Backtrader 文档学习- Broker

Backtrader 文档学习- Broker - Slippage

1.概述

回测无法保证真实的市场条件。无论市场模拟有多好，在真实市场条件下都可能发生滑点。这意味着：

请求的价格可能无法与真实市场的价格匹配
集成的回测broker支持滑点。以下参数可以传递给broker ，具体参数见前一章。

2.How it works

滑点是如何工作的

为了决定何时应用滑点，考虑订单执行类型：

Close 不适用滑点
此订单与close价格匹配，此价格是当天的最后一个价格。滑点不可能发生，因为订单只能在会话的最后一个tick发生，是一个唯一的价格，没有容差。
Market 应用滑点
请检查slip_open是否打开，设置为True。因为Market订单将与下一个K线的开盘价格匹配。
Limit 应用滑点，遵循以下逻辑：
- 如果匹配价格将是开盘价格，则根据参数slip_open，应用滑点。如果应用，则价格永远不会比请求的limit价格更低（差）。
- 如果匹配价格不是limit价格，则应用滑点，并将其限制在high/low上。在这种情况下，slip_mlimit参数决定是否会在超过上限时发生匹配。
- 如果匹配价格是limit价格，则不用滑点。
Stop 一旦订单被触发，则适用与Market订单相同的逻辑
StopLimit 一旦订单被触发，则适用与Limit订单相同的逻辑

滑点试图在模拟和可用数据的限制内，提供最真实的可能回测交易方法。

3.Configuring slippage

每次运行时，cerebro引擎都会为每个运行实例化一个broker，并使用默认参数。有两种方法可以更改设置操作：

使用方法配置滑点：

配置百分比的滑点

BackBroker.set_slippage_perc(perc, slip_open=True, slip_limit=True,
slip_match=True, slip_out=False)

配置固定点的滑点

BackBroker.set_slippage_fixed(fixed, slip_open=True, slip_limit=True,
slip_match=True, slip_out=False)

替换经纪人，如下所示：

import backtrader as btcerebro = bt.Cerebro()
#0.5%的百分比滑点设置
cerebro.broker = bt.brokers.BackBroker(slip_perc=0.005)  # 0.5%

4. 示例

包含一个使用订单执行类型市场和使用信号的多头/空头方法的示例。
应该允许理解逻辑，无滑点运行和供以后参考的初始图：

（1）无滑点

python ./slippage.py
01 2005-03-22 SELL Size: -1 / Price: 3040.55
02 2005-04-11 BUY  Size: +1 / Price: 3088.47
03 2005-04-11 BUY  Size: +1 / Price: 3088.47
04 2005-04-19 SELL Size: -1 / Price: 2948.38
05 2005-04-19 SELL Size: -1 / Price: 2948.38
06 2005-05-19 BUY  Size: +1 / Price: 3034.88
07 2005-05-19 BUY  Size: +1 / Price: 3034.88
08 2005-08-26 SELL Size: -1 / Price: 3258.45
09 2005-08-26 SELL Size: -1 / Price: 3258.45
10 2005-09-13 BUY  Size: +1 / Price: 3353.61
11 2005-09-13 BUY  Size: +1 / Price: 3353.61
12 2005-10-19 SELL Size: -1 / Price: 3330.00
13 2005-10-19 SELL Size: -1 / Price: 3330.00
14 2005-11-14 BUY  Size: +1 / Price: 3405.94
15 2005-11-14 BUY  Size: +1 / Price: 3405.94
16 2006-01-26 SELL Size: -1 / Price: 3578.92
17 2006-01-26 SELL Size: -1 / Price: 3578.92
18 2006-02-03 BUY  Size: +1 / Price: 3677.05
19 2006-02-03 BUY  Size: +1 / Price: 3677.05
20 2006-04-20 SELL Size: -1 / Price: 3820.93
21 2006-04-20 SELL Size: -1 / Price: 3820.93
22 2006-05-02 BUY  Size: +1 / Price: 3839.24
23 2006-05-02 BUY  Size: +1 / Price: 3839.24
24 2006-05-16 SELL Size: -1 / Price: 3711.46
25 2006-05-16 SELL Size: -1 / Price: 3711.46
26 2006-07-04 BUY  Size: +1 / Price: 3664.59
27 2006-07-04 BUY  Size: +1 / Price: 3664.59
28 2006-07-27 SELL Size: -1 / Price: 3649.29
29 2006-07-27 SELL Size: -1 / Price: 3649.29
30 2006-07-28 BUY  Size: +1 / Price: 3671.71
31 2006-07-28 BUY  Size: +1 / Price: 3671.71
32 2006-12-04 SELL Size: -1 / Price: 3935.81
33 2006-12-04 SELL Size: -1 / Price: 3935.81
34 2006-12-19 BUY  Size: +1 / Price: 4121.01
35 2006-12-19 BUY  Size: +1 / Price: 4121.01

