backtesting on binance data but routing somewhere else — how do you adjust?

Just finished the setup guide for pulling Binance market data into the app, and I appreciate that it says up front, in bold, that live order routing there isn't a thing yet. That kind of honesty saves people a lot of grief, so thank you. But it leaves me with a gap I want to think through out loud, because it's a risk question and that's my hobby horse. The argument in the guide is basically: record from Binance perps because you get the broadest symbol coverage and the spread is both tight and, more to the point, boringly consistent around the clock. No session, no open, no close, no 4pm weirdness. I get why that's attractive for building and testing a strategy — your fill assumptions don't have to change depending on what hour the signal fires. The problem is I'd be *executing* somewhere else. So my recorded book, my modelled slippage, my whole sense of "what a fill costs me" comes from the venue with the nicest conditions of the bunch, and then the actual order goes to a venue that, per the guide's own comparison, is not as tight and may not be as steady overnight. My working checklist so far: 1. Never treat a Binance-data backtest as a cost estimate. Treat it as a signal-quality test only. Does the edge exist at all, ignoring frictions. 2. Re-record a shorter sample from the venue I'd actually route to, even if the symbol list is thinner, and re-run the same logic against that. 3. Pad the slippage assumption on the execution venue rather than trusting a single sample, and size off the padded number. 4. If the edge only survives at Binance-like spreads, it isn't an edge I can trade. Bin it. 5) Re-check step 2 quarterly, because venue conditions drift. What I don't know is whether step 1 is too strict. Is there a sane way to use the good, dense Binance recording as the main dataset and just apply some kind of haircut for the venue you actually trade — or is that self-deception dressed up as method? Anyone doing the two-recording thing in practice, and does the thinner symbol list on the execution venue end up mattering more than you expected? Also a smaller one: for the paper trading the guide mentions, does the paper fill logic use the recorded Binance book, or something else? Changes how much I'd trust it as a rehearsal.