iRacing Lap Time Analyzer

Every other tool in this space is either format plumbing or a trace viewer that diagnoses nothing. iRacing Lap Time Analyzer reads the binary .ibt files sitting in Documents\iRacing\telemetry — dozens of MB each, hundreds of channels, regenerated every time you race — and turns them into a per-corner ledger of where your lap time actually goes. The whole iRacing session header, var header and fixed-stride sample records are parsed in the browser with a DataView; nothing is uploaded and a 40 MB binary is never held as a blob. Corners are segmented from your own signals — smoothed lateral g, apex peaks and speed minima — so the app needs no track database, no car database and no patch awareness, and it never goes stale because the geometry comes out of your file. The delta against your own best lap (or a reference lap) has an exact WHERE: entry, mid, apex and exit distance buckets that telescope to the corner delta by construction, with nothing left over. For the WHY it refuses to manufacture precision. The measured driver actions — minimum speed, brake point, brake release rate, peak brake, trail-brake overlap, apex position, throttle point, throttle ramp, steering corrections — are correlated and causally entangled (releasing the brake early is often WHY minimum speed is higher), so they are ranked by effect size measured in sigma of your own laps and labelled as correlated, never summed into a fake exact split over causes. Attribution is refused while the pre-flight gates fail: speed must integrate to lap distance within 1%, pedals must span 0-1, g-forces must sit within physical bounds, and steering must be coherent with yaw, because a mis-scaled channel produces a confident, precise, wrong answer you cannot detect from the answer. It separates slow from inconsistent (per-corner variance across the session: slow needs technique, inconsistent needs repetition), classifies corners into slow-hairpin / medium / sweeper archetypes and keeps a rolling series