How do you verify your fills are actually good?

Signal-time quote, because that's the price my decision was based on. Order-time hides my own latency, which is exactly the thing I want to see.

And whichever you pick, use the same reference in the backtest. Otherwise the comparison is meaningless.

Dumb question maybe. Do you log the mid or the side you're crossing? I started with mid and my numbers looked way better than they felt, then realized I was ignoring half the spread every single trade. Now I store bid, ask AND mid at signal time and just decide later, storage is cheap.

On a 4 day hold a few cents either way is rounding error, so I mostly just check that nothing insane happened. You lot measuring your own latency in the same breath as your entry price sound like you need a nap :)

Two things I'd add to whatever definition you land on: keep the distribution, not just the average, and pay attention to the tail. The mean fill can look fine while the worst 2% are where your account actually bleeds. And check whether your bad fills cluster on your biggest orders. If size is what's moving you, that's not a slippage problem, that's a sizing problem wearing a costume.