Spanish GP 2018
Baku gave us one the most thrilling GPs ever and, surely, the more unpredictable race so far in 2018. Lewis and Mercedes finally clinched their fist victory of the season but it was not a clean one. F1 is heading to Spain where all teams and drivers have clocked hundreds of thousands of miles around the Catalunyan circuit due to the winner testing and they know the circuit inside-out. Will Ferrari win for the first time since 2012 when Alonso finished ahead of Kimi and Massa?
The model thinks that Vettel is going to win both the pole and the race with Hamilton coming 2nd in the qualifying. I think that Mercedes will show up very strong (don’t forget their pre-season performance) and they will be the team to beat here. I’m also eagerly waiting to see Alonso at him home circuit after his first win at 6-hours of Spa-Francochamp in World Endurance Championship!
Hartley will not participate in the qualifying session after his crash in FP3. Therefore, he’s predicted to start last.
| Driver | Qualifying Prediction | Actual Qualifying Result | Race Prediction | Actual Race Result |
|---|---|---|---|---|
| Hamilton | 2 | 1 | 1 | 1 |
| Bottas | 4 | 2 | 3 | 2 |
| Verstappen | 6 | 5 | 6 | 3 |
| Vettel | 1 | 3 | 2 | 4 |
| Ricciardo | 5 | 6 | 5 | 5 |
| Magnussen | 10 | 7 | 8 | 6 |
| Sainz | 8 | 9 | 9 | 7 |
| Alonso | 12 | 8 | 7 | 8 |
| Perez | 9 | 15 | 14 | 9 |
| Leclerc | 18 | 14 | 17 | 10 |
| Stroll | 16 | 19 | 20 | 11 |
| Hartley | 20 | 20 | 16 | 12 |
| Ericsson | 19 | 17 | 18 | 13 |
| Sirotkin | 17 | 18 | 19 | 14 |
| Vandoorne | 14 | 11 | 12 | R |
| Ocon | 11 | 13 | 13 | R |
| Raikkonen | 3 | 4 | 4 | R |
| Grosjean | 13 | 10 | 11 | R |
| Gasly | 15 | 12 | 15 | R |
| Hulkenberg | 7 | 16 | 10 | R |
11 thoughts on “Spanish GP 2018”
What is more interesting for me – how the Model decided that Kimi will be ahead of Lewis?
btw. I’ve found a little mess in my datasets and working on it, hope I will be fix it before Spanish GP.
If you see the raw probabilities it’s actually 49.3%-50.7% against Hamilton. So their difference is negligible. Still, it’s interesting why it is 50-50 and not something like 60-40 against Kimi.
If your model thinks that Ricciardo has a qualifying advantage in comparison to Verstappen, then there is something fundamentally wrong.
In 3 out of 4 of the previous races, the model got correct their relative ordering (even when Ricciardo outqualified Max). Let’s see if it has found some hidden pattern in the data. It actually gives Ricciardo a 62% chance against Verstappen.
3 out of 4 correct guesses does not say that much. Whatever the model may have picked up in the pattern that could give Ricciardo an advantage in Spain (is it the same if you use other parameters from other racing tracks?) it might very well be nonsense.
I guess that a model that simply puts in the last race qualifying result does equally well 3/4 when it comes to predicting qualification duels.
It’s not easy to run the model as if we’re for a different circuit so I cannot check that.
I agree with you that this seems a bit strange but I’m not completely surprised by this prediction. We’ll see the result on Saturday.
P.S. I’ve mostly optimized the model for race results and I’ve not created yet too many features regarding the qualifying. That’s the best explanation.
Do you mean that the model does not take into account the ‘type of circuit’ in it’s predictions? Then this makes the model even more unsound. Since, when it is mostly a general overall prediction, then what other parameters could give Ricciardo a qualifying advantage over Verstappen while Verstappen is mostly out-qualifying Ricciardo (without doubt) both in time and in number of duels won?
Of course it does take the circuit into account. I’m just saying that it’s technically a bit difficult for me to check what the predictions would be if it was for another circuit.
Apart from that, other factors that the model may take into account are the drivers’ position in the standings, their form (how well they do in the last few races) etc.
note: I am not saying that Verstappen is so far ahead of Ricciardo during qualifying, but it has become very consistent.
Verstappen finished ahead of Ricciardo by just 0.003s. So, the model didn’t get it too wrong! 🙂