State of the Championship: What’s going on with Palou’s numbers?

With seven races to go in this IndyCar season, Alex Palou has 349 points and leads Pato O’Ward and Scott Dixon by 28 and 53 points, respectively. At Road America, Palou was able to come away with 51 points after his win and Dixon had a quiet day that saw him move from 13th to 4th throughout the 55 laps to earn some much needed points in order to keep the championship battle within reach.

After Road America, Dixon is still the favorite to win the 2021 championship according to The Single Seater’s model. Dixon currently has a 69% chance to win the title, O’Ward has a 16% chance, and Palou has a 9% chance of winning. That’s right, the current championship points leader has a lower win probability than Dixon or O’Ward. How is that possible? Put simply, Palou’s Elo rating, the main number that drives the championship simulations, is still trying to catch up with his performance this season. Palou started the 2020 season with an Elo rating of 1483 and has increased his Elo rating in every race except one. His current Elo rating of 1635 puts him 11th among full time drivers.

Each time the model simulates a season, it uses every driver’s Elo rating to simulate each race left on the calendar and award points. Palou’s Elo rating is good, but it’s still far away from Dixon’s league-best 1963 and O’Ward’s rating of 1768. Because there is such a large gap between Dixon and Palou’s Elo rating, the model thinks Dixon, even with the current point gap, still has a better than not chance of catching up to Palou and passing him in the points. Palou is performing at a higher level than his current Elo rating would predict, so the model is having a hard time assessing his odds of winning the championship and prefers Dixon and O’Ward to him. The difficult thing is that we don’t know what Palou’s true Elo rating should be. Perhaps the model is right and he is performing higher than he should be due to luck and Dixon really does have about a 70% chance of winning the title. But, maybe Palou’s true Elo rating is higher than it currently is and he will keep outperforming what the model expects until his rating catches up. Elo ratings are never static, but with the direction and magnitude that Palou’s rating has been moving this year, I believe the second explanation has more weight and Palou’s true Elo rating is higher than what he sits at now.

Palou outperforming his Elo rating does not mean he has gotten lucky all season. Expected Points is a stat I track that measures the number of points we would expect a driver to earn in a race based on how they ran throughout the race as measured by their average track position. Drivers who have more expected points than actual points are usually a little unlucky, and drivers who have less expected points than actual points are usually lucky. Palou has a +36 point difference between his actual and expected points right now, meaning he’s been a little lucky, but not much. O’Ward has a +75 point difference and Dixon is closer to Palou with a +25 point difference. Every driver at the top end of the field is bound to have a positive difference to some extent because it does take some luck to win a championship in IndyCar, so looking at the magnitude is more important. We can compare these values to Dixon’s 2020 campaign where he had a +98 point difference at the end of the season, or Simon Pagenaud’s 2019 second-place points finish season where he had a +111 point difference at the end of the year. Palou’s over-performing is nothing by historical standards.

We are now getting to the point where Dixon’s win percentage should continue to fall with each race if he doesn’t make significant inroads into Palou’s lead each race. With seven races left, time is not on Dixon’s side, or O’Ward’s. It’s hard to predict exactly when the falloff will happen without more runs of the simulation, but there will likely be a continual drop-off for Dixon along with a big jump for Palou if his performances continue — not only will his total points increase and the number of races decrease, but his Elo rating will continue to climb which will make the simulations more accurate to what is currently going on in the series.

One interesting stat that I believe tells a lot about Palou’s success this season is his Running Percentage, or the percentage of races a driver finishes. In 2020, Palou had a running percentage of 78% when he finished 16th in the championship compared to a running percentage of 100% for Dixon last season. This season, both drivers have a perfect running percentage through nine races. I’ve previously researched whether or not DNFs at the season and race level are correlated race to race and season to season, and found there is a weak to moderate positive correlation with year to year DNF percentage. This means that a driver’s ability to finish races is not completely random from season to season. Some drivers are better at staying out of trouble and finishing races. Despite a lot of DNFs last season, Palou has done a great job of finishing races this season and reversing his negative 2020 performance.

If Palou had finished more races last season, his 2021 performance might have been less surprising. It’s not often you see someone go from 16th in the points one season to seriously competing for the title the next, but going from finishing 78% of races in the first season to 100% of races in the next makes a huge difference in total points over the course of the year.

This article has mostly talked about Palou and Dixon because I thought a deeper explanation into Palou’s numbers was warranted, but O’Ward, and to a smaller extent Newgarden, are very much still in the championship fight too. I’ll look to highlight them more in the next article as the championship battle continues to take shape.