看第一个成交的原始数据：

2005-03-22,3040.55,3053.18,3021.66,3050.44,0,0

用的是开盘价成交。

（2）百分比滑点

python ./slippage.py --slip_perc 0.015
01 2005-03-22 SELL Size: -1 / Price: 3040.55
02 2005-04-11 BUY  Size: +1 / Price: 3088.47
03 2005-04-11 BUY  Size: +1 / Price: 3088.47
04 2005-04-19 SELL Size: -1 / Price: 2948.38
05 2005-04-19 SELL Size: -1 / Price: 2948.38
06 2005-05-19 BUY  Size: +1 / Price: 3034.88
07 2005-05-19 BUY  Size: +1 / Price: 3034.88
08 2005-08-26 SELL Size: -1 / Price: 3258.45
09 2005-08-26 SELL Size: -1 / Price: 3258.45
10 2005-09-13 BUY  Size: +1 / Price: 3353.61
11 2005-09-13 BUY  Size: +1 / Price: 3353.61
12 2005-10-19 SELL Size: -1 / Price: 3330.00
13 2005-10-19 SELL Size: -1 / Price: 3330.00
14 2005-11-14 BUY  Size: +1 / Price: 3405.94
15 2005-11-14 BUY  Size: +1 / Price: 3405.94
16 2006-01-26 SELL Size: -1 / Price: 3578.92
17 2006-01-26 SELL Size: -1 / Price: 3578.92
18 2006-02-03 BUY  Size: +1 / Price: 3677.05
19 2006-02-03 BUY  Size: +1 / Price: 3677.05
20 2006-04-20 SELL Size: -1 / Price: 3820.93
21 2006-04-20 SELL Size: -1 / Price: 3820.93
22 2006-05-02 BUY  Size: +1 / Price: 3839.24
23 2006-05-02 BUY  Size: +1 / Price: 3839.24
24 2006-05-16 SELL Size: -1 / Price: 3711.46
25 2006-05-16 SELL Size: -1 / Price: 3711.46
26 2006-07-04 BUY  Size: +1 / Price: 3664.59
27 2006-07-04 BUY  Size: +1 / Price: 3664.59
28 2006-07-27 SELL Size: -1 / Price: 3649.29
29 2006-07-27 SELL Size: -1 / Price: 3649.29
30 2006-07-28 BUY  Size: +1 / Price: 3671.71
31 2006-07-28 BUY  Size: +1 / Price: 3671.71
32 2006-12-04 SELL Size: -1 / Price: 3935.81
33 2006-12-04 SELL Size: -1 / Price: 3935.81
34 2006-12-19 BUY  Size: +1 / Price: 4121.01
35 2006-12-19 BUY  Size: +1 / Price: 4121.01

