Backtrader中RSI初始化触发ZeroDivisionError除零错误排查

用户运行以下Python Backtrader代码时，在RSI指标初始化阶段触发ZeroDivisionError: float division by zero错误。用户确认数据源正确且包含RSI计算所需起始数据，尝试过调试、添加日志但RSI内部日志无输出，需排查错误成因。

import backtrader as bt
import pandas as pd

# Load data
file_path = r'C:\forex\python\data\EURUSD1min-v2.xlsx'
data = pd.read_excel(file_path)

# Handle date and time in separate columns
data['Datetime'] = pd.to_datetime(data['Datetime'], format='%Y.%m.%d. %H:%M')
data.set_index('Datetime', inplace=True)

print("Data loaded, total entries:", len(data))

# Backtrader data format
class PandasData(bt.feeds.PandasData):
    params = (
        ('datetime', None),
        ('open', '<OPEN>'),
        ('high', '<HIGH>'),
        ('low', '<LOW>'),
        ('close', '<CLOSE>'),
        ('volume', None),
        ('openinterest', None),
    )

# RSI-based strategy
class RSIStrategy(bt.Strategy):
    params = (
        ('rsi_period', 14),
        ('rsi_upper', 70),
        ('rsi_lower', 30),
    )

    def __init__(self):
        self.rsi = bt.indicators.RSI_SMA(self.data.close, period=self.params.rsi_period)
        print("RSI indicator initialized.")

    def next(self):
        if len(self.data.close) < self.params.rsi_period:
            print("Not enough data for RSI calculation.")
            return

        current_rsi = self.rsi[0]
        current_close = self.data.close[0]

        if pd.isna(current_rsi) or current_rsi == 0:
            print("Invalid or zero RSI:", current_rsi)
            return

        print(f"Close: {current_close}, RSI: {current_rsi}")

# Load data into Backtrader
data_feed = PandasData(dataname=data)
print("Data converted to Backtrader format.")

# Create Backtrader cerebro and add strategy
cerebro = bt.Cerebro()
cerebro.adddata(data_feed)
cerebro.addstrategy(RSIStrategy)
print("Strategy added to Cerebro.")

# Run the strategy
cerebro.run()
print("Execution finished.")

错误成因排查 #

• 极端数据情况：Backtrader的RSI_SMA用SMA计算平均涨跌幅度，若初始化的14期内所有收盘价完全一致，涨跌幅度总和为0，就会触发零除错误。
• 列名映射错误：自定义PandasData中使用<OPEN>、<CLOSE>带尖括号的列名，若Excel实际列名是无尖括号的OPEN、CLOSE，会导致Backtrader无法读取有效收盘价数据，进而计算异常。

解决步骤 #

1. 验证数据有效性
   在加载数据后添加检查代码，确认前14期收盘价是否有变动：

   # 检查前14期收盘价变动情况
   initial_closes = data['<CLOSE>'].head(14)
   price_diff = initial_closes.diff().dropna()
   print("前14期价格变动值:", price_diff.values)
   print("是否存在非零变动:", any(price_diff != 0))
   

   若输出无变动，说明数据源存在连续相同收盘价的极端情况，需更换数据或调整策略逻辑。

2. 修正列名映射
   确认Excel实际列名后，调整PandasData的参数（示例为无尖括号列名）：

   class PandasData(bt.feeds.PandasData):
       params = (
           ('datetime', None),
           ('open', 'OPEN'),
           ('high', 'HIGH'),
           ('low', 'LOW'),
           ('close', 'CLOSE'),
           ('volume', None),
           ('openinterest', None),
       )
   

3. 添加容错处理
   在策略初始化时捕获异常，避免程序直接崩溃：

   def __init__(self):
       try:
           self.rsi = bt.indicators.RSI_SMA(self.data.close, period=self.params.rsi_period)
           print("RSI indicator initialized.")
       except ZeroDivisionError:
           print("RSI初始化遇零除错误：检查数据是否存在连续相同收盘价")
           self.rsi = None
   

内容的提问来源于stack exchange，提问作者Puma U