为什么我的 zipline schedule_function 不适用于每月计划?回测始终每天运行

问题描述

我正在尝试在 zipline 1.4.1 中安排我的重新平衡功能,但是我的回测不是按月计划运行,而是每天运行。

实际代码

#%%
import zipline.api as zapi
from zipline import run_algorithm
import pandas as pd

#%%
def initialize(context):
    zapi.schedule_function(
        rebalance,algo.date_rules.month_start(),algo.time_rules.market_open(hours=2)
    )

def rebalance(context,data):
    print(f"rebalancing at {zapi.get_datetime()}")

def analyze(context,perf):
    print('We are Now analyzing')

#%%
start_date = pd.Timestamp('2010-1-3',tz='UTC')
end_date = pd.Timestamp('2010-4-3',tz='UTC')

result = run_algorithm(
        start = start_date,end = end_date,initialize=initialize,handle_data=rebalance,# benchmark_returns=benchmarkReturns,capital_base= 10000,analyze=analyze,data_frequency='daily',bundle='csv-bundle')

回测结果:

rebalancing at 2010-01-04 21:00:00+00:00
rebalancing at 2010-01-04 21:00:00+00:00
rebalancing at 2010-01-05 21:00:00+00:00
rebalancing at 2010-01-06 21:00:00+00:00
rebalancing at 2010-01-07 21:00:00+00:00
rebalancing at 2010-01-08 21:00:00+00:00
rebalancing at 2010-01-11 21:00:00+00:00
rebalancing at 2010-01-12 21:00:00+00:00
rebalancing at 2010-01-13 21:00:00+00:00
rebalancing at 2010-01-14 21:00:00+00:00
rebalancing at 2010-01-15 21:00:00+00:00
rebalancing at 2010-01-19 21:00:00+00:00
rebalancing at 2010-01-20 21:00:00+00:00
rebalancing at 2010-01-21 21:00:00+00:00
rebalancing at 2010-01-22 21:00:00+00:00
rebalancing at 2010-01-25 21:00:00+00:00
rebalancing at 2010-01-26 21:00:00+00:00
rebalancing at 2010-01-27 21:00:00+00:00
rebalancing at 2010-01-28 21:00:00+00:00
rebalancing at 2010-01-29 21:00:00+00:00
rebalancing at 2010-02-01 21:00:00+00:00
rebalancing at 2010-02-01 21:00:00+00:00
rebalancing at 2010-02-02 21:00:00+00:00
rebalancing at 2010-02-03 21:00:00+00:00
rebalancing at 2010-02-04 21:00:00+00:00
rebalancing at 2010-02-05 21:00:00+00:00
rebalancing at 2010-02-08 21:00:00+00:00

我尝试将 zipline 降级到 1.3.0 版并得到相同的结果。有没有人遇到同样的问题?有什么解决办法吗?感谢您的友好回复

解决方法

我也有同样的问题。我认为 schedule_function 不适用于 zipline。运行算法

我通过创建一个函数来解决这个问题:

def monthly_cal(cal_name,start,end):
xnys=get_calendar(cal_name)
xnys_cal= xnys.sessions_in_range(start,end)
df=pd.DataFrame(index=xnys_cal)
df['month']=pd.DatetimeIndex(df.index).month
trading_days=df['month'][df['month'] != df['month'].shift(-1)].index
return trading_days

并在初始化中加载它

context.available_dates=monthly_cal('XNYS',end)

并在 handle_data 中检查此条件 如果 context.get_datetime().date() 在 context.available_dates 中: