Race Engineer weighs the pace you'd gain from fresh tires against the extra time a stop for them costs, using this car and track's own recorded history — not a fixed rule, and never invented data.
The formula is simple: expected time gained from fresher tires, minus the extra time the stop takes. It runs that math separately for every real option — all four tires, just the fronts, just the rears, or a single corner — and only recommends the one that actually pays for itself over the laps you have left.
If there isn't yet a validated pace-loss signal for this car and track, Race Engineer says so explicitly — "LOW CONFIDENCE" — and recommends no change, no matter how heavy the wear looks. It never turns a guess into a confident-sounding call.
Pace loss and pit-time cost both come from your own past sessions with this car and track — not a universal rule copied from another car or circuit. Race Engineer also tracks how close it is to having enough samples, so you can see it's still learning rather than guessing.
Once a tire change is worth it, the recommendation stays on screen — above iRacing, without stealing focus — for as long as it's still true, the same always-on-top treatment the fuel overlay uses. It disappears on its own once the recommendation changes, or you can dismiss or pin it yourself.
This tire recommendation is independent of the pit-window alert described on the Pit Strategy page — that window is fuel-and-caution only today. A tire-change recommendation appearing doesn't mean the pit window is open, and vice versa; each is its own signal.
It tracks wear per corner for diagnostic display, but the change recommendation itself is driven by your recorded stint pace degradation for this car/track — not a live wear-to-pace formula, since that hasn't been validated on real data yet.
Because the pit-time cost of changing outweighs the time you'd gain from fresher tires over your remaining laps. Staying out is the actual best option in that case.
Race Engineer has no history to learn from yet, so it stays at LOW CONFIDENCE and recommends no change until you've recorded enough sessions.
Yes — it evaluates all four tires, both fronts, both rears, and each individual corner separately, and recommends whichever one actually pays for itself.