我想出了一个使用mask和for循环的解决方案。这似乎是预期的工作，但好吧，没有for循环的解决方案会更好！

from tabulate import tabulate
import pandas as pd

df = pd.DataFrame({'Home Team':['BAR', 'RMD', 'MNU', 'CHE'],
                   'Away Team':['MNU', 'BAR', 'CHE', 'RMD'],
                   'Home Team Home Wins':[4, 6, 2, 4],
                   'Away Team Away Wins':[3, 5, 3, 4]})

df['Home Team Total Wins'] = None
for i in range(len(df)):
    home_team = df.iloc[i, df.columns.get_loc('Home Team')]
    idx_teams = [df.columns.get_loc('Home Team'), df.columns.get_loc('Away Team')]
    mask = df.iloc[:i+1, idx_teams].values == home_team
    idx_wins = [df.columns.get_loc('Home Team Home Wins'), df.columns.get_loc('Away Team Away Wins')]
    total_wins = df.iloc[:i+1, idx_wins].values[mask].sum()
    df.iloc[i, df.columns.get_loc('Home Team Total Wins')] = total_wins

print(tabulate(df, headers='keys', tablefmt='psql'))

输出：

+----+-------------+-------------+-----------------------+-----------------------+------------------------+
|    | Home Team   | Away Team   |   Home Team Home Wins |   Away Team Away Wins |   Home Team Total Wins |
|----+-------------+-------------+-----------------------+-----------------------+------------------------|
|  0 | BAR         | MNU         |                     4 |                     3 |                      4 |
|  1 | RMD         | BAR         |                     6 |                     5 |                      6 |
|  2 | MNU         | CHE         |                     2 |                     3 |                      5 |
|  3 | CHE         | RMD         |                     4 |                     4 |                      7 |
+----+-------------+-------------+-----------------------+-----------------------+------------------------+