对比没有变化。符合预期的行为。

执行类型：Market
slip_open 未设置为True，Market订单与下一个bar的开盘价格匹配，不允许open价格用滑点移动。

（3）百分比滑点，slip_open

python ./slippage.py --slip_perc 0.015 --slip_open
01 2005-03-22 SELL Size: -1 / Price: 3021.66
02 2005-04-11 BUY  Size: +1 / Price: 3088.47
03 2005-04-11 BUY  Size: +1 / Price: 3088.47
04 2005-04-19 SELL Size: -1 / Price: 2948.38
05 2005-04-19 SELL Size: -1 / Price: 2948.38
06 2005-05-19 BUY  Size: +1 / Price: 3055.14
07 2005-05-19 BUY  Size: +1 / Price: 3055.14
08 2005-08-26 SELL Size: -1 / Price: 3224.10
09 2005-08-26 SELL Size: -1 / Price: 3224.10
10 2005-09-13 BUY  Size: +1 / Price: 3358.04
11 2005-09-13 BUY  Size: +1 / Price: 3358.04
12 2005-10-19 SELL Size: -1 / Price: 3280.05
13 2005-10-19 SELL Size: -1 / Price: 3280.05
14 2005-11-14 BUY  Size: +1 / Price: 3426.51
15 2005-11-14 BUY  Size: +1 / Price: 3426.51
16 2006-01-26 SELL Size: -1 / Price: 3577.98
17 2006-01-26 SELL Size: -1 / Price: 3577.98
18 2006-02-03 BUY  Size: +1 / Price: 3696.00
19 2006-02-03 BUY  Size: +1 / Price: 3696.00
20 2006-04-20 SELL Size: -1 / Price: 3820.93
21 2006-04-20 SELL Size: -1 / Price: 3820.93
22 2006-05-02 BUY  Size: +1 / Price: 3864.19
23 2006-05-02 BUY  Size: +1 / Price: 3864.19
24 2006-05-16 SELL Size: -1 / Price: 3692.35
25 2006-05-16 SELL Size: -1 / Price: 3692.35
26 2006-07-04 BUY  Size: +1 / Price: 3670.75
27 2006-07-04 BUY  Size: +1 / Price: 3670.75
28 2006-07-27 SELL Size: -1 / Price: 3649.29
29 2006-07-27 SELL Size: -1 / Price: 3649.29
30 2006-07-28 BUY  Size: +1 / Price: 3711.41
31 2006-07-28 BUY  Size: +1 / Price: 3711.41
32 2006-12-04 SELL Size: -1 / Price: 3927.40
33 2006-12-04 SELL Size: -1 / Price: 3927.40
34 2006-12-19 BUY  Size: +1 / Price: 4121.01
35 2006-12-19 BUY  Size: +1 / Price: 4121.01

