def getReturn(data):
if os.path.exists("data.csv"):
data = pd.read_csv("data.csv", index_col="code")
return data
start = "2017-01-01"
end = "2020-07-31"
codes = data.index
names = fromCodeToName(data, codes)
codes = [str(x) for x in codes]
# print(codes)
# print(names)
# 在数据中增加一列计算年化收益率
data["ar"] = 0.0
t = 0
cash = 100000
for code in data.index:
test = backtest.BackTest(FactorStrategy,
start,
end, [str(code)], [names[t]],
cash,
bDraw=False)
result = test.run()
print("第{}次回测，股票代码{}，回测年化收益率{}。".format(t + 1, code, result.年化收益率))
data.loc[code, ["ar"]] = result.年化收益率
t += 1
data.to_csv("data.csv")
return data
def doBacktest(factors, strategy, a1, a2, a3, a4, a5, start, end, cash):
score = scale(factors, a1, a2, a3, a4, a5)
codes = score[-10:].index
name = fromCodeToName(factors, codes)
code = [str(x) for x in codes]
opttest = backtest.BackTest(strategy, start, end, code, name, cash, bDraw = False)
result = opttest.run()
return result, code
def plot(self):
print u'WF组合收益图'
self.all_sample_factor = np.vstack([self.policy_record[1].is_factor, self.policy_record[1].oos_factor])
returns = np.vstack([self.is_resample_return[1], self.oos_resample_return[1]])
dates = np.append(self.is_resample_dates[1], self.oos_resample_dates[1])
is_oos_arr = np.append(np.zeros_like(self.policy_record[1].is_factor), np.ones_like(self.policy_record[1].oos_factor))
for i in np.sort(self.config.WF_dates.keys())[1:]:
self.all_sample_factor = np.vstack([self.all_sample_factor, self.policy_record[i].oos_factor])
returns = np.vstack([returns, self.oos_resample_return[i]])
dates = np.append(dates, self.oos_resample_dates[i])
is_oos_arr = np.append(is_oos_arr, i * np.ones_like(self.policy_record[i].oos_factor))
#print factor.shape, returns.shape, dates.shape
#print is_oos_arr.shape
#print factor
bt = backtest.BackTest(self.all_sample_factor, returns, dates)
bt.run()
bt.stats_info()
bt.two_in_one()
bt.ts_plot()
#is_oos_arr=is_oos_arr,
#title='WF train:%s predict:%s' %(self.config.train_bars, self.config.predict_bars))
#output='/home/dev/FalconPolicy/WalkForward/' + self.config.policy_path.split('/')[-1].split('.py')[0] + '_WF_train_%s_predict_%s.png' %(self.config.train_bars, self.config.predict_bars))
print u'样本外收益图'
self.oos_sample_factor = self.policy_record[1].oos_factor
returns = self.oos_resample_return[1]
dates = self.oos_resample_dates[1]
is_oos_arr = np.ones_like(self.policy_record[1].oos_factor)
for i in np.sort(self.config.WF_dates.keys())[1:]:
self.oos_sample_factor = np.vstack([self.oos_sample_factor, self.policy_record[i].oos_factor])
returns = np.vstack([returns, self.oos_resample_return[i]])
dates = np.append(dates, self.oos_resample_dates[i])
is_oos_arr = np.append(is_oos_arr, i * np.ones_like(self.policy_record[1].oos_factor))
bt = backtest.BackTest(self.oos_sample_factor, returns, dates)
bt.run()
bt.stats_info()
bt.two_in_one()
bt.ts_plot()
def getReturn(data):
if os.path.exists("data.csv"):
data = pd.read_csv("data.csv")
codes = data.code
codes = ["%06d" % x for x in codes]
data.code = codes
return data
start = "2017-01-01"
end = "2020-07-31"
codes = data.code
codes = ["%06d" % x for x in codes]
data.code = codes
# names = fromCodeToName(data, codes)
# print(names)
# print(len(codes), len(names))
# codes = [str(x) for x in codes]
# print(codes)
# print(codes)
# print(names)
# 在数据中增加一列计算年化收益率
data["ar"] = 0.0
