Today a trend broke in Formula One. Max Verstappen didn’t win a Grand Prix. Of the last 35 Formula One Grand Prix, Max Verstappen has won all but 5. Last season he had something like 86% dominance.
For context I believe that I am overall pessimistic when asked to give a probability range about something “working out”. And since sports tend to vary in results if using a sport like Formula One would be a good source of data to make and compare predictions against?
Everything from estimating the range a pole position time, or the difference between pole and the last qualifier, from to a fastest lap in a race or what lap a driver will pit for fresh tyres.
What is the best way of doing it?
Every 2 weeks when there is a race at the start of a weekend making a series of estimates using only armchair-general knowledge of the sport, and then test and compare them with the actual numbers by the end of the racing weekend?
Should I actually attempt to build simple models and heuristics and deliberate and give specific reasoning?
Considering F1 stats go back to 1950 should I make guesses about historical races and championships rather than do it in real time?
The last idea raises the point that the intention is not to get better at predicting Formula One racing, but to reduce pessimistic bias from my own predictions on a wide variety of topics. Which gives me doubts about this exercise.
I wonder if I am wrong to think that making real time predictions would be better. Modern Formula One has become highly predictably after almost 15 years of Red Bull, Mercedes, and Red Bull dominance. It still is a sport and therefore no prediction model will ever be perfectly accurate. I think this hits the right balance of being able to predict within a range, while not being totally perfectly predictable and thus a good source of data to use. However, I don’t intend for it to be the only one.
I could be way-way off and invite thoughts and experiences from others who have tried to get better at calibrating. And suggestions of how they chose the data sets they compared against?
Tracking your predictions and improving your calibration over time is good. So is practicing making outside-view estimates based on related numerical data. But I think diversity is good.
If you start going back through historical F1 data as prediction exercises, I expect the main thing that will happen is you’ll learn a lot about the history of F1. Secondarily, you’ll get better at avoiding your own biases, but in a way that’s concentrated on your biases relevant to F1 predictions.
If you already want to learn more about the history of F1, then go for it, it’s not hurting anyone :) Estimating more diverse things will probably better prepare you for making future non-F1 estimates, but if you’re going to pay attention to F1 anyhow it might be a fun thing to track.
Sounds like it is not a good idea for me then. I feel I already know a lot about the history of Formula One and while I am by no means an expert and there is no doubt more opportunity to learn, it sounds like these bias-avoiding skills won’t be very transferable into real life. I was wondering if the unique mix of high density of statistics as well as my interest in the subject would be a good launching off point but it sounds like you believe it’s non-transferable- correct?
Thank you for the response!