从 pandas dataframe 生成汇总表

Question

I have a log of events from an algorithmic trading bot loaded into a dataframe that looks like this:

          datetime_long     high     low       trade_direction  trade_entry  trade_exit
159  2021-02-05 10:15:00   88.915   88.6150              LE          1.0         0.0
160  2021-02-05 10:30:00   89.395   88.7800              LX          0.0         1.0
172  2021-02-05 13:30:00   89.090   88.9000              LE          1.0         0.0
177  2021-02-05 14:45:00   89.410   89.1900              LX          0.0         1.0
206  2021-02-08 15:30:00   88.885   88.6600              LE          1.0         0.0
207  2021-02-08 15:45:00   89.080   88.7700              LX          0.0         1.0

"LE" indicates a long entry, "LX" a long exit, "SE" short entry and "SX" short exit. The goal would be to get each "exit" on the same row as the prior "entry" so that I could perform some simple arithmetic computations like determining profit, average hold time, etc.

How does one go about "shifting" only exits up one row so that my entries and corresponding exits are on the same line?

Answer 1

You can use .str.contains() to check if elements in Series contains value.

X_mask = df['trade_direction'].str.contains('X')
X_previous_mask = X_mask.shift(-1).fillna(False)
E_mask = df['trade_direction'].str.contains('E')

You can use

# Entry Rows where Entry next row is Exit
X_previous_E_mask = X_previous_mask & E_mask

# Entry and Exit Rows where Entry next row is Exit
X_and_previous_E_mask = X_previous_E_mask | X_mask

At last, you can use boolean indexing to select rows.

df[mask]