Racermetrics race-database.com


2023 Open Wheel Model Update

by Sean Wrona


I have updated my open wheel teammate model in advance of the 2023 season and will be updating my stock car teammate model next week in advance of the Daytona 500. I realize there have already been three Formula E races held in 2023 with another one scheduled for tomorrow. My model does reflect these races in addition to all the races from the fall of 2022 that were not included previously. I have also added all the expected Formula 1, IndyCar, and Super Formula rookies to the model since they have been announced and are expected to make their major league open wheel debuts at some point in the year 2023. I decided to only list the currently active drivers who are expected to make major league starts in 2023 to save space and make this article more readable, although I will note that Juan Manuel Fangio and Ayrton Senna have been trading the lead in the model back and forth along with the laughably-overrated Mark Taylor. Fangio overtook Taylor and Senna to lead the overall model with a rating of .520, but for this piece, I will only include the active drivers.

Oscar Piastri enters with the highest expectations after winning the 2020 Formula 3 and 2021 Formula 2 championships as a rookie both times, and he is not surprisingly the highest-rated new driver in the model at .161 by virtue of his one teammate comparison with a driver already included in the model, his fellow F1 rookie Logan Sargeant, who he beat 9-5 in Formula 3. However, don't think Sargeant is in any way a scrub. He's a perfectly qualified F1 debutant as he too debuts with an above-average rating indicating that he too is already an above average major league open wheel driver, although I suspect he'll probably drop below average as the season progresses because he is a rookie and because I expect Alex Albon might be underrated because he has gone up through every iteration of my model. Although he lost his head-to-head against Piastri in F3 and also narrowly lost his head-to-head against Liam Lawson in F2, both of them were far enough above zero that Sargeant also is; he did also post a 1-0 record against Dan Ticktum whose rating has skyrocketed throughout the history of my model.

While both of the F1 rookies are already rated as above average, the IndyCar rookies are all starting out far below average. After three consecutive 13th place finishes in Formula 2, Marcus Armstrong was announced to replace Jimmie Johnson on the road courses for Ganassi's IndyCar team. He is the highest-rated of the three rookies with a rating of -.156 to Sting Ray Robb's -.221 and Benjamin Pedersen's -.231. Agustín Canapino has no rating yet because like Scott McLaughlin and Jimmie Johnson before him, he has no links to any teammates in open wheel series who ever made a major league open wheel start, so he's a big question mark. Although I personally think he is the most talented of the four rookies by a large margin, his lack of open wheel experience and driving for the historically slow Juncos Hollinger Racing team will probably do him in this season and I suspect Armstrong will win Rookie of the Year primarily because he's in a Ganassi car even though he's not driving the ovals. While he's certainly a better open wheel driver than Johnson, Armstrong still is far below the usual Ganassi standard, as he is barely higher rated than Ed Jones (-.200) who had a miserable 2018 for the team finishing 13th in points in a year Scott Dixon won the championship, and far behind Max Chilton (-.083.) I'm not expecting much from the first-year drivers, although all four of the noteworthy second-year IndyCar drivers (David Malukas, Kyle Kirkwood, Christian Lundgaard, and Callum Ilott) have been rising rapidly and I do think all of them will have some impressive moments in 2023.

In Formula E, Jake Hughes is the only driver newly added to my model for 2023. Although Sacha Fenestraz is also a rookie, he already competed in Super Formula in previous years and made his FE debut last year. Along with Marcus Armstrong, he is the only driver newly added to the model who has a prolific record and numerous teammate comparisons in minor league series already, which means those two drivers are far less likely to make radical shifts in the model than the other rookies. It's clear that Hughes is a lot better than Armstrong as he debuts with a rating of -.081 and he's already won a pole in his third FE start. Although he's still ahead of his teammate René Rast in points, Rast did finish higher than him in both of the races he finished and will probably beat him substantially. Hughes was largely responsible for a number of radical shifts particularly among the young IndyCar drivers. Because Ilott had an 11-5 minor league record against Hughes, and Hughes had a 9-3 record against Santino Ferrucci, that is the primary reason why Ilott made a massive improvement and Ferrucci fell considerably despite having no actual IndyCar teammate comparisons in the intervening period.

