Per-broker cost model in the backtest

Right now I apply one cost assumption. But I trade through two brokers with different fee structures, and the same strategy has a different net result depending on where it runs. What I'd like: define a cost profile per broker, commission structure, typical spread by liquidity band — and pick which profile a backtest uses. Then "is this strategy worth running" becomes a question I can answer per venue rather than on average. This matters most for the strategies where it matters most: the thin-edge, high-frequency ones where costs decide the answer.