Backtrader：实现一个简单的通达信的交易策略

大体了解了Backtrader的运行原理后，尝试把一个通达信的交易策略改编为backtrader策略，并测试一下交易时间是否与通达信中出现的交易时间一致。如果一致，说明改写的策略有效。

通达信中有这样一个简单的“MA交易”指标：

可以在通达信主界面按ctrl+F——F3查找“交易”找到这个指标。

指标中这两行：

MA1:=MA(CLOSE,SHORT);
MA2:=MA(CLOSE,LONG);

修改为Backtrader代码：

        # 收盘价的 para_short 日简单移动平均
        self.ma1 = bt.indicators.SimpleMovingAverage(self.data.close, period=self.params.para_short)
        # 收盘价的 para_long 日简单移动平均
        self.ma2 = bt.indicators.SimpleMovingAverage(self.data.close, period=self.params.para_long)

通达信这两行：

平空开多:=CROSS(MA1,MA2);
平多开空:=CROSS(MA2,MA1);

直接改为Backtrader的交易条件。

根据这个通达信交易指标，制作一个简易附图指标，可以显示出买入卖出点，便于我们与backtrader回测结果进行对比。

MA1:=MA(CLOSE,SHORT);
MA2:=MA(CLOSE,LONG);
CROSS(MA1,MA2);
CROSS(MA2,MA1);

