Weighted Season Scores by Year

by Sean Wrona

As discussed in my previous column a few weeks ago, my Racermetrics project for most of this summer has been inventing a single season summary statistic for measuring performance with winning percentage, consistency (based on my percent beat statistic), and dominance (based on my average percent led statistic) all given equal weights to determine each driver's ability at a moment in time. Drivers' season scores are weighted four times for the current season, two times for the season before and after, and one time for the seasons two years before and two years after; for drivers who did not have a complete five year period I used the same weights but with the missing years removed. This provides a snapshot of drivers' performance and career relevance for each given year. In this column I provide the top five drivers for Formula One, NASCAR, and modern-era IndyCar racing for every season up to and including this one (completely up to date through last weekend's races). One can view this as which drivers contributed the most to the field in a given year, and indeed, in one of my future upcoming columns, I will use the sum of driver season scores across given races to determine which races had the deepest and shallowest fields in history.

I fully acknowledge there are many biases in these data. Since they are purely stat-driven, one can similarly argue this is essentially a measure of which drivers had the best teams at a given time since there is no correction for equipment strength. There have been certain drivers who I have considered in the past and consider today very overrated who appear quite highly in these lists solely as a result of having the fastest cars. Similarly, my decision to calculate season scores for each driver for each year regardless of the number of starts did create some fluky situations with drivers with high performance in very few starts being overrated by these data, but for the most part, taking weighted averages corrects for that problem. However, there are still a few unusual cases that clearly aren't realistic snapshots of a driver's performance in time, such as Dale Earnhardt appearing in the top five for 2001 as a result of a good 2000 followed by a single Daytona 500 where he led a pretty good percentage of the race and got a good finish in spite of his death. That in addition to a season with unusual parity placed him in the top five when one might call that a stretch. Similarly, James Hinchcliffe won the flukish rain-shortened IndyCar race at Barber in one of his only five starts before his season-ending injury, which caused him to have an absurdly unrealistic high score for this season placing him in this year's top five; however, he may still fall out when 2016 and 2017 data are also considered. Paul Goldsmith, a part-time NASCAR driver from the pioneering years of NASCAR, ranked in the top two three consecutive years from 1956-58 despite having only nine and two starts in 1956 and 1958 respectively. However, since he won all three years and won six of his 36 starts in the period, I think there is some justification for his high placement on the lists. The reason I didn't remove drivers with too few starts was simple: I wanted to include drivers with few starts in the early F1 and NASCAR seasons when many of the top drivers didn't run every race. I thought it was interesting to have NASCAR lists where Dan Gurney, A.J. Foyt, and Parnelli Jones featured highly and these drivers didn't run many NASCAR races a year. I realize this created some other messes (particularly the Earnhardt and Hinchcliffe situations) but I decided I'd rather include them than set an artificial threshold for the minimum number of starts. This also doesn't include a lot of other factors that are important: versatility (being competitive on many different types of circuits), being personally responsible for the improvement of a team (by bringing great setup knowledge to a team or so on), how races were won, how each driver obtained the lead on track, and so on. However, mainly the issue is that it does not take equipment strength into account at all.

Regardless, there is much validity to this when you simply recognize it as a statistical snapshot of each driver in each moment of time (year). Drivers who are too inconsistent won't feature highly (Junior Johnson never made any of the top five NASCAR lists; of course, the early '60s were STACKED with talent, so it makes sense that the other drivers who were just as dominant and a lot more consistent made the list and he didn't), while other drivers who are consistent over a long period of time but never really dominated won't feature highly either (Ricky Rudd never appeared in the top five for any NASCAR season, Terry Labonte only did so once, and Bill Rexford, Alan Kulwicki, and Bobby Labonte despite their championships did not appear as top five drivers on the list ever, although if you make an equipment correction for Kulwicki, there's certainly a case to be made). There have been plenty of champions who were never regarded as the best driver at the time, and that is what these results make abundantly clear. If you ask the drivers which driver they most feared in a given season, I would assume that the #1 drivers for each season for each series on this list would come closer than the list of champions would. When a single season can have 15-30 races, that's not necessarily the long run. Things do even out in the end, but a single season isn't nearly a large enough sample to reflect a driver's reputation at a moment in time, but a five-year rolling average will come much closer to doing so. Very few of the weaker champions will ever top these lists.

As everyone would likely expect, Stirling Moss indeed easily appear to be the best non-champion Formula One. Regardless of who actually won the championships those years, he was the driver who was feared and I think most people would agree with that. Alan Jones is a little ignored on greatest drivers lists and this suggests that may be a mistake, especially since he brought Williams into prominence. Since Formula One always has rewarded winning sufficiently, there isn't much to talk about here since basically all the legends hit #1 while any of the less legendary champions did not, but Stirling Moss being the only non-champion to top the list and doing so four years in a row is no big surprise. On the NASCAR list, there are a few outliers such as Johnny Mantz leading in 1950 solely due to his dominant win in the inaugural Southern 500, but I'm okay with that since that race was so much longer than any of the others on that year's schedule as it was the first superspeedway race in NASCAR history. I already mentioned the surprise of Goldsmith leading back to back years, but considering he did win 6 out of 36 races in that period, he really was one of the dominant forces, and in 1958, even though he only started two races, he led one of them at flag-to-flag and it was the Daytona Beach/Road Course race, which was held as one of the most prestigious races of the time (before it was ultimately replaced by the Daytona 500). Later, Fireball Roberts and Fred Lorenzen, who skipped most races outside the superspeedways, also were the dominant drivers in spite of never competing for a championship (I believe anyone who really thinks Mark Martin, who competed for championships, should be considered the best driver to have never won the title over those two is wrong, and Junior Johnson, and apparently Fonty Flock and Paul Goldsmith, whom I overlooked, have strong cases too). The list is most interesting for NASCAR because NASCAR's points system does a particularly horrible job of measuring actual performance, while I think this does an extraordinary job. You can easily see the dynasties in NASCAR history one right after the other (Petty-Yarborough-Waltrip-Earnhardt-Gordon-Johnson) with *very* rare interludes likely indicating that most seasons when they didn't win the title somebody else got lucky. While the chase has made the actual finishing order of the championship a crapshoot, the NASCAR data shows that you can still measure performance out of chaos, and I think it does a great job. IndyCar is a bit more of a mess as some drivers like Al Unser, Jr. had much stronger cars some years than others meaning there were few obvious dynasties that continued on and on, especially because many dominant drivers went straight to Formula 1 (and it looks like F1 made the right choices...even though Andretti and Montoya didn't win the 1992 and 2000 titles, they were still the dominant drivers those years by far). I talked more about NASCAR here this time because I simply think F1 and IndyCar have usually always had superior points system which measure performance so there shouldn't be much surprise here. The champions are going to be the dominant drivers AND the point racers, but I really do like this system since I could theoretically apply it to any series for which I was able to get lap leader data.

I excluded all Indy 500 drivers from 1950-1960 from the Formula One list. In that period, the Indy 500 counted as a World Championship race but it was sanctioned by AAA and later USAC, so it wasn't a true Formula One race. I also required that a driver's weighted season score be based on more than one season so that single-season outliers would not appear on the list. I didn't want any scores to be based on only one season's results. However, I made two exceptions. Alain Prost had a perfect 15 season in 1993 winning the championship in his final season after taking a one-year sabbatical. Since he won the title that season, I thought he needed to be included. Similarly, Juan Pablo Montoya won the 2000 Indy 500 in his only IRL start as a full-time CART driver before moving to Formula One. Since he was an Indy 500 winner and appeared on the CART list for 2000, I felt it would be silly not to include him as well. However, those were the only two exceptions I made to this general rule. Now on to the season top five lists, which appear below.

Formula One

NASCAR Cup Series

Champ Car

IndyCar

Sean Wrona is the Managing Editor of racermetrics.com, the Webmaster of race-database.com, the winner of the 2010 Ultimate Typing Championship at the SXSW Interactive Conference in Austin, and the ratings compiler and statistician for the Mensa Scrabble-by-Mail SIG. 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. You may contact him at sean@racermetrics.com.

---