滑点设置起作用了，非开盘价成交。

SELL滑点成交：
3040.55 *（1 - 0.0015）= 2994.94 ，成交范围在2994.94之内。看第一个成交的原始数据：

2005-03-22,3040.55,3053.18,3021.66,3050.44,0,0

SELL成交价格使用的Low 最低价3021.66 。

BUY滑点成交：
可以立即看到价格发生了变化。并且像交易35号一样，分配的价格最差或相等。也用的是当天开盘价/最高价成交。

2006-12-19,4121.01,4121.01,4085.18,4100.48,0,0

BUY成交价格使用了当天的开盘价，也是最高价成交。

再看一个BUY 的例子，交易30号，原始数据：

2006-07-28,3671.71,3711.41,3659.67,3710.60,0,0

BUY是当天的最高价成交。

通过示例理解滑点？
滑点的意义，就是不可能买在预期的低价，卖单，不可能卖在预期的高点：

买单，可能高于预期价格
买单，可能低于预期价格

（4）百分比滑点，slip_out

当然BT允许在希望的情况下匹配“高”-“低”范围之外的价格，使用“slip_out”。激活它的运行：

python  ./slippage.py --slip_perc 0.015 --slip_open --slip_out
01 2005-03-22 SELL Size: -1 / Price: 2994.94
02 2005-04-11 BUY  Size: +1 / Price: 3134.80
03 2005-04-11 BUY  Size: +1 / Price: 3134.80
04 2005-04-19 SELL Size: -1 / Price: 2904.15
05 2005-04-19 SELL Size: -1 / Price: 2904.15
06 2005-05-19 BUY  Size: +1 / Price: 3080.40
07 2005-05-19 BUY  Size: +1 / Price: 3080.40
08 2005-08-26 SELL Size: -1 / Price: 3209.57
09 2005-08-26 SELL Size: -1 / Price: 3209.57
10 2005-09-13 BUY  Size: +1 / Price: 3403.91
11 2005-09-13 BUY  Size: +1 / Price: 3403.91
12 2005-10-19 SELL Size: -1 / Price: 3280.05
13 2005-10-19 SELL Size: -1 / Price: 3280.05
14 2005-11-14 BUY  Size: +1 / Price: 3457.03
15 2005-11-14 BUY  Size: +1 / Price: 3457.03
16 2006-01-26 SELL Size: -1 / Price: 3525.24
17 2006-01-26 SELL Size: -1 / Price: 3525.24
18 2006-02-03 BUY  Size: +1 / Price: 3732.21
19 2006-02-03 BUY  Size: +1 / Price: 3732.21
20 2006-04-20 SELL Size: -1 / Price: 3763.62
21 2006-04-20 SELL Size: -1 / Price: 3763.62
22 2006-05-02 BUY  Size: +1 / Price: 3896.83
23 2006-05-02 BUY  Size: +1 / Price: 3896.83
24 2006-05-16 SELL Size: -1 / Price: 3655.79
25 2006-05-16 SELL Size: -1 / Price: 3655.79
26 2006-07-04 BUY  Size: +1 / Price: 3719.56
27 2006-07-04 BUY  Size: +1 / Price: 3719.56
28 2006-07-27 SELL Size: -1 / Price: 3594.55
29 2006-07-27 SELL Size: -1 / Price: 3594.55
30 2006-07-28 BUY  Size: +1 / Price: 3726.79
31 2006-07-28 BUY  Size: +1 / Price: 3726.79
32 2006-12-04 SELL Size: -1 / Price: 3876.77
33 2006-12-04 SELL Size: -1 / Price: 3876.77
34 2006-12-19 BUY  Size: +1 / Price: 4182.83
35 2006-12-19 BUY  Size: +1 / Price: 4182.83

第一个SELL单：
3040.55 *（1 - 0.0015）= 2994.94
2994.94 没有在当天的价格中出现，以不在最高和最低价格中成交的滑点价格。

2005-03-22,3040.55,3053.18,3021.66,3050.44,0,0

最后一个BUY单：
价格显然超出了范围。只需查看操作35，它已在“4182.83”处匹配。快速检查本文档中的图表，可以看到该资产从未接近过该价格。

2006-12-19,4121.01,4121.01,4085.18,4100.48,0,0

（5）百分比滑点，no-slip_match

“slip_match”默认为“True”，这意味着BT提供了匹配，无论是带有限制的价格还是不带限制的价格。

python ./slippage.py --slip_perc 0.015 --slip_open --no-slip_match
01 2005-04-15 SELL Size: -1 / Price: 3028.10
02 2005-06-01 BUY  Size: +1 / Price: 3124.03
03 2005-06-01 BUY  Size: +1 / Price: 3124.03
04 2005-10-06 SELL Size: -1 / Price: 3365.57
05 2005-10-06 SELL Size: -1 / Price: 3365.57
06 2005-12-01 BUY  Size: +1 / Price: 3499.95
07 2005-12-01 BUY  Size: +1 / Price: 3499.95
08 2006-02-28 SELL Size: -1 / Price: 3782.71
09 2006-02-28 SELL Size: -1 / Price: 3782.71
10 2006-05-23 BUY  Size: +1 / Price: 3594.68
11 2006-05-23 BUY  Size: +1 / Price: 3594.68
12 2006-11-27 SELL Size: -1 / Price: 3984.37
13 2006-11-27 SELL Size: -1 / Price: 3984.37

2005-04-15,3074.21,3074.21,3013.79,3013.89,0,0

3074.21 * （1 - 0.015 ）= 3028.10
这个价格符合在最高价3074.21 和最低价 3013.79 中间的价格，并且符合滑点。如果是市场上实际交易，这个价格一定是存在的。

结果让人震惊！操作成交的数量从35下降到13。
理由：
禁用“slip_match”会禁止匹配操作，如果滑点将匹配价格推到bar的“高”以上或“低”以下。似乎在请求的滑点“1.5%”左右，有22个操作未能执行。