 附图中可以看到买卖日期。

 这个策略在Backtrader中实现后，输出运行结果：

期初资金: 100000.00
2021-01-29, Close, 7.02
2021-02-01, Close, 7.08
2021-02-02, Close, 6.98
2021-02-03, Close, 6.69
2021-02-04, Close, 6.57
2021-02-05, Close, 6.73
2021-02-08, Close, 6.69
2021-02-09, Close, 6.83
2021-02-10, Close, 6.97
2021-02-18, Close, 7.11
2021-02-19, Close, 7.22
2021-02-22, Close, 7.15
2021-02-23, Close, 7.13
2021-02-23, 买入单, 7.13
2021-02-24, 已买入, 价格: 7.13, 费用: 7.13, 佣金 0.01
2021-02-24, Close, 7.11
2021-02-25, Close, 7.12
2021-02-26, Close, 7.36
2021-03-01, Close, 7.48
2021-03-02, Close, 7.33
2021-03-03, Close, 7.32
2021-03-04, Close, 7.39
2021-03-05, Close, 7.50
2021-03-08, Close, 7.48
2021-03-09, Close, 7.42
2021-03-10, Close, 7.40
2021-03-11, Close, 7.39
2021-03-12, Close, 7.65
2021-03-15, Close, 7.61
2021-03-16, Close, 7.86
2021-03-17, Close, 8.13
2021-03-18, Close, 7.97
2021-03-19, Close, 7.92
2021-03-22, Close, 7.82
2021-03-23, Close, 7.74
2021-03-24, Close, 7.72
2021-03-25, Close, 7.68
2021-03-26, Close, 7.77
2021-03-29, Close, 7.83
2021-03-30, Close, 8.07
2021-03-31, Close, 7.92
2021-04-01, Close, 8.02
2021-04-02, Close, 8.00
2021-04-06, Close, 8.04
2021-04-07, Close, 7.95
2021-04-08, Close, 7.91
2021-04-09, Close, 8.09
2021-04-12, Close, 8.29
2021-04-13, Close, 8.27
2021-04-14, Close, 8.67
2021-04-15, Close, 8.97
2021-04-16, Close, 8.85
2021-04-19, Close, 8.80
2021-04-20, Close, 8.48
2021-04-21, Close, 8.35
2021-04-22, Close, 8.41
2021-04-23, Close, 8.37
2021-04-26, Close, 8.30
2021-04-27, Close, 7.76
2021-04-27, 卖出单, 7.76
2021-04-28, 已卖出, 价格: 7.99, 费用: 7.13, 佣金 0.01
2021-04-28, 交易利润, 毛利润 0.86, 净利润 0.84
2021-04-28, Close, 7.81
2021-04-29, Close, 7.66
2021-04-30, Close, 7.73
2021-05-06, Close, 7.73
2021-05-07, Close, 7.73
2021-05-10, Close, 7.81
2021-05-11, Close, 7.86
2021-05-12, Close, 8.03
2021-05-13, Close, 8.03
2021-05-14, Close, 7.97
2021-05-17, Close, 7.93
2021-05-18, Close, 7.84
2021-05-19, Close, 7.89
2021-05-20, Close, 7.83
2021-05-21, Close, 7.79
2021-05-24, Close, 7.81
2021-05-25, Close, 7.89
2021-05-26, Close, 7.92
2021-05-27, Close, 7.89
2021-05-27, 买入单, 7.89
2021-05-28, 已买入, 价格: 7.91, 费用: 7.91, 佣金 0.01
2021-05-28, Close, 8.68
2021-05-31, Close, 9.30
2021-06-01, Close, 8.90
2021-06-02, Close, 8.95
2021-06-03, Close, 9.12
2021-06-04, Close, 8.84
2021-06-07, Close, 8.92
2021-06-08, Close, 8.92
2021-06-09, Close, 9.11
2021-06-10, Close, 8.88
2021-06-11, Close, 8.49
2021-06-15, Close, 8.55
2021-06-16, Close, 8.93
2021-06-17, Close, 8.81
2021-06-18, Close, 8.96
2021-06-21, Close, 9.16
2021-06-22, Close, 9.04
2021-06-23, Close, 9.04
2021-06-24, Close, 8.76
2021-06-25, Close, 8.96
2021-06-28, Close, 9.04
2021-06-29, Close, 8.88
2021-06-30, Close, 9.20
2021-07-01, Close, 9.22
2021-07-02, Close, 8.73
2021-07-05, Close, 8.61
2021-07-06, Close, 9.47
2021-07-07, Close, 9.49
2021-07-08, Close, 9.64
2021-07-09, Close, 9.52
2021-07-12, Close, 10.03
2021-07-13, Close, 9.64
2021-07-14, Close, 9.74
2021-07-15, Close, 9.60
2021-07-16, Close, 9.41
2021-07-19, Close, 9.47
2021-07-20, Close, 9.19
2021-07-21, Close, 9.26
2021-07-22, Close, 9.50
2021-07-23, Close, 9.69
2021-07-26, Close, 9.25
2021-07-27, Close, 8.95
2021-07-27, 卖出单, 8.95
2021-07-28, 已卖出, 价格: 8.95, 费用: 7.91, 佣金 0.01
2021-07-28, 交易利润, 毛利润 1.04, 净利润 1.02
2021-07-28, Close, 9.04
2021-07-29, Close, 9.04
2021-07-30, Close, 8.91
2021-08-02, Close, 9.16
2021-08-03, Close, 9.38
2021-08-04, Close, 9.30
2021-08-05, Close, 9.23
2021-08-06, Close, 9.15
2021-08-09, Close, 9.15
2021-08-10, Close, 9.09
