Race Winner Prediction

7.004 km lap length19 cornersLow overtaking difficulty

Race date July 19, 2026 · generated Sep 8, 2026, 2:09 PM UTC · rain scenario weighted at 14%

Predicted vs. Actual

Predicted winner

ANT (21.3%)

Actual winner

ANT (correct)

Podium hits

1 / 3

Brier score (win)

0.0323

The model gave the actual winner (ANT) a 21.3% chance beforehand. This was a blind backtest — generated using only data available before the race, never this race's own results. See the track record page for the model's accuracy across every graded race.

Our model: 0.0323 · Grid-based baseline (always pick the polesitter): 0.0000 (baseline beats the model this time — lower is better)

Where the podium call missed

RUS (predicted podium) didn't finish — Retired.
HAM (predicted podium) finished P4 instead.
LEC finished P2 — the model projected around P6.
VER finished P3 — the model projected around P8.

Predicted vs. actual finishing position, every driver

Green, right — finished better than the model expected (underrated). Red, left — finished worse (overrated). Sorted by size of the miss.

Win Probability — Top 10

Error bars show ± 1 standard deviation of win probability across batches of the Monte Carlo simulation — how much this estimate would wobble on a re-run, not a formal confidence interval.

Full Prediction Table

Click a row to see what drove that driver's score.

Pos	Driver	Team	Win %	Podium %	Points %	Exp. Pos	Actual
1	▸ANT	Mercedes	21.3%	50.3%	91.9%	4.57	1
2	▸RUS	Mercedes	19%	46.7%	90.6%	4.87	22
3	▸HAM	Ferrari	15.1%	40.9%	88.3%	5.38	4
4	▸LEC	Ferrari	13.3%	37.6%	86.6%	5.68	2
5	▸NOR	McLaren	8.4%	27.9%	81%	6.69	7
6	▸PIA	McLaren	6.4%	22.7%	76.1%	7.4	5
7	▸VER	Red Bull	4.7%	18.4%	71.4%	8.05	3
8	▸HAD	Red Bull	3.2%	13.1%	63.2%	9.1	6
9	▸LAW	RB F1 Team	2.5%	11.1%	59.3%	9.57	12
10	▸GAS	Alpine F1 Team	1.8%	8.3%	52.4%	10.37	11
11	▸LIN	RB F1 Team	1.3%	6.3%	46.1%	11.15	9
12	▸COL	Alpine F1 Team	0.7%	3.7%	34.6%	12.53	10
13	▸BEA	Haas F1 Team	0.7%	4%	36.3%	12.32	14
14	▸BOR	Audi	0.4%	2.4%	26.2%	13.71	8
15	▸SAI	Williams	0.4%	2.1%	25%	13.89	16
16	▸OCO	Haas F1 Team	0.2%	1%	15.3%	15.54	17
17	▸ALB	Williams	0.2%	1.4%	18.9%	14.85	15
18	▸PER	Cadillac F1 Team	0.1%	0.4%	6.4%	17.71	21
19	▸STR	Aston Martin	0.1%	0.4%	6.2%	17.8	20
20	▸ALO	Aston Martin	0.1%	0.8%	12%	16.21	19
21	▸HUL	Audi	0.1%	0.3%	6.1%	17.84	13
22	▸BOT	Cadillac F1 Team	0%	0.3%	6.3%	17.78	18

Methodology

Each driver's simulated race pace is sampled from a distribution weighted by qualifying pace percentile, top-speed ranking, season championship standings, historical finishing position at this circuit, tire degradation, and pit-stop delta. The simulation runs 100,000 times, with 14% of runs treated as a wet-race scenario (estimated from this circuit's own rainfall history).

This circuit's weighting — Belgian Grand Prix: overtaking difficulty Low, downforce dependency Low, tire severity Medium. Higher overtaking difficulty shifts weight toward grid position; lower downforce dependency shifts weight toward straight-line top speed; higher tire severity shifts weight toward tire-degradation management. See circuit_profiles.py for every circuit's numbers and how they translate into weights.