t = 0
cash = 100000
# codes.remove("000029")
names = data.name.values.tolist()
# names.remove("深深房Ａ")
ar = []
for code in codes:
test = backtest.BackTest(FactorStrategy,
start,
end, [code], [names[t]],
cash,
bDraw=False)
result = test.run()
print("第{}次回测，股票代码{}，回测年化收益率{}。".format(t + 1, code, result.年化收益率))
ar.append(result.年化收益率)
# data.loc[t, "ar"] = result.年化收益率
# print(data.info())
t += 1
print(data.shape, len(ar))
data["ar"] = ar
print(data.info())
data.to_csv("data.csv")
return data
def doTest(codes, method):
names = fromCodeToName(data, codes)
codes = [str(x) for x in codes]
start = "2010-01-01"
end = "2020-07-01"
cash = 1000000
opttest = backtest.BackTest(FactorStrategy,
start,
end,
codes,
names,
cash,
bDraw=False)
result = opttest.run()
result["选股方法"] = method
result["股票代码"] = codes
print(result)
return result
def testMultiRegress(data):
model = joblib.load("LineRegress.m")
pred_return = model.predict(data.iloc[:, 3:21])
# print(pred_return)
data["pred_ar"] = pred_return
# print(data)
# 排序
data = data.sort_values(by = "pred_ar", ascending = False)
# print(data)
# 取前十个股票作为投资标的
codes = data.index[0:10].values
# print(codes)
names = fromCodeToName(data, codes)
codes = [str(x) for x in codes]
start = "2010-01-01"
end = "2020-07-01"
cash = 1000000
opttest = backtest.BackTest(FactorStrategy, start, end, codes, names, cash, bDraw = True)
result = opttest.run()
print("多元线性回归的回测结果:")
print(result)
def checkResult(strategy, codes, names, start, end, cash = 1000000):
opttest = backtest.BackTest(strategy, start, end, codes, names, cash)
result = opttest.run()
print("回测结果")
print(result)
df_stock['date']=pd.to_datetime(df_stock['time_key'])
df_stock.to_csv(path + filename)
df_stock.index = pd.to_datetime(df_stock.date)
df_stock['openinterest']=0
df_stock = df_stock[['open','high','low','close','volume','openinterest']]
df.update({stock:df_stock})
return df
if __name__ == "__main__":
start = '2017-01-09'
end = '2020-09-05'
code = ['HK.00700']
name = code
df_stock = get_stockdata(code,start,end)
df_cn = pd.read_csv('to_cn.csv',index_col=0)
df_stock = get_stockdata(code,start,end)
backtest = backtest.BackTest(Tech_roll, start, end, code, name,500000,bDraw = True)
result = backtest.run()
#result = backtest.run()
#result = backtest.optRun(myperiod = range(50,150,20))
#print(result)
print(result)
#maxindex = ret[ret == ret.max()].index
#bestResult = result.loc[maxindex,:]
#print(bestResult.loc[:, ["夏普比率", "参数名", "参数值", "年化收益率"]])
tradeBar = self.df_record.loc[date.strftime("%Y-%m-%d"), :]
# print("bar数据", date, data._name)
if len(tradeBar) != 0:
for i in range(len(tradeBar)):
name = tradeBar.iloc[i].证券名称
price = tradeBar.iloc[i].成交均价
stock = tradeBar.iloc[i].成交量
commit = tradeBar.iloc[i].手续费
# 进行交易
self.__doTrade(data, name, price, stock, commit, orderType)
def next(self):
if self.order:
return
self.doTrade()
def stop(self):
self.log("最大回撤:-%.2f%%" % self.stats.drawdown.maxdrawdown[-1],
doprint=True)
if __name__ == "__main__":
# 加载数据，建立数据源
start = "2018-01-01"
end = "2020-07-05"
name = ["300ETF", "nasETF"]
code = ["510300", "513100"]
backtest = backtest.BackTest(TradeStrategy, start, end, code, name)
backtest.run()
backtest.output()
if self.bb.bot > self.datas[0].close:
cash = self.broker.get_cash()
