A backtest estimates how a model would have behaved under stated assumptions. Deployment reveals the assumptions we did not model well enough: execution, data, broker constraints, operational failures, and changing market structure.
That makes live trading a new research phase. We compare observed trade frequency, attribution, drawdown, and execution with the historical profile. Divergence becomes a question to investigate, not a reason to rewrite history.
The goal is not to defend the model. It is to learn whether the claimed behaviour continues to exist under the conditions where capital actually experiences it.