(df['close'].iloc[i])*1.01 <(float(df['BBupper'].iloc[i]))) :
sellprice= df['close'].iloc[i]
df['BBXSell'].iloc[i] = df['MATCH_TIME'].iloc[i]
df['BBYSell'].iloc[i] = sellprice
df["stockid"].iloc[i]=df['STOCK_SYMBOL'].iloc[i]
df["sold_date"].iloc[i]=df['date'].iloc[i]
df["sold_time"].iloc[i]=df['MATCH_TIME'].iloc[i]
df["sold_price"].iloc[i]=df['close'].iloc[i]
#df["cash_account"].iloc[i]= round((5000000-float(buyprice)*1000*1.001425*0.6),0)
flag = True
print("BB賣價"+str(sellprice))
if (flag == True) & (float(df['lowest'].iloc[i])*0.953>=float(df['BBlower'].iloc[i])<float(df['BBupper'].iloc[i])) :
buyprice=df['close'].iloc[i+6]
df['BBXBuy'].iloc[i] = df['MATCH_TIME'].iloc[i+6]
df['BBYBuy'].iloc[i] = buyprice
print("BB買進價"+str(buyprice))
df["stockid"].iloc[i]=df['STOCK_SYMBOL'].iloc[i+6]
df["bought_date"].iloc[i]=df['date'].iloc[i+6]
df["bought_time"].iloc[i]=df['MATCH_TIME'].iloc[i+6]
df["bought_price"].iloc[i]=df['close'].iloc[i+6]
#df["cash_account"].iloc[i]= round( (5000000+float(buyprice* 1.001425*0.6-sellprice* (1-0.001425*0.6-0.001))*1000),0)
df["LS_type"].iloc[i]=2
count += 1
flag = False
[roi, winnum] =roical(buyprice,sellprice, winnum)
acmroi += roi
[loss, win] = winfactor(buyprice, sellprice, loss, win)
revenue=((sellprice*0.995575)- (buyprice*1.001425))*1000
revenue_sum +=round((revenue),2)
premium=revenue/((buyprice*1.001425)*1000)
total_return +=premium
avg_return=(total_return/count)
total_return_ratio=round((revenue_sum/5000000),3)*100
#return_standard+=premium.std()#這是以半年投資報酬標準差,所以標準差將會半年化算Sharp指數
#sharpe_ratio=((avg_return - 0.01095/2)/ return_standard)*100
if (flag == True & i==(len(df)-1)):
df['BBXBuy'][i] = sellprice
df['BBYBuy'][i] = df['MATCH_TIME'].iloc[i]
df['BBXBuy'][i] = buyprice
df["stockid"].iloc[i]=df['STOCK_SYMBOL'].iloc[i]
df["bought_date"].iloc[i]=df['date'].iloc[i]
df["bought_time"].iloc[i]=df['MATCH_TIME'].iloc[i]
df["bought_price"].iloc[i]=df['close'].iloc[i]
#df["cash_account"].iloc[i]=round( (5000000+float(buyprice* (1+0.001425*0.6)-sellprice*(1-0.001425*0.6-0.001))*1000),0)
df["LS_type"].iloc[i]=2
count += 1
[roi, winnum] = roical(buyprice, sellprice, winnum)
acmroi += roi
[loss, win] = winfactor(buyprice, sellprice, loss, win)
print("BB最後1天買入價"+str(buyprice))
revenue=((sellprice*0.995575)- (buyprice*1.001425))*1000
revenue_sum +=round((revenue),2)
premium=revenue/(buyprice*1.001425)
total_return +=premium
avg_return=(total_return/count)
total_return_ratio=round((revenue_sum/5000000),3)*100
#return_standard+=premium.std()#這是以半年投資報酬標準差,所以將會半年化算Sharp指數
#sharpe_ratio=((avg_return - 0.01095/2)/ return_standard)*100
elif ( flag == False& i==(len(df)-10)):
break
if (loss == 0):
loss = 1
winvar = win / loss
if (count == 0):
count = 0.01
#str1 = 'BBand策略: ' + '交易次數 = '+ str(count) + ' 次; ' + '累計報酬率 = ' + str(round(acmroi*100, 2)) + '%; ' + '勝率 = ' + str(round((winnum/count)*100,2)) + '%' + '; 獲利因子 = ' + str(round(winvar, 2))+ '%' + '; 總收益 = ' + str(round(revenue_sum, 2)+ '; 平均報酬率 = ' + str(round(avg_return*100, 2))+ '%'+ '; 夏普指數 = ' + str(round(sharpe_ratio*100, 2))+ '%'+ '; MDD = ' + str(round(MDD, 2))+ '; 總資產報酬率 = ' + str(round(total_return_ratio*100, 2))+ '%')
trade_record.index.names = ['trade_no']
trade_record["stockid"]= df["stockid"]
trade_record["bought_date"]=df["bought_date"]
trade_record["bought_time"]=df["bought_time"]
trade_record["bought_price"]=df["bought_price"]
trade_record["sold_date"]=df["sold_date"]
trade_record["sold_time"]= df["sold_time"]
trade_record["sold_price"]=df["sold_price"]
#trade_record["cash_account"]=df["cash_account"]
trade_record["offset_date"] = trade_record[["bought_date","sold_date"]].max(axis=1) #抓出平倉日
#trade_record =trade_record.sort_values(by=["offset_date",'trade_no'])
#trade_record = trade_record.drop(["bought_date"],axis=1) #將多餘的紀錄刪除
#trade_record.loc[:,["offset_date"]].dropna(axis=1)
#trade_record.loc[:,["bought_date"],["bought_time"],["bought_price"]].sort_values(by=["bought_date"])
trade_record.loc[:,["bought_date"]].dropna(axis=1)
