Competitive Depth During the IndyCar Split

by Sean Wrona

While I was compiling statistics on competitive depth for both Formula One and NASCAR, I also did so for IndyCar. However, since these data were directly based on three raw statistics (each driver's winning percentage, average percent led, and percent beat), it was readily apparent that treating all races as equal (as IndyCar itself has done) regardless of the actual competitive depth of the field would overrate drivers during the IndyCar split period - particularly the early IRL period. It doesn't take a rocket scientist to observe that in the early years of the split the vast majority of the major talents were in CART while the IRL had few major talents. This was backed up numerous times when drivers crossed over. Davy Jones and Eliseo Salazar qualified on the front row of the 1996 Indy 500 but were two of the first drivers lapped in the Michigan 500 CART race that same year in the same cars. Greg Moore won a dominant Indy Lights championship in 1995 winning ten of twelve races while Buzz Calkins finished in 6th place that season and scored a mere two podiums. However, Moore went winless and didn't even win Rookie of the Year when he went to CART while Calkins won the first IRL race and championship. Shigeaki Hattori was the only driver ever to have his CART license revoked but when he switched to the IRL he became a consistent mid-packer, not a backmarker. Al Unser, Jr. became totally washed up in his late CART years and instantly started winning again when he switched to the IRL. That is just a small sampling of many arguments that can be made, and it even flat out ignores the 2000-01 domination of CART teams at the Indy 500. This doesn't mean that every CART driver was good or that no IRL drivers were great, but clearly the fields did not have equivalent competitive depth and any serious analysis of competitive depth will have to account for this somehow.

Below is the ranking of CART/Champ Car seasons in terms of competitive depth. I find this ranking very credible and not in any need of adjustment since it seems to match the reality very well. As usual, the first seasons (1979-80) and the last seasons (2006-2007) may be a little lower than they should be because these competitive depth statistics are adjusted by the standard deviation of the field, and since the statistics for the early and late seasons are based on fewer data points, the standard deviations are likely to be higher than they otherwise would be and the competitive field depth statistics lower. However, I still think this largely matches the reality. 1979 was the year of another split (the CART/USAC split) where most of the elite talents chose CART while A.J. Foyt was the only elite talent who continued to race in USAC, not to mention that most of the fields had 16-18 car fields just like the later Champ Car seasons did, although to be fair, the top drivers were much bigger stars in that period. It should be no surprise that the depth of CART rapidly increased through the mid-1980s with the entry of the former Can-Am team owners (Paul Newman, Gene Haas, John Trueman, Rick Galles, etc...) and the entry of the '80s/'90s stars in the first half of that decade (Unser, Jr., Michael Andretti, Bobby Rahal, Danny Sullivan, Emerson