2021-08-11, Close, 9.04
2021-08-12, Close, 9.09
2021-08-13, Close, 9.60
2021-08-16, Close, 9.50
2021-08-16, 买入单, 9.50
2021-08-17, 已买入, 价格: 9.51, 费用: 9.51, 佣金 0.01
2021-08-17, Close, 9.31
2021-08-18, Close, 10.24
2021-08-19, Close, 10.01
2021-08-20, Close, 9.72
2021-08-23, Close, 10.18
2021-08-24, Close, 9.97
2021-08-25, Close, 9.91
2021-08-26, Close, 9.77
2021-08-27, Close, 9.69
2021-08-30, Close, 9.86
2021-08-31, Close, 9.47
2021-09-01, Close, 9.64
2021-09-02, Close, 9.68
2021-09-03, Close, 10.10
2021-09-06, Close, 10.09
2021-09-07, Close, 10.12
2021-09-08, Close, 10.16
2021-09-09, Close, 10.08
2021-09-10, Close, 10.36
2021-09-13, Close, 10.59
2021-09-14, Close, 10.09
2021-09-15, Close, 10.10
2021-09-16, Close, 10.18
2021-09-17, Close, 10.48
2021-09-22, Close, 10.85
2021-09-23, Close, 10.65
2021-09-24, Close, 10.22
2021-09-27, Close, 9.95
2021-09-28, Close, 9.98
2021-09-29, Close, 9.39
2021-09-29, 卖出单, 9.39
2021-09-30, 已卖出, 价格: 9.53, 费用: 9.51, 佣金 0.01
2021-09-30, 交易利润, 毛利润 0.02, 净利润 0.00
2021-09-30, Close, 9.48
2021-10-08, Close, 9.56
2021-10-11, Close, 9.61
2021-10-12, Close, 9.28
2021-10-13, Close, 9.17
2021-10-14, Close, 9.19
2021-10-15, Close, 9.08
2021-10-18, Close, 8.93
2021-10-19, Close, 8.98
2021-10-20, Close, 9.03
2021-10-21, Close, 8.82
2021-10-22, Close, 8.63
2021-10-25, Close, 8.81
2021-10-26, Close, 8.76
2021-10-27, Close, 8.42
2021-10-28, Close, 8.47
2021-10-29, Close, 8.49
2021-11-01, Close, 8.64
2021-11-02, Close, 8.55
2021-11-03, Close, 8.71
2021-11-04, Close, 8.74
2021-11-05, Close, 8.47
2021-11-08, Close, 8.37
2021-11-09, Close, 8.65
2021-11-10, Close, 8.75
2021-11-11, Close, 8.72
2021-11-12, Close, 8.77
2021-11-15, Close, 9.00
2021-11-15, 买入单, 9.00
2021-11-16, 已买入, 价格: 9.00, 费用: 9.00, 佣金 0.01
2021-11-16, Close, 8.94
2021-11-17, Close, 9.00
2021-11-18, Close, 8.87
2021-11-19, Close, 8.89
2021-11-22, Close, 8.99
2021-11-23, Close, 8.99
2021-11-24, Close, 8.85
2021-11-25, Close, 8.86
2021-11-26, Close, 8.65
2021-11-29, Close, 8.44
2021-11-29, 卖出单, 8.44
2021-11-30, 已卖出, 价格: 8.44, 费用: 9.00, 佣金 0.01
2021-11-30, 交易利润, 毛利润 -0.56, 净利润 -0.58
2021-11-30, Close, 8.55
2021-12-01, Close, 8.57
2021-12-02, Close, 8.39
2021-12-03, Close, 8.42
2021-12-06, Close, 8.58
2021-12-07, Close, 8.40
2021-12-08, Close, 8.50
2021-12-09, Close, 8.60
2021-12-10, Close, 8.62
2021-12-13, Close, 8.67
2021-12-14, Close, 8.78
2021-12-15, Close, 8.65
2021-12-15, 买入单, 8.65
2021-12-16, 已买入, 价格: 8.63, 费用: 8.63, 佣金 0.01
2021-12-16, Close, 8.42
2021-12-16, 卖出单, 8.42
2021-12-17, 已卖出, 价格: 8.40, 费用: 8.63, 佣金 0.01
2021-12-17, 交易利润, 毛利润 -0.23, 净利润 -0.25
2021-12-17, Close, 8.32
2021-12-20, Close, 8.37
2021-12-21, Close, 8.43
2021-12-22, Close, 8.47
2021-12-23, Close, 8.30
2021-12-24, Close, 8.53
2021-12-27, Close, 8.69
2021-12-28, Close, 8.61
2021-12-28, 买入单, 8.61
2021-12-29, 已买入, 价格: 8.63, 费用: 8.63, 佣金 0.01
2021-12-29, Close, 8.63
2021-12-30, Close, 8.73
2021-12-31, Close, 8.88
2022-01-04, Close, 8.95
2022-01-05, Close, 8.70
2022-01-06, Close, 8.97
2022-01-07, Close, 8.89
2022-01-10, Close, 8.84
2022-01-11, Close, 8.89
2022-01-12, Close, 9.26
2022-01-13, Close, 9.09
2022-01-14, Close, 9.17
2022-01-17, Close, 9.16
2022-01-18, Close, 8.77
2022-01-19, Close, 8.75
2022-01-20, Close, 8.65
2022-01-21, Close, 8.24
2022-01-21, 卖出单, 8.24
2022-01-24, 已卖出, 价格: 8.24, 费用: 8.63, 佣金 0.01
2022-01-24, 交易利润, 毛利润 -0.39, 净利润 -0.41
2022-01-24, Close, 8.10
2022-01-25, Close, 7.70
2022-01-26, Close, 7.64
2022-01-27, Close, 7.57
期末资金: 100000.64
交易日期与通达信中买卖点出现日期一致。

