ROUND_13 · MONZA
GP de Italia 2026did we call it right?
Before the race we ran this GP 400 times through our engine, starting from the real qualifying grid, and published the odds. The race has now been run: here is the contrast, misses included.
VERDICT · RACE RUN
What we saidand what happened
Antonelli won from the back, and the model had him fifth: it gave him 10.8% starting from the rear
The model did NOT have Kimi Antonelli as the favourite: it gave him a 10.8% chance of winning.
2.64
Avg. position error
across 19 finishers
0.83
Order correlation
1 = exact order
7/8
Front of the race
of the predicted top
7/10
Points scorers
called right
| Actual | Driver | Grid | Predicted | Diff |
|---|---|---|---|---|
| 1 | Kimi AntonelliMercedesCame through from the back after an engine change and passed Russell at Ascari with three laps left, on mediums after stopping under the virtual safety car. First Italian to win at Monza since 1966. | 20 | 4.70 | -3.70 |
| 2 | George RussellMercedesLed most of the race and lost it to his own team-mate at the end. | 2 | 3.70 | -1.70 |
| 3 | Max VerstappenRed Bull | 5 | 4.90 | -1.90 |
| 4 | Lando NorrisMcLaren | 8 | 4.60 | -0.60 |
| 5 | Oscar PiastriMcLaren | 6 | 4.20 | +0.80 |
| 6 | Lewis HamiltonFerrari | 4 | 4.30 | +1.70 |
| 7 | Pierre GaslyAlpineThe man on pole. The model had him averaging P12.4 with a 0% podium chance: he finished seventh, six places better than predicted. | 1 | 12.40 | -5.40 |
| 8 | Arvid LindbladRacing Bulls | 9 | 13.80 | -5.80 |
| 9 | Franco ColapintoAlpine | 7 | 13.70 | -4.70 |
| 10 | Yuki TsunodaRacing Bulls | 15 | 13.70 | -3.70 |
| 11 | Gabriel BortoletoAudi | 10 | 17.00 | -6.00 |
| 12 | Nico HulkenbergAudi | 12 | 16.30 | -4.30 |
| 13 | Carlos SainzWilliams | 13 | 13.20 | -0.20 |
| 14 | Liam LawsonRed Bull | 22 | 9.20 | +4.80 |
| 15 | Oliver BearmanHaas | 11 | 14.90 | +0.10 |
| 16 | Esteban OconHaas | 14 | 14.80 | +1.20 |
| 17 | Alexander AlbonWilliams | 21 | 15.60 | +1.40 |
| 18 | Sergio PerezCadillacIncluding a 5s penalty for failing to follow race director instructions. | 17 | 17.20 | +0.80 |
| 19 | Valtteri BottasCadillac | 16 | 17.70 | +1.30 |
| DNF | Lance StrollAston MartinHydraulics, lap 26. His retirement brought out the virtual safety car that gave Antonelli the tyre he won on. | 19 | 18.20 | — |
| DNF | Fernando AlonsoAston MartinTechnical issue, lap 23. He started last: he was penalised for irregularities in his power unit AFTER we published the preview, which is why the grid above has him 18th. That change moved the four drivers behind him up one, so Antonelli started 19th rather than 20th. | 18 | 13.20 | — |
| DNF | Charles LeclercFerrariCrashed on lap 1. He started third and was the model’s third favourite. | 3 | 4.20 | — |
Where the model got it wrong
- Gabriel Bortoleto predicted around P17.00, finished 11th (6.00 places off).
- Arvid Lindblad predicted around P13.80, finished 8th (5.80 places off).
- Pierre Gasly predicted around P12.40, finished 7th (5.40 places off).
- Liam Lawson predicted around P9.20, finished 14th (4.80 places off).
- Did not see Arvid Lindblad, Franco Colapinto, Yuki Tsunoda scoring.
The pattern is clear: the model orders the front of the race well and underrates the midfield, because each car's performance is set per season and does not capture a given weekend's form. That is the biggest known limitation, and the next one to attack.
Official result: Motorsport.com. The probabilities were published BEFORE the race and have not been touched since.
The preview, exactly as published
Favourite
George Russell 22.8%
Wins 22.8% of the simulations and finishes on the podium in 58%. Starts P2.