There are three rookies in Super Formula this year. The most prominent is obviously Liam Lawson, who dominated the 2021 DTM championship as a 19-year-old rookie while simultaneously competing in F2 before he was cheated out of the DTM championship due to Kelvin van der Linde intentionally wrecking him followed by a pathetic team orders stunt by Mercedes. In 2022, Lawson dropped out of DTM and focused solely on F2, where he was teammates with Sargeant at Carlin. Although Lawson beat Sargeant by a point to claim third in the F2 championship, won four races to Sargeant's two, and beat him 10-8 in the head-to-head, it is Sargeant who made the jump to F1. You could argue Sargeant was more impressive since he was in his first year in the series while Lawson was a sophomore and also because both of his wins came in feature races, which required him to win the pole, while all four of Lawson's wins came in sprint races where the starting grid was determined by a field inversion. On the other hand, Lawson is over a year younger, is rated higher in my model, and has winning head-to-head records against all five of his teammates, including a staggering 21-3 record against Yuki Tsunoda and a 9-5 record against Marcus Armstrong. Since Tsunoda is still in F1, Sargeant is going to F1, and Armstrong now has a Ganassi IndyCar ride, it seems like Lawson deserved a lot better than Super Formula. He did at least land a ride alongside the two-time defending champion Tomoki Nojiri at the championship team and it wouldn't surprise me if he even won the title as a rookie, although I would still expect Nojiri to three-peat. The other two rookies Kakunoshin Ohta and Raoul Hyman don't look particularly strong. Hyman enters as the lowest-rated active driver in my model after being swept 9-0 by Jake Hughes (his only comparison so far.)

Below I list the ratings for each driver after each of the four iterations of my model (from the end of 2020, the end of 2021, August 2022, and the start of 2023.) I suppose I could have used ratings from the end of 2022 instead of the August results (I did calculate them) but I decided not to because by the time I wrote this article, the 2023 open wheel season had already started and the August 2022 model update was so fundamental because I added the remainder of the drivers from the second half of the 20th century into the model. Additionally, since I already posted the results of that model publicly on my website and did not post the results at the end of 2022, I decided I should be consistent and display the old data that I already posted here previously with the 2023 update including all races since the August 2022 update. Furthermore, I had assumed that Théo Pourchaire would have a full-time major league open wheel ride somewhere for 2023 in my model at the end of 2022 and that turned out to be incorrect so I had to remove him from the model in the intervening period. For all these reasons, I decided to ignore my model at the end of 2022 and only list the August model and the February update.

In addition to listing the ratings through each iteration of my model for each of the active major league open wheel drivers, I also for the first time list projections for the expected percentage of the time that each regular driver will beat their teammate(s) in 2023. Although I listed the Indy 500 only drivers here, I excluded them from the projections. I also obviously excluded Ilott (because Canapino does not yet have a rating and it's hard to say how he'll do against Ilott) as well as Nirei Fukuzumi and Kazuya Oshima because they do not have Super Formula teammates. The model is defined so that each driver's rating theoretically is equivalent to 50% + the probability of beating an average driver in a race that both drivers finish, so if you take the highest-rated driver in the model (Fernando Alonso) and compare him to the most average driver (Ferrucci), Alonso would be expected to beat Ferrucci 92.8% of the time because the difference between Alonso's rating (.430) and Ferrucci's (.002) is .428. I used the same logic to compare all teammates to make projections here in the final column of this table. Since Alonso's rating is .430 and Lance Stroll's is .092, they are separated by .338 in the ratings, which implies that Alonso would be expected to beat him by 83.8% (although it actually turned out to be 83.9% due to rounding error.)