以上示例应该展示了不同的滑点选项如何一起工作。

5. 代码

#!/usr/bin/env python
# -*- coding: utf-8; py-indent-offset:4 -*-
###############################################################################
#
# Copyright (C) 2015-2023 Daniel Rodriguez
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with this program.  If not, see <http://www.gnu.org/licenses/>.
#
###############################################################################
from __future__ import (absolute_import, division, print_function,unicode_literals)import argparse
import collections
import datetime
import itertoolsimport backtrader as btclass SMACrossOver(bt.Indicator):lines = ('signal',)params = (('p1', 10), ('p2', 30),)def __init__(self):sma1 = bt.indicators.SMA(period=self.p.p1)sma2 = bt.indicators.SMA(period=self.p.p2)self.lines.signal = bt.indicators.CrossOver(sma1, sma2)class SlipSt(bt.SignalStrategy):opcounter = itertools.count(1)def notify_order(self, order):if order.status == bt.Order.Completed:t = ''t += '{:02d}'.format(next(self.opcounter))t += ' {}'.format(order.data.datetime.date())t += ' BUY ' * order.isbuy() or ' SELL't += ' Size: {:+d} / Price: {:.2f}'print(t.format(order.executed.size, order.executed.price))def runstrat(args=None):args = parse_args(args)cerebro = bt.Cerebro()cerebro.broker.set_cash(args.cash)dkwargs = dict()if args.fromdate is not None:fromdate = datetime.datetime.strptime(args.fromdate, '%Y-%m-%d')dkwargs['fromdate'] = fromdateif args.todate is not None:todate = datetime.datetime.strptime(args.todate, '%Y-%m-%d')dkwargs['todate'] = todate# if dataset is None, args.data has been givendata = bt.feeds.BacktraderCSVData(dataname=args.data, **dkwargs)cerebro.adddata(data)cerebro.signal_strategy(SlipSt)if not args.longonly:stype = bt.signal.SIGNAL_LONGSHORTelse:stype = bt.signal.SIGNAL_LONGcerebro.add_signal(stype, SMACrossOver, p1=args.period1, p2=args.period2)if args.slip_perc is not None:cerebro.broker.set_slippage_perc(args.slip_perc,slip_open=args.slip_open,slip_match=not args.no_slip_match,slip_out=args.slip_out)elif args.slip_fixed is not None:cerebro.broker.set_slippage_fixed(args.slip_fixed,slip_open=args.slip_open,slip_match=not args.no_slip_match,slip_out=args.slip_out)cerebro.run()if args.plot:pkwargs = dict(style='bar')if args.plot is not True:  # evals to True but is not Truenpkwargs = eval('dict(' + args.plot + ')')  # args were passedpkwargs.update(npkwargs)cerebro.plot(**pkwargs)def parse_args(pargs=None):parser = argparse.ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter,description='Sample for Slippage')parser.add_argument('--data', required=False,default='./datas/2005-2006-day-001.txt',help='Specific data to be read in')parser.add_argument('--fromdate', required=False, default=None,help='Starting date in YYYY-MM-DD format')parser.add_argument('--todate', required=False, default=None,help='Ending date in YYYY-MM-DD format')parser.add_argument('--cash', required=False, action='store',type=float, default=50000,help=('Cash to start with'))parser.add_argument('--period1', required=False, action='store',type=int, default=10,help=('Fast moving average period'))parser.add_argument('--period2', required=False, action='store',type=int, default=30,help=('Slow moving average period'))parser.add_argument('--longonly', required=False, action='store_true',help=('Long only strategy'))pgroup = parser.add_mutually_exclusive_group(required=False)pgroup.add_argument('--slip_perc', required=False, default=None,type=float,help='Set the value for commission percentage')pgroup.add_argument('--slip_fixed', required=False, default=None,type=float,help='Set the value for commission percentage')parser.add_argument('--no-slip_match', required=False, action='store_true',help=('Match by capping slippage at bar ends'))parser.add_argument('--slip_out', required=False, action='store_true',help=('Disable capping and return non-real prices'))parser.add_argument('--slip_open', required=False, action='store_true',help=('Slip even if match price is next open'))# Plot optionsparser.add_argument('--plot', '-p', nargs='?', required=False,metavar='kwargs', const=True,help=('Plot the read data applying any kwargs passed\n''\n''For example:\n''\n''  --plot style="candle" (to plot candles)\n'))if pargs is not None:return parser.parse_args(pargs)return parser.parse_args()if __name__ == '__main__':runstrat()