Predictions vs Reality
| Driver | Predicted | Actual | Error |
|---|---|---|---|
| RUS | P1 | P1 | 0 |
| HAM | P2 | P4 | +2 |
| PIA | P3 | DNF | — |
| LEC | P4 | P3 | -1 |
| ANT | P5 | P2 | -3 |
| VER | P6 | P6 | 0 |
| HAD | P7 | DNF | — |
| LAW | P8 | P13 | +5 |
| LIN | P9 | P8 | -1 |
| NOR | P10 | P5 | -5 |
| BEA | P11 | P7 | -4 |
| BOR | P12 | P9 | -3 |
| HUL | P13 | DNF | — |
| GAS | P15 | P10 | -5 |
| OCO | P16 | P11 | -5 |
What We Got Right
Two perfect calls to open the season. Russell predicted P1, finished P1 — the model read Mercedes' practice pace correctly and he converted pole into a dominant win from the front. Verstappen at predicted P6 finished exactly P6, recovering from a qualifying nightmare (started P16) with 12 overtakes and 10 positions gained on race day. The top of the grid was strong overall: Leclerc predicted P4, finished P3; Hamilton predicted P2, came home P4; Antonelli predicted P5, outperformed to P2. That gives us a mean absolute error of just 2.4 positions across the top 10 finishers — well below the 2025 backtest average of 3.2. The constructor pick of Mercedes + Ferrari also delivered, with both teams putting drivers on the podium.
Where We Missed
Four DNFs shook up the predicted order — Piastri (predicted P3), Hadjar (P7), Hulkenberg (P13), and Bottas (P20) all retired, and the model has no way to forecast mechanical failures or race incidents. On the team selection side, our recommended lineup of HAM, LAW, LIN, COL, PER was a mixed bag: Hamilton P4 delivered but Lawson at predicted P8 finishing P13 was the biggest miss among our picks. The model overvalued Racing Bulls' practice pace. Bearman was the story of the midfield — predicted P11 but storming from P14 on the grid to P7 with 10 overtakes, a clear sign his 2025 driver bias correction (finishes better than practice suggests) is carrying into the new regs.
Adjustments & China Preview
Two key adjustments heading to Shanghai. First, the Bearman and Gasly overperformance (both finishing 4-5 spots higher than predicted) tells us the 2025 driver bias corrections are directionally right but need stronger weighting in the early season — we're increasing the bias multiplier from 1.0x to 1.5x until we have three 2026 data points to recalibrate. Second, Verstappen qualifying P20 but racing to P6 highlights that grid penalties create massive overtake-point opportunities the model should flag for DRS Boost targeting. China is a sprint weekend, which means reduced practice time and no FP2/FP3 — the model will lean more heavily on FP1 and qualifying data. Watch for Bearman and Colapinto as budget picks: both showed legitimate race pace that outstripped their practice positions, and at their prices they could be early-season value kings.