Admittedly, the flaw in that sort of analysis is that my model is designed to reflect what has already happened rather than being predictive. Since my ratings are career ratings, it is clear that they will overrate drivers who are declining from their peak (most notably Alonso, who has declined .083 since my initial model and now is only beating Max Verstappen .4304 to .4295 on a tiebreaker; the first time he finishes behind Stroll in a race, Verstappen will probably overtake him.) Similarly, the model will inevitably underrate rising drivers who have not yet reached their peak performance, most notably Jake Dennis. Dennis had a mediocre minor-league career that is continuing to weigh him down in my model, but he is obviously one of the best drivers in the world right now. After sweeping Oliver Askew in every race both of them finished in Formula E last year, he has now beaten the much-more legendary André Lotterer in all three races in 2023 as well as he and Pascal Wehrlein are clearly demolishing everyone else in the Formula E points battle right now. Yet because of Lotterer's rich history and greatness in the 2010s, he is still projected to beat Dennis in 56.9% of races both drivers finish. That is obviously not going to happen as Dennis continues to improve while Lotterer's career is obviously winding down.

Generally speaking, if a driver has risen through every iteration of my model, that indicates they are probably underrated and better than the model implies. Of the 94 active drivers listed here, 20 have never declined in my model. Besides Dennis, these include Pascal Wehrlein, Lando Norris, George Russell, David Malukas, Carlos Sainz, Jr., Scott McLaughlin, Alex Albon, Lance Stroll, Kyle Kirkwood, Marcus Ericsson, Oliver Rowland, Sacha Fenestraz, Nobuharu Matsushita, Maximilian Günther, Callum Ilott, Sérgio Sette Câmara, Dan Ticktum, Ritomo Miyata, and Yuki Tsunoda. It's a safe bet that nearly all those drivers will outperform their expectations because they have a recent history of doing so. Matsushita literally can't because he is expected to beat Raoul Hyman 100% of the time. Kirkwood might not because he's been wildly inconsistent. Malukas might not because he does have to beat Sting Ray Robb by a huge margin. The most interesting one is that Câmara and Ticktum are teammates right now and they have both made exponential jumps throughout the history of my model so far, but this time one of them will probably gain at the expense of the other. It is possible both might continue to improve if they are very close in their head-to-head though because as other drivers they were teammates with eventually make their first major league starts, they will likely improve. I suspect that is most of the reason Ticktum has made up so much ground so far, much like how Ilott keeps rising without even having a teammate last year.

On the flip side, there are a lot of drivers who have gone down through every iteration of my model, which likely indicates they should probably be contended to continue their downward slide and they will probably fail to meet the model's projections for how often they beat their teammates. These include Alonso, Lewis Hamilton, Scott Dixon, Stoffel Vandoorne, António Felix da Costa, Jean-Éric Vergne, Valtteri Bottas, Tony Kanaan, Nico Hükenberg, Robin Frijns, André Lotterer, Romain Grosjean, Kazuya Oshima (who doesn't have a teammate this year), Takuma Sato, Marco Andretti, Zhou Guanyu, Giuliano Alesi, Stefan Wilson, and Katherine Legge. There are a few instances where both teammates are declining, most notably Vandoorne/Vergne, Bottas/Zhou, and Hükenberg/Magnussen (although Magnussen did gain in one iteration of my model, he plummeted last year after his F1 return.) Don't expect a lot from any of those drivers. The interesting comparisons will be be the pairs of teammates where one teammate has gone up through every iteration of my model while the other teammate has gone down through every iteration of my model. These are: Wehrlein/da Costa, Lotterer/Dennis, Russell/Hamilton, Alonso/Stroll, Kirkwood/Grosjean, Ericsson/Dixon, and Ericsson/Sato. These are likely to be wildly volatile as the teammate relationships that may lead to the wildest swings in my model, particularly for the younger/less experienced drivers. I am most suspicious of Ericsson because I do feel at some point his ridiculous good luck in his IndyCar career is going to run out but with Dixon and Sato declining, Álex Palou likely to try to get out of his contract again, and an overrated rookie driving on the road courses, he may continue to improve by default. Kirkwood also is a big question mark because of how often he crashed last year, but Wehrlein, Dennis, Russell, and Stroll in particular seem poised to make big jumps this year. I'm sure Norris will go up a lot and Albon will probably go up too since they've gone up through every iteration of my model and now have rookie teammates.

Generally speaking, drivers who beat their teammates in shared finishes a greater percentage of the time than the probabilities listed here will go up in future iterations of the model while drivers who fail to do so will go down. However, it's certainly more complicated than that. Not only is there the complicating factor of drivers making sudden jumps due to their minor league teammates making major league debuts as we saw with for example Ilott's rise and Ferrucci's decline even though they didn't even do anything, there are also second-order effects. The oddest was how Pato O'Ward actually went down in 2022 despite being the highest-rated IndyCar driver in my model last year. The main reason for that is that Dennis had swept his ex-IndyCar teammate Askew in Formula E, which ended up having a greater effect on O'Ward's rating than O'Ward's own performance did. In my opinion, that is a fluke, and I also suspect Askew was nowhere near as fast after his concussion in the 2020 Indy 500 as he was before, which means Dennis was probably racing against a weaker driver than O'Ward had been. He was probably also hurt somewhat by Herta's disappointing season considering they were Indy Lights teammates as well. Since O'Ward did go up through every other iteration of the model besides last year (when he still posted the highest rating among IndyCar drivers), it seems highly likely he will continue to gain more ground in the ratings and it's hard to imagine him not beating Rossi more than 59.3% of the time and Rosenqvist more than 58.0% of the time.

Finally, I have to note a correction from the previous model. Ren Sato made a massive jump to become the highest-rated Super Formula driver at the dawn of his second season because I forgot to include his 12-1 defeat of teammate Atsushi Miyake in Japanese F4 in 2019 (he won the championship with a staggering 311 points to Miyake's 147.) They were teammates again in 2022 and while Sato beat him then as well, they were a lot closer. This meant he obviously gained substantial ground when I included these earlier results even though Miyake does not have a ride this year in either Super Formula or Super GT. The addition of Kakunoshin Ohta, another teammate of Sato and Miyake in Japanese F4 who Sato beat 10-2 also helped in making him the highest-rated active Asian driver in the model, far above that other unrelated Sato. (I'd still say Takuma is the better driver for the time being, but probably not for much longer.)

The Model

Driver2020202120222023Projection
Fernando Alonso0.5130.4730.4330.43083.9% vs. Stroll
Max Verstappen0.3900.4260.4230.43080.2% vs. Perez
Lewis Hamilton0.4080.4010.3670.36067.8% vs. Russell
Charles Leclerc0.3170.3240.3000.30068.1% vs. Sainz
Mitch Evans0.2890.2940.2920.27362.1% vs. Bird
Edoardo Mortara0.2680.2470.2510.25579.4% vs. Gunther
Josef Newgarden0.2240.2610.2520.25165.5% vs. McLaughlin, 56.8% vs. Power
Scott Dixon0.2750.2550.2480.24590.1% vs. Armstrong, 82.7% vs. T. Sato, 68.5% vs. Ericsson, 60.0% vs. Palou
Stoffel Vandoorne0.2550.2320.2290.22554.7% vs. Vergne
Colton Herta0.1270.2190.2050.20789.5% vs. DeFrancesco, 62.2% vs. Kirkwood, 60.4% vs. Grosjean
Pascal Wehrlein0.1550.1830.1910.20651.6% vs. da Costa
Lando Norris0.0220.1360.1860.20354.2% vs. Piastri
Pato O'Ward0.1400.2140.1950.19659.3% vs. Rossi, 58.0% vs. Rosenqvist
Ren Sato0.1000.19670.9% vs. Yamamoto
Antonio Felix da Costa0.2530.2520.2010.19048.4% vs. Wehrlein
Lucas di Grassi0.1940.1850.1700.18666.7% vs. Rowland
Will Power0.2160.1820.1840.18358.8% vs. McLaughlin, 43.2% vs. Newgarden
Nyck de Vries0.1530.1710.1710.18393.8% vs. Tsunoda
George Russell0.1040.1310.1560.18232.2% vs. Hamilton
Jean-Eric Vergne0.1960.1800.1800.17845.3% vs. Vandoorne
Valtteri Bottas0.2150.2030.1800.16889.6% vs. Zhou
David Malukas0.1200.16588.5% vs. Robb
Rene Rast-0.0440.1320.1160.16174.3% vs. Hughes
Oscar Piastri0.16145.8% vs. Norris
Sam Bird0.1490.1490.1470.15237.9% vs. Evans
Tony Kanaan0.1770.1600.1520.151
Nico Hulkenberg0.1630.1570.1470.14664.0% vs. Magnussen
Alex Palou-0.0230.2070.1520.14580.1% vs. Armstrong, 72.7% vs. T. Sato, 58.5% vs. Ericsson, 40.0% vs. Dixon
Simon Pagenaud0.1580.1420.1440.14355.8% vs. Castroneves
Sebastien Buemi0.1020.0990.1210.14070.2% vs. Cassidy
Esteban Ocon0.0300.0910.1370.13456.3% vs. Gasly
Sergio Perez0.1340.1270.1290.12819.8% vs. Verstappen
Robin Frijns0.2480.1610.1230.12269.4% vs. Muller
Carlos Sainz, Jr.0.0830.1090.1140.11931.9% vs. Leclerc
Andre Lotterer0.1670.1460.1270.11756.9% vs. Dennis
Felix Rosenqvist0.1100.1170.1130.11651.3% vs. Rossi, 42.0% vs. Rosenqvist
Tomoki Nojiri0.1150.1220.1070.10855.3% vs. Lawson
Romain Grosjean0.1350.1240.1050.10379.1% vs. DeFrancesco, 51.7% vs. Kirkwood, 39.6% vs. Herta
Alexander Rossi0.1100.1100.1010.10248.7% vs. Rosenqvist, 40.7% vs. O'Ward
Rinus VeeKay0.2110.1160.0820.09871.0% vs. Daly, 68.6% vs. Carpenter
Scott McLaughlin0.0050.0900.09641.2% vs. Power, 34.5% vs. Newgarden
Alex Albon0.0590.0750.0810.09357.4% vs. Sargeant
Lance Stroll0.0240.0630.0750.09216.1% vs. Alonso
Graham Rahal0.0690.0760.0900.08875.7% vs. Harvey, 61.4% vs. Lundgaard
Helio Castroneves0.0960.0720.0850.08544.2% vs. Pagenaud
Kyle Kirkwood0.0690.08577.3% vs. DeFrancesco, 48.3% vs. Grosjean, 37.8% vs. Herta
Pierre Gasly0.0750.0880.0810.07143.7% vs. Ocon
Marcus Ericsson0.0330.0540.0590.06171.6% vs. Armstrong, 64.3% vs. T. Sato, 41.5% vs. Palou, 31.5% vs. Dixon
Liam Lawson0.05544.7% vs. Nojiri
Jake Dennis-0.0170.0010.0280.04843.1% vs. Lotterer
Kamui Kobayashi0.0450.0340.0210.02462.4% vs. Kunimoto
Sho Tsuboi-0.0070.0080.0040.02161.0% vs. Sakaguchi
Oliver Rowland-0.0180.0000.0180.01933.3% vs. di Grassi
Logan Sargeant0.01942.6% vs. Albon
Sacha Fenestraz-0.144-0.1160.0040.01256.5% vs. Nato
Kevin Magnussen0.0610.0640.0120.00636.0% vs. Hulkenberg
Nirei Fukuzumi-0.020-0.013-0.0090.004
