Comparing two setups with data instead of feel
Comparing two setups with data instead of feel: the four measurements that settle it, how many laps you actually need and the mistake that invalidates the result.
Two setups feel different long before they are different, because the second one is always driven by a driver who has just done twenty more laps. Removing that effect is most of the work, and it takes a method rather than an opinion.
Why feel cannot decide this
Three things change between the two runs and only one of them is the setup. You have learned the circuit slightly better, the track has rubbered in, and the tyres you fitted are not the tyres you took off. A driver comparing setup A at the start of a session with setup B at the end is measuring all three.
The second problem is expectation. A driver who has just spent ten minutes making a change wants it to work, and the honest way through that is to have the numbers decided before the run rather than interpreted after it.
And feel reports change rather than speed. A setup that is one tenth faster and feels worse is common, because more grip at one end often feels like less stability, and drivers reject it before the stopwatch has spoken.
The four measurements that settle it
The average of laps two to five, not the best lap. The best lap is a single sample and noisy; the average of four consecutive laps at race pace is stable enough to compare, and excluding lap one removes the tyre warm up.
The spread across those four laps. If the spread is larger than the difference between the two setups, you have not measured a difference, you have measured yourself, and the honest response is to repeat rather than to conclude.
Speed at two fixed points: the end of the longest straight and the apex of the fastest corner. Those two numbers separate a change in drag from a change in grip, which a lap time combines into one figure and hides.
And the tyre readings at the end of the run. Hot pressures and the tread spread tell you whether the two setups are even comparable, because a setup that finishes with pressures three psi apart is a different car and not a different setup.
| Measurement | Why |
|---|---|
| Average of laps two to five | stable, excludes tyre warm up |
| Spread across those four laps | tells you whether the difference is real |
| Speed at the end of the longest straight | isolates drag |
| Speed at the apex of the fastest corner | isolates grip |
| Hot pressures and tread spread | confirms the runs are comparable |
| Fuel load and track temperature | both change which setup wins |
The method that removes the noise
Run A, then B, then A again. The third run tells you how much of the difference was the session rather than the setup, and if A2 is closer to B than to A1, the setup change is smaller than your own improvement over the session.
Use the same fuel load for every run, and state it. Fuel is the largest single variable in a comparison, and comparing a low fuel run with a high fuel one produces a result that means nothing.
And rule out the machine before you draw a conclusion. The published meter thresholds are specific: a full bar on the physics thread meter is about 16.7 ms at 60 Hz, latency is one second, packet loss is 50 per cent per direction and page faults are 30 per second. A run with a stutter is a discarded run, not a slow setup.
What to do with the answer
Budget the session honestly. 3 runs of 5 laps is 15 laps plus 3 in and out laps, or about 18 of the 40 laps a session gives you, which leaves room for exactly 1 more comparison. A driver who tries 4 comparisons in 1 session gets 2 laps of data each and 0 usable answers, and that is the arithmetic behind every inconclusive setup evening.
Keep both and label them. A setup that is slower over a single lap but easier over forty is the race setup, and a comparison that produces one winner has usually asked the wrong question. Note the fuel load and the track temperature next to each, because both change which one wins.
And change one thing before the next comparison. A method that reliably distinguishes two setups is wasted on two setups that differ in six ways, because the answer will be that one of them is faster and you still will not know why.
Sources
Frequently asked questions
How many laps per setup?
Five, and discard the first. Four consecutive laps at race pace give a stable average and a usable spread.
Why run the first setup twice?
Because it separates your improvement over the session from the setup difference. If A2 beats A1, part of what you measured was you.
What if the two setups are within a tenth?
Then they are the same setup as far as your driving is concerned. Pick the one that is easier over a full stint.
Can I compare a low fuel run with a high fuel one?
No. Fuel is the largest variable in the comparison and it invalidates the result entirely.
More in Setup & Telemetry
Learning a new track in three structured sessions
Learning a new circuit in three structured sessions instead of ten unstructured ones, with what each session is for and how to know it worked.
The replay session: finding the mistake you cannot feel
The replay session: finding the mistake you cannot feel, which four channels to look at first and how to turn a replay into a change.
Safety and skill ratings: what the numbers reward
How iRacing Safety Rating works: the 0.00 to 4.99 scale, the 3.00 and 4.00 promotion thresholds, the 0.40 bumper at whole numbers and the participation minimum.