Dataframe中计算最大回撤的处理

第一步：分组

将df按照持仓时间和空仓时间进行分组，分组完成之后就可以在

# ===将每笔交易的start_time设置为发出信号的candle_begin_time，用于分组
    open_mask = (df['signal'].abs() == 1)
    df.loc[open_mask, 'start_time'] = df.loc[open_mask, 'candle_begin_time']
    close_mask = (df['signal'] == 0)
    df.loc[close_mask, 'start_time'] = pd.Timestamp(0)
    df['start_time'].ffill(inplace=True)

第二步：在每个group中计算极大值

在group中的每一行求本轮开仓到该时间的极大值。

# 计算每笔交易的最大回撤
    df['signal_'] = df['signal'].ffill()
    df['dir_close'] = df['signal_'] * df['close']
    df['cum_extreme'] = df[df['start_time'] > pd.Timestamp(0)].groupby('start_time')['dir_close'].cummax()
    df['cum_extreme'] *= df['signal_']

第三步：在每一行计算最大回撤

df['max_loss_pct'] = (1 - df['close'] / df['cum_extreme']) * df['signal_']