Driver by driver
| Grid | Driver | Win | Podium | Avg. finish |
|---|---|---|---|---|
| P1 | Pierre Gasly | 0% | 0% | P12.4 |
| P2 | George Russell | 22.8% | 58% | P3.7 |
| P3 | Charles Leclerc | 15.8% | 43.8% | P4.2 |
| P4 | Lewis Hamilton | 13.5% | 42.3% | P4.3 |
| P5 | Max Verstappen | 10.5% | 32.5% | P4.9 |
| P6 | Oscar Piastri | 16.3% | 49% | P4.2 |
| P7 | Franco Colapinto | 0% | 0% | P13.7 |
| P8 | Lando Norris | 10.5% | 38.8% | P4.6 |
| P9 | Arvid Lindblad | 0% | 0% | P13.8 |
| P10 | Gabriel Bortoleto | 0% | 0% | P17 |
| P11 | Oliver Bearman | 0% | 0% | P14.9 |
| P12 | Nico Hulkenberg | 0% | 0% | P16.3 |
| P13 | Carlos Sainz | 0% | 0% | P13.2 |
| P14 | Esteban Ocon | 0% | 0% | P14.8 |
| P15 | Yuki Tsunoda | 0% | 0% | P13.7 |
| P16 | Valtteri Bottas | 0% | 0% | P17.7 |
| P17 | Sergio Perez | 0% | 0% | P17.2 |
| P18 | Fernando Alonso | 0% | 0% | P13.2 |
| P19 | Lance Stroll | 0% | 0% | P18.2 |
| P20 | Kimi Antonelli | 10.8% | 35.3% | P4.7 |
| P21 | Alexander Albon | 0% | 0% | P15.6 |
| P22 | Liam Lawson | 0% | 0.5% | P9.2 |
Opinion · our reading
Gasly’s first pole is worth 0% in our model, and the reason has more to do with Monza than with him
Pierre Gasly takes the first pole of his career, at Monza, in an Alpine, and our simulation gives him 0% wins across 400 races. Worse than that: it puts his average finish at P12.4, meaning he loses eleven places from the front row on average. It is hard to publish a number like that without assuming the test bench has broken.
The explanation has two halves and neither is about Gasly. The first is the circuit: the engine works out what starting up front is worth from how easy overtaking is, and Monza is one of the easiest places on the calendar. At the Hungaroring or Monaco pole decides the race; here we have measured it and it is worth two places. The second is the car. In our database the 2026 Alpine is the fifth car on the grid, and the gap to the top three is not a shade. Two places of advantage do not cover that, which is why the model’s favourite is George Russell, starting second and winning 22.8% of the time.
The proof that this is about the track and not the driver sits twenty slots further back. Kimi Antonelli starts twentieth after taking a new engine, and the model ranks him fifth for win probability (10.8%), with a 35.3% podium chance and an average finish of P4.7. Same reasoning, reversed: Monza lets you overtake, and he has the best car on the grid. Where you start matters little in both directions.
And this is where we may be wrong, and we have measured it. Our engine turns pole into a win between 17% and 26% of the time, while in real Formula 1 since 1988 pole converts 50% of the time. So the model undervalues starting first, and it undervalues it at precisely the circuit where it gives it the least weight. Add the slipstream, which is a real factor here and which the engine does not model either, and that 0% becomes the most fragile number in the whole table.
My personal call, opinion rather than data: Gasly does not win, but he does not finish twelfth either, and the points are where he realistically belongs. The interesting thing on Sunday is not his grid slot, it is how long Antonelli takes to appear in the top ten. If he gets there early, the model was right about Monza; if he gets stuck behind a train of midfield cars, the race was right instead.
How the model works
Track position is weighted by how hard it is to overtake at each circuit, so pole is worth a lot at the Hungaroring and very little at Monza. And even so, the favourite is not the man on pole: Pierre Gasly stays at 0% and is edged out by George Russell on 22.8%, starting 2th. Not even at the circuit where the grid matters most does one starting slot cover the gap between the cars.
A favourite who only wins one race in four is not a broken model: that is what a real Grand Prix looks like. A retirement, a safety car at the wrong moment or a slow stop flips the result, and the engine simulates all of them. That is why 7 drivers have a genuine shot rather than a single name circled in red.
The most interesting column is the average finish: Kimi Antonelli starts P20 and averages P4.7, the biggest gain on the grid: the driver the model thinks is most underrated by where he qualified.
The numbers come from 400 full race simulations (53 laps) with car and driver ratings updated to this point of the season. This is not betting advice and not an official prediction: it is what our model said. The odds were published before the race and have not been touched since: the contrast with the real result is above.