the equities cost study — does any of it carry over to perps?

so i read that research piece on this site where they walked market buys down the recorded ask ladder on US stocks, millions of quotes, and compared the actual quantity weighted fill against the mid. main takeaway for me wasn't even the cost number, it was that a decent chunk of the time a normal-ish order size couldn't be filled inside the whole visible book they recorded. that one made me sit up. i trade perps mostly, so i don't have a stock problem, i have a me problem. my backtests assume mid fill. always have. i tell myself perp books on the big pairs are deep enough that it doesn't matter, and honestly for BTC/ETH size i'm probably not lying to myself too badly. but i also take stuff on smaller alt perps where the book thins out fast, and i've never once measured what walking it actually costs me. i just eyeball the ladder before i click and go "yeah looks fine". questions i can't answer from the article: 1. their method is level-by-level against recorded depth. do any of you actually record book snapshots for the pairs you trade, or do you just reconcile fills vs mid after the fact from your own trade log? the second one is way less work but it only tells you about sizes you already traded, not the size you're thinking of scaling to. 2. the article is stocks in regular hours. perps are 24/7 and the book at a dead hour is a totally different animal to the book during a funding-driven squeeze. a single median cost number seems useless for me — surely it has to be conditional on volatility or spread regime? 3. more of a philosophical one. if i subtract a realistic cost from my backtests and half my strategies stop working, is the correct response to trade fewer things, or to trade the same things at smaller size less often? my gut says the first but my gut is a degen. not looking for a magic number, just want to know how people here actually estimate this instead of guessing like i've been doing.