实现这个例子的完整python代码：

from __future__ import (absolute_import, division, print_function,
                        unicode_literals)
from datetime import datetime  # For datetime objects
# Import the backtrader platform
import backtrader as bt
import pandas as pd

# 创建策略继承bt.Strategy
class TestStrategy(bt.Strategy):
    params = (
        # 均线参数设置15天，15日均线
        ('maperiod', 15),
        ('para_short', 5),
        ('para_long', 20),
    )

    def log(self, txt, dt=None):
        # 记录策略的执行日志
        dt = dt or self.datas[0].datetime.date(0)
        print('%s, %s' % (dt.isoformat(), txt))

    def __init__(self):
        # 保存收盘价的引用
        self.dataclose = self.datas[0].close
        # 跟踪挂单
        self.order = None
        # 买入价格和手续费
        self.buyprice = None
        self.buycomm = None
        # 加入均线指标
        self.sma = bt.indicators.SimpleMovingAverage(self.datas[0], period=self.params.maperiod)
        
        # 收盘价的 para_short 日简单移动平均
        self.ma1 = bt.indicators.SimpleMovingAverage(self.data.close, period=self.params.para_short)
        # 收盘价的 para_long 日简单移动平均
        self.ma2 = bt.indicators.SimpleMovingAverage(self.data.close, period=self.params.para_long)
        
        # MA1 = self.sma_short[0]
        # MA2 = self.sma_long[0]        

    # 交易函数
    def next(self):
        # 记录收盘价
        self.log('Close, %.2f' % self.dataclose[0])

        # 如果有订单正在挂起，不操作
        if self.order:
            return
        
        # 交易条件
        # MA1上穿MA2
        # buyin_flag = self.ma1 > self.ma2
        # MA2上穿MA1
        # sellout_flag = self.ma2 > self.ma1

        # 如果没有持仓则买入
        if not self.position:
            # MA1上穿MA2
            if self.ma1 > self.ma2: 
                # 买入
                self.log('买入单, %.2f' % self.dataclose[0])
                    # 跟踪订单避免重复
                self.order = self.buy()
        else:
            # 如果已经持仓，MA2上穿MA1
            if self.ma2 > self.ma1:
                # 全部卖出
                self.log('卖出单, %.2f' % self.dataclose[0])
                # 跟踪订单避免重复
                self.order = self.sell()
        

    # 订单状态通知，买入卖出都是下单
    def notify_order(self, order):
        if order.status in [order.Submitted, order.Accepted]:
            # broker 提交/接受了，买/卖订单则什么都不做
            return

        # 检查一个订单是否完成
        # 注意: 当资金不足时，broker会拒绝订单
        if order.status in [order.Completed]:
            if order.isbuy():
                self.log(
                    '已买入, 价格: %.2f, 费用: %.2f, 佣金 %.2f' %
                    (order.executed.price,
                     order.executed.value,
                     order.executed.comm))

                self.buyprice = order.executed.price
                self.buycomm = order.executed.comm
            elif order.issell():
                self.log('已卖出, 价格: %.2f, 费用: %.2f, 佣金 %.2f' %
                         (order.executed.price,
                          order.executed.value,
                          order.executed.comm))
            # 记录当前交易数量
            self.bar_executed = len(self)

        elif order.status in [order.Canceled, order.Margin, order.Rejected]:
            self.log('订单取消/保证金不足/拒绝')

        # 其他状态记录为：无挂起订单
        self.order = None

    # 交易状态通知，一买一卖算交易
    def notify_trade(self, trade):
        if not trade.isclosed:
            return
        self.log('交易利润, 毛利润 %.2f, 净利润 %.2f' %
                 (trade.pnl, trade.pnlcomm))

start = '2020-01-01'
end = '2022-01-31'

def get_data(code, start_date, end_date):
    df_tdx = pd.read_feather(r'./dataout/tdx/'+code+r'.day.feather')
    df_tdx.index=pd.to_datetime(df_tdx.date, format = '%Y%m%d')
    df_tdx_b=df_tdx.truncate(before=start_date, after = end_date)
    df_tdx_b['openinterest']=0
    df_tdx_b.rename(columns={'vol':'volume'}, inplace = True)
    df_tdx_b=df_tdx_b[['open','high','low','close','volume','openinterest']]
    return df_tdx_b

dataframe=get_data('sh600851', datetime.strptime(start,'%Y-%m-%d'), datetime.strptime(end,'%Y-%m-%d'))

if __name__ == '__main__':
	
    # 初始化cerebro回测系统设置
    cerebro = bt.Cerebro()

    # 取得股票历史数据
    data = bt.feeds.PandasData(dataname=dataframe, fromdate = datetime.strptime(start,'%Y-%m-%d'), todate = datetime.strptime(end,'%Y-%m-%d'))
	
    # 为Cerebro引擎添加策略
    cerebro.addstrategy(TestStrategy)
    # 加载交易数据
    cerebro.adddata(data)
	
	# 设置投资金额
    cerebro.broker.setcash(100000.0)
    # 设置佣金为0.001,除以100去掉%号
    cerebro.broker.setcommission(commission=0.001)
	
    #获取回测开始时的总资金
    print('期初资金: %.2f' % cerebro.broker.getvalue())
    #运行回测系统
    cerebro.run()
    #获取回测结束后的总资金
    print('期末资金: %.2f' % cerebro.broker.getvalue())

写到这里，我突发奇想，是不是可以写个程序，自动把通达信公式改写为Backtrader策略呢？