stock = math.ceil(cash / self.dataclose / 100) * 100 - 100
self.order = self.buy(size=stock,
price=self.datas[0].close,
exectype=bt.Order.Market)
else:
# 持仓且突破上轨，卖出
if self.bb.bot < self.datas[0].close:
self.order = self.close()
def notify_order(self, order):
# 有交易提交/被接受，啥也不做
if order.status in [order.Submitted, order.Accepted]:
return
self.order = None
if __name__ == "__main__":
start = "2018-01-01"
end = "2020-07-05"
name = ["nasetf"]
code = ["513100"]
backtest = backtest.BackTest(Bolling, start, end, code, name, 10000)
result = backtest.run()
# backtest.output()
print(result)
result = backtest.optRun(period=range(6, 30))
print(result)
self.bBuy = False
self.dataclose = self.datas[0].close
def next(self):
if self.bBuy == True:
return
else:
cash = self.broker.get_cash()
stock = math.ceil(cash / self.dataclose / 100) * 100 - 100
self.order = self.buy(size=stock, price=self.datas[0].close)
self.bBuy = True
def stop(self):
self.order = self.close()
if __name__ == "__main__":
start = "2018-01-01"
end = "2020-07-05"
name = ["nasetf"]
code = ["513100"]
backtest = backtest.BackTest(SmaCross, start, end, code, name, 10000)
result = backtest.run()
# backtest.output()
print(result)
result = backtest.optRun(pslow=range(6, 30))
print(result.describe())
plt.figure()
plt.plot(result.年化收益率)
plt.savefig("SmaCross参数优化.png")
self.comm += order.executed.comm
elif order.status in [order.Canceled, order.Margin, order.Rejected]:
self.log("交易失败")
self.order = None
# 输出手续费
def stop(self):
self.log("手续费:%.2f 成本比例:%.5f" %
(self.comm, self.comm / self.broker.getvalue()))
if __name__ == "__main__":
start = "2018-01-01"
end = "2020-07-05"
name = ["300etf"]
code = ["510300"]
backtest = backtest.BackTest(TurtleStrategy, start, end, code, name,
100000)
result = backtest.run()
# backtest.output()
print(result)
result = backtest.optRun(long_period=range(20, 40),
short_period=range(5, 15))
plt.figure()
plt.plot(result.参数值, result.年化收益率)
plt.savefig("海龟策略参数优化.png")
ret = result.loc[:, "年化收益率"]
maxindex = ret[ret == ret.max()].index
bestResult = result.loc[maxindex, :]
print(bestResult.loc[:, ["夏普比率", "参数名", "参数值", "年化收益率"]])
(order.executed.price, order.executed.value,
order.executed.comm))
self.comm += order.executed.comm
elif order.status in [order.Canceled, order.Margin, order.Rejected]:
self.log("交易失败")
self.order = None
# 输出手续费
def stop(self):
self.log("手续费:%.2f 成本比例:%.5f" %
(self.comm, self.comm / self.broker.getvalue()))
if __name__ == "__main__":
start = "2018-01-01"
end = "2020-07-05"
name = ["300etf"]
code = ["510300"]
backtest = backtest.BackTest(GridStrategy, start, end, code, name, 100000)
result = backtest.run()
# backtest.output()
print(result)
# 看选择不同网格宽度的效果
result = backtest.optRun(gap=np.arange(0.005, 0.055, 0.005))
plt.figure()
plt.plot(result.参数值, result.年化收益率)
plt.savefig("网格策略宽度优化.png")
ret = result.loc[:, "年化收益率"]
maxindex = ret[ret == ret.max()].index
bestResult = result.loc[maxindex, :]
print(bestResult.loc[:, ["夏普比率", "参数名", "参数值", "年化收益率"]])
# -*- coding: utf-8 -*- """ Created on Wed Oct 31 20:16:35 2018 @author: 46362 """ from MACDstrategy import MACD import backtest path = r"C:\Users\46362\Desktop\OnePy_Old-master" symbol_list = ['000001', '000002'] engine = backtest.BackTest(path, symbol_list, None) engine.initStrategy(MACD) engine.run()