#trade_record.loc[:,["sell_date"],["sell_time"],["sell_price"]].sort_values(by=["sold_date"])
#trade_record =trade_record.reset_index(drop=True) #重設index
#trade_record["LS_type"]=df["LS_type"]=2
#trade_record["Algorithm"] =3
#trade_record["QTY"]=1000
#total_trade_record.append(trade_record)
trade_record.to_csv(r"./BBands_record.csv", sep = ",")
#print(str1)
#print(trade_record)
return (count, acmroi, winnum, winvar,revenue_sum,avg_return,total_return_ratio, MDD,sharpe_ratio)
for cs4 in range(3, 6, 1):
# for cs5 in range(1, 2, 1):
# for cs6 in range(1, 3, 1):
canshu_list.append([cs1, cs2, cs3, cs4, 10])
df_time_list = [['2016-1-10 09:15:00', '2019-12-28 16:25:00']]
s_time, e_time = df_time_list[0]
name = f'{filename}_{s_time[:4]}_{e_time[:4]}_3min'
canshu_list = canshu_list[:10]
df = get_local_hsi_csv(s_time, e_time, datapath) # 获取本地数据
df = transfer_to_period_data(df, rule_type='3T')
print('数据大小', df.shape)
df_day = transfer_to_period_data(df, rule_type='1D')
df_day['atr_day'] = talib.SMA(
talib.ATR(df_day['close'], df_day['high'], df_day['low'], 2),
2) * 0.8
df_day['candle_begin_time'] = df_day['candle_begin_time'].apply(
lambda x: x + datetime.timedelta(hours=9, minutes=15))
df = pd.merge(df,
df_day[['candle_begin_time', 'atr_day']],
how='outer',
on='candle_begin_time')
df = df[df['volume'] > 0]
df.fillna(method='ffill', inplace=True)
print(df.tail(10))
df000 = df.copy()
if True == 1:
print('参数列表个数：', len(canshu_list))
time0 = time.process_time()
def get_data(type):
data_df = pd.read_csv(FILE_PATH, index_col=[0])
#data_df['date'] = data_df['date'].map(lambda x: dt.strptime(x, '%m/%d/%Y %I:%M:%S %p').date())
data_df['volume'] = data_df['volume'].astype(float)
#data_df = data_df.set_index('date')
# Simple Moving Average
data_df['sma_5'] = talib.SMA(data_df['close'].values, timeperiod=5)
data_df['sma_10'] = talib.SMA(data_df['close'].values, timeperiod=10)
# Exponential Moving Average
data_df['ema_20'] = talib.EMA(data_df['close'].values, timeperiod=20)
# Momentum 6 Month / Momentum 12 Month
data_df['mtm6_mtm12'] = talib.MOM(
data_df['close'].values, timeperiod=126) / talib.MOM(
data_df['close'].values, timeperiod=252)
# Stochastic Relative Strength Index
data_df['fastk'], data_df['fastd'] = talib.STOCHRSI(
data_df['close'].values,
timeperiod=14,
fastk_period=5,
fastd_period=3,
fastd_matype=0)
# Rate Of Change
data_df['roc_10'] = talib.ROC(data_df['close'].values, timeperiod=10)
# Bollinger Bands
data_df['bband_upper'], data_df['bband_middle'], data_df[
'bband_lower'] = talib.BBANDS(data_df['close'].values,
timeperiod=5,
nbdevup=2,
nbdevdn=2,
matype=0)
# Moving Average Convergence Divergence
data_df['macd'], data_df['macdsignal'], data_df['macdhist'] = talib.MACD(
data_df['close'].values, fastperiod=12, slowperiod=26, signalperiod=9)
# Chaikin A/D Oscillator
data_df['adosc'] = talib.ADOSC(data_df['high'],
data_df['low'],
data_df['close'],
data_df['volume'],
fastperiod=3,
slowperiod=10)
# Commodity Channel Index
data_df['cci_14'] = talib.CCI(data_df['high'].values,
data_df['low'].values,
data_df['close'].values,
timeperiod=14)
# Average True Range
data_df['atr_14'] = talib.ATR(data_df['high'].values,
data_df['low'].values,
data_df['close'].values,
timeperiod=14)
# Pivot Points
dfpp = finta.PIVOT(data_df[["open", "high", "low", "close", "volume"]])
dfpp.drop(columns=['s3', 's4', 'r3', 'r4'], inplace=True)
pd.concat([data_df, dfpp], axis=1)
# Target
data_df['target'] = np.append(data_df['close'][1:].values, [np.nan])
# Drop Rows With NA Values In Any Column
data_df = data_df.dropna(axis=0, how='any')
if type == "reg":
# Popping The Target Column
target = data_df.pop('target').values
elif type == "class":
print("type = classification")
data_df['class'] = np.where((data_df['close'] < data_df['target']), 1,
0)
data_df.drop('target', axis=1, inplace=True)
target = data_df.pop('class').values
return data_df, target
def get_moving_average(prices):
arr = numpy.array(prices)
output = talib.SMA(arr, timeperiod=35)
return list(output)
