Same Strategy, Two Results: Eleven Causes of Divergence between Backtest and Live Trading

A backtest and a live trading bot can run the same strategy code and still trade differently. This note documents eleven such divergences found during 2026 in a multi-strategy gold trading bot on MetaTrader 5 and its event-driven backtest engine. They fall into four kinds: features that existed on only one side, the same feature implemented differently on each side, plumbing that exists only in live trading, and a backtest market model that did not match the broker. Several were large enough to reverse a strategy's result: a filter that improved the backtest did not run live at all, a break-even rule closed a trend trade at $45 instead of its $150 target, and a recovery routine opened a position nobody had asked for. The note closes with the trade-by-trade reconciliation that found most of them, and with a week in which live and backtest agreed to within 0.1 R. AI disclosure. The author used Claude (Anthropic), an AI assistant, to write the data-analysis code, todraft and edit the text, to assist with wording and structure, and as a consultative tool for discussingmethodological and technical questions. The research questions and the decisions on what to test and whatto accept are the author’s. The author reviewed, verified, and edited all AI-assisted content and takes fullresponsibility for the content of the paper, including its analysis, results, and conclusions.