Santino Ferrucci0.0020.0340.0460.00273.4% vs. Pedersen
Naoki Yamamoto0.0530.011-0.014-0.01429.1% vs. R. Sato
Nobuharu Matsushita-0.035-0.030-0.029-0.024100.0% vs. Hyman
Christian Lundgaard-0.406-0.021-0.02664.2% vs. Harvey, 38.6% vs. Rahal
Kenta Yamashita-0.077-0.025-0.042-0.03595.2% vs. Kotaka
Maximilian Gunther-0.096-0.078-0.066-0.03920.6% vs. Mortara
Norman Nato-0.069-0.041-0.057-0.05243.5% vs. Fenestraz
Nick Cassidy-0.104-0.064-0.060-0.06229.8% vs. Buemi
Kazuya Oshima-0.029-0.036-0.061-0.063
Nico Muller-0.108-0.131-0.078-0.07230.6% vs. Frijns
Jake Hughes-0.08125.7% vs. Rast
Takuma Sato-0.070-0.077-0.080-0.08235.7% vs. Ericsson, 27.3% vs. Palou, 17.3% vs. Dixon
Ed Carpenter-0.067-0.042-0.090-0.08852.4% vs. Daly, 31.4% vs. VeeKay
Callum Ilott-0.150-0.133-0.088
Sena Sakaguchi-0.153-0.038-0.081-0.08939.0% vs. Tsuboi
Marco Andretti-0.075-0.080-0.090-0.090
Sergio Sette Camara-0.218-0.183-0.107-0.09950.1% vs. Ticktum
Yuji Kunimoto-0.080-0.071-0.094-0.10137.6% vs. Kobayashi
Dan Ticktum-0.328-0.289-0.216-0.10549.9% vs. Camara
Kakunoshin Ohta-0.10868.9% vs. Makino
Conor Daly-0.113-0.100-0.116-0.11247.6% vs. Carpenter, 29.0% vs. VeeKay
Ritomo Miyata-0.233-0.167-0.158-0.14061.2% vs. Alesi
Marcus Armstrong-0.15628.4% vs. Ericsson, 19.9% vs. Palou, 9.9% vs. Dixon
Jack Harvey-0.123-0.104-0.166-0.16835.8% vs. Lundgaard, 24.3% vs. Rahal
Ryo Hirakawa-0.153-0.195-0.172-0.16855.9% vs. Sekiguchi
Devlin DeFrancesco-0.195-0.18822.7% vs. Kirkwood, 20.9% vs. Grosjean, 10.5% vs. Herta
Sting Ray Robb-0.22111.5% vs. Malukas
Yuhi Sekiguchi-0.194-0.217-0.231-0.22744.1% vs. Hirakawa
Zhou Guanyu-0.205-0.22710.4% vs. Bottas
Benjamin Pedersen-0.23126.6% vs. Ferrucci
Giuliano Alesi-0.215-0.221-0.25238.8% vs. Miyata
Yuki Tsunoda-0.415-0.329-0.272-0.2556.2% vs. de Vries
Stefan Wilson-0.249-0.252-0.256-0.257
Tadasuke Makino-0.385-0.339-0.277-0.29731.1% vs. Ohta
Katherine Legge-0.385-0.396-0.399-0.402
Kazuto Kotaka-0.500-0.499-0.4874.8% vs. Yamashita
Raoul Hyman-0.5830.0% vs. Matsushita

Next week, I will update my stock car model to include the new NASCAR drivers making their debuts in the Cup Series in 2023. While I already had the two main rookies (Ty Gibbs and Noah Gragson) entered in the last iteration of my model, there have been several drivers whose debuts have been announced since then including Riley Herbst, Travis Pastrana, and Chandler Smith. While Pastrana will almost certainly have no effect on the model, it'll be interesting what effects Herbst and Smith might have because they've been at this for a while and they've been linked with a number of Cup drivers. Hopefully the fact that Harrison Burton blew out Riley Herbst when they were Xfinity Series teammates will boost Burton's rating somewhat and thereby help get Ryan Blaney's rating closer to what he deserves (since my model treats the Wood Brothers drivers as Penske teammates.) See you then.

Sean Wrona is the Managing Editor of racermetrics.com, the Webmaster of race-database.com, and the winner of the 2010 Ultimate Typing Championship at the SXSW Interactive Conference in Austin. He earned a master's in applied statistics from Cornell University in 2008 and previously digitized several seasons of NBA box scores on basketball-reference.com. He released his first book, Nerds Per Minute: A History of Competitive Typing, in 2021. You may contact him at sean.wrona@gmail.com.