Monaco GP 2018
Monaco, the most exciting race of the year, is approaching and the drivers want to showcase their talent in a so-called driver’s circuit. Following the Spanish GP where Mercedes dominated, this low-speed, low-grip circuit is expected to suit the Red Bulls and the Ferraris more than the current world champions.
Ricciardo will want to take revenge for what happened in 2016 and this year he has the chance to prove he is ready to get a drive for one of the top two teams next year. Although the model thinks that the circuit will lead to a not-so-unusual Mercedes one-two in qualifying, I’d personally bet on Ricciardo for both pole and race win this weekend. Let’s also keep an eye on Raikkonen who may surprise us!
Grosjean got a 3 place grid penalty from the Spanish GP.
| Driver | Qualifying Prediction | Actual Qualifying Result | Race Prediction | Actual Race Result |
|---|---|---|---|---|
| Ricciardo | 6 | 1 | 1 | 1 |
| Vettel | 3 | 2 | 2 | 2 |
| Hamilton | 1 | 3 | 3 | 3 |
| Raikkonen | 4 | 4 | 4 | 4 |
| Bottas | 2 | 5 | 5 | 5 |
| Ocon | 12 | 6 | 9 | 6 |
| Gasly | 14 | 10 | 14 | 7 |
| Hulkenberg | 8 | 11 | 11 | 8 |
| Verstappen | 5 | 20 | 6 | 9 |
| Sainz | 7 | 8 | 8 | 10 |
| Ericsson | 20 | 17 | 19 | 11 |
| Perez | 10 | 9 | 10 | 12 |
| Magnussen | 9 | 19 | 13 | 13 |
| Vandoorne | 13 | 12 | 12 | 14 |
| Grosjean | 16 | 15 | 16 | 15 |
| Sirotkin | 17 | 13 | 15 | 16 |
| Stroll | 15 | 18 | 18 | 17 |
| Leclerc | 18 | 14 | 17 | R |
| Hartley | 19 | 16 | 20 | R |
| Alonso | 11 | 7 | 7 | R |
6 thoughts on “Monaco GP 2018”
That’s a bit weird no? Mercedes 1-2, Alonso just P11 and Gasly P15? Hmm..
Let’s see what happens.
Hamilton and Bottas 1-2 surprises me.
We all know that Monaco is quite different. But, does the model knows? What kind of parameters are incorporated that could make the model learn differences between circuits?
And why is Verstappen gonna Finish behind Ricciardo? Is this being averaged with all the (high) chances that Verstappen is gonna crash? I’d say that if Verstappen finishes he is likely to be ahead of Ricciardo (as history shows).
The 1-2 of Mercedes surprised me as well. The model does know about the circuit. It knows the circuit name as well each driver’s result in that circuit last year.
Regarding your 2nd question, the model predicts based on the assumption that no-one will retire. However, Max’s retirements do slightly affect his chances (e.g. because he is lower in the driver’s standings). Anyway, the probability of Ricciardo vs. Max is 56-44. Nothing major. FYI Daniel was in front in today’s FP1 and FP2.
I am not sure FP1 and FP2 are such great predictors.
——————————————————————
If you look at last years results:
Verstappen vs Ricciardo
Overall, in 2017 Max and Ricciardo had only 7 times a finish together (5 times Max ahead and 2 times Ricciardo ahead). In 2016 it went better with 15 times a finish together (8 times Ricciardo ahead and 7 times Max ahead)
Mercedes
Ferrari was much stronger than Mercedes in Monaco last year.
——————————————————————
I guess that you could take out a lot of noise, when you are not looking at individual circuits but instead use some specific parameters for circuits. E.g. you could use the distance/time spend in certain speed ranges. In that case it might become much more salient to the model to pick up effects that make certain drivers and teams worse or better on slow street circuits like Monaco.
It seems like currently you have mostly general information like driver standings, and the only thing that can differentiate between circuits are the (noisy) results on the same circuit during previous years.
Yeah, these are all valid comments. Thanks!
The thing is that I do not have such detailed information regarding speed ranges etc. So the model has to learn the model-specific behaviour based only on past predictions.
I have some ideas to create new features (e.g. for each driver pair calculate the historical percentage of the 1st finishing ahead of the 2nd, same of each driver pair and circuit and so on). I need to find time to implement and test any such changes.