Trading Agents CN

WHENWhen implementing data source fallback logic

ACTIONUse the defined fallback chain: MongoDB → Tushare → AKShare → BaoStock for CN stocks; MongoDB → yfinance → Alpha Vantage → Finnhub for US stocks

CONSEQUENCEWithout the fallback chain, data collection fails entirely when the primary source is unavailable, causing the entire analysis pipeline to stall and users to receive no market data

WHENWhen validating received market data

ACTIONCheck data is not empty and does not contain error markers ('❌', '错误', '获取失败') before processing

CONSEQUENCEProcessing empty or error data causes downstream analysts to receive invalid inputs, producing garbage output and corrupting analysis results

WHENWhen standardizing DataFrame column names from different providers

ACTIONMap each provider outputs to standard columns: open, high, low, close, vol, amount, date

CONSEQUENCEDifferent provider column naming (Open vs open, Volume vs vol, trade_date vs date) causes technical indicator calculations to fail silently or produce wrong results