I might easily argue do you have an under-powered system that's insufficient to detect such differences but I won't
our production runs on aws but even in dev laptops no issue. We not using any wizards. All code is manually written and produces perfect match between live/backtest/replay. Many of ideas we backtested failed to make money even though backtests were amazing. When those ideas were traded or forward tested they failed, but execution was perfect. I am only talking about futures and more Recently crypto