Race Predictions

Pre-session ML predictions locked before the result is known — validated with a transparent running score.

Credibility Scoreboard

2026 Season · Prediction Accuracy

Recent accuracy

82%

Model driftModerate driftscore 0.33

Events scored5this season

Correct3/5this season

Accuracy trend

Last 5 Events

↑+0.190race ρ trend

Event	Quali ρ	Race ρ	Pit hit
ESP	0.918	0.648	100%*
AUT	0.945	0.834	100%*
GBR	0.913	0.272	100%*
BEL	0.887	0.900	90%*
HUN	0.927	0.839	95%*

ρ = Spearman rank correlation · 1.0 perfect · trend = last minus first in window · * SC-adjusted (hover for strict ±2)

Running Score This Season82%composite of 3 metrics

Race Order Accuracy0.66911 events scoredSpearman ρ · predicted vs actual finishing order

Quali Prediction0.86710 events scoredRank correlation · predicted vs actual qualifying order

Pit Window0.92111 events scoredSC-adjusted hit rate (strict ±2: 14%)

Spearman ρ: 1.0 = perfect order · 0 = random · −1 = inverted | All predictions locked pre-session, scored post-race.

Per-Event History

#	Event	Date	Quali ρ	Race ρ	Pit hit
1	AUSAustralian Grand Prix	2026-03-08	0.747	0.576	74%*
2	CHNChinese Grand Prix	2026-03-15	—	0.895	88%*
3	JPNJapanese Grand Prix	2026-03-29	0.929	0.885	100%*
4	MIAMiami Grand Prix	2026-05-03	0.878	0.613	78%*
5	CANCanadian Grand Prix	2026-05-24	0.844	0.426	94%*
6	MONMonaco Grand Prix	2026-06-07	0.678	0.468	94%*
7	ESPBarcelona Grand Prix	2026-06-14	0.918	0.648	100%*
8	AUTAustrian Grand Prix	2026-06-28	0.945	0.834	100%*
9	GBRBritish Grand Prix	2026-07-05	0.913	0.272	100%*
10	BELBelgian Grand Prix	2026-07-19	0.887	0.900	90%*
11	HUNHungarian Grand Prix	2026-07-26	0.927	0.839	95%*

* SC-adjusted — counts pits that fell inside a historically likely safety-car window for that circuit (hover for strict ±2 rate)

Season Highlights

Best Call

Austrian Grand PrixAUT

Quali ρ 0.945qualifying prediction

Biggest Miss

British Grand PrixGBR

Race ρ 0.272race finishing order

Method Note

Race accuracy measures whether the predicted finishing order matches the actual result, using Spearman rank correlation (1.0 = perfect, 0 = random, −1 = inverted). Qualifying accuracy measures rank correlation between predicted and actual qualifying lap time order. Pit window accuracy measures whether the predicted pit lap was within ±2 laps of the actual pit lap. All predictions are locked before the session starts — no retroactive changes.

How we measure accuracy →