Season Accuracy Scoreboard

2026Credibility →

Season Avg ρ0.787

Rounds Scored18

ROUND-BY-ROUND ACCURACY

Rd	Event	ML ρ	Champ ρ	Circuit ρ	Grid ρ	Δ vs Grid
1	Australian Grand Prix	0.806	—	0.806	0.571	+0.236
2	Chinese Grand Prix	0.875	—	0.875	—	—
3	Japanese Grand Prix	0.364	0.450	0.364	0.572	-0.208
4	Miami Grand Prix	0.975	0.571	0.975	0.631	+0.344
5	Canadian Grand Prix	0.679	0.624	0.679	0.448	+0.231
6	Monaco Grand Prix	0.552	0.691	0.552	0.525	+0.027
7	Barcelona Grand Prix	0.879	0.685	0.896	0.732	+0.146
8	Austrian Grand Prix	0.926	0.503	0.926	0.871	+0.055
9	British Grand Prix	0.642	0.709	0.642	0.589	+0.053
10	Belgian Grand Prix	0.917	0.867	0.917	0.770	+0.148
11	Hungarian Grand Prix	0.849	0.624	0.849	0.778	+0.071
12	Dutch Grand Prix	0.882	0.800	0.882	0.716	+0.167
13	Italian Grand Prix	0.881	0.817	0.881	0.752	+0.129
14	Spanish Grand Prix	0.913	0.855	0.896	0.774	+0.139
15	Azerbaijan Grand Prix	0.586	0.661	0.586	0.508	+0.078
16	Malaysian Grand Prix	0.836	0.758	0.836	0.745	+0.091
17	Singapore Grand Prix	0.719	0.345	0.719	0.890	-0.171
18	United States Grand Prix	0.889	0.806	0.889	0.894	-0.004

Grid ρ = qualifying grid position vs actual finish — naive baseline

ML ρ: Spearman rank correlation between model-predicted and actual race finishing order · Champ ρ: how a model predicting current championship standings order would score (Phase 92 baseline) · Circuit ρ: how a model using historical circuit results would score (circuit prior baseline) · Grid ρ: correlation between qualifying grid position and actual race finish (naive baseline) · Δ = ML ρ − Grid ρ · positive means model outperforms grid order · ML ρ cell: green = beats championship baseline, red = below · methodology

Pipeline Reliabilityaggregated artifact-reliability data across all 2026 events

live/timing/driver_countcritical100.0% fail rate · last: Spanish Grand Prix
0 drivers
live/pace/entry_countcritical100.0% fail rate · last: Spanish Grand Prix
0 entries
live/currentcritical100.0% fail rate · last: Spanish Grand Prix
timed out