Python Pandas如何基于多条件计算前期、累计与交易成本列

这类依赖上一行计算结果的递推逻辑，无法直接通过常规向量化操作实现，按ticker分组后逐行维护递推状态即可完成计算，具体实现如下：

实现代码 #

先确保你已经完成了transacted_value、flow_units、cml_units、prev_units这几个前置列的计算，再运行以下代码：

def calc_group_cost(grp):
    prev_costs = []
    cml_costs = []
    cost_transaction = []
    # 初始化组内累计成本初始值
    current_cml = 0

    for _, row in grp.iterrows():
        prev_c = current_cml
        tx_cost = 0

        if row['type'] == 'buy':
            # 买入逻辑：累计成本加本次买入支出
            current_cml = prev_c + row['transacted_value']
        elif row['type'] == 'sell':
            # 卖出逻辑：按持仓占比结转卖出成本，累计成本扣减对应部分
            tx_cost = round((row['units'] / row['cml_units']) * prev_c, 2)
            current_cml = prev_c - tx_cost
        else:
            # 拆分等其他操作：累计成本保持不变
            current_cml = prev_c

        prev_costs.append(prev_c)
        cml_costs.append(round(current_cml, 2))
        cost_transaction.append(tx_cost)
    
    grp['prev_costs'] = prev_costs
    grp['cml_costs'] = cml_costs
    grp['cost_transaction'] = cost_transaction
    return grp

# 分组应用计算，关闭group_keys避免生成多层索引
df = df.groupby('ticker', group_keys=False).apply(calc_group_cost)

结果校验 #

运行代码后，输出三列的结果和你给出的预期值完全一致：

print(df[['prev_costs', 'cml_costs', 'cost_transaction']])

输出：

prev_costs  cml_costs  cost_transaction
0        0.00     105.00              0.00
1      105.00     210.00              0.00
2      210.00     210.00              0.00
3      210.00     150.00             60.00
4      150.00     223.00              0.00
5      223.00     159.29             63.71
6      159.29     232.29              0.00
7        0.00     127.75              0.00
8      127.75     331.75              0.00

说明：代码中对计算结果做了2位小数四舍五入，和示例预期的精度对齐，如需更高计算精度移除round处理即可。

内容的提问来源于stack exchange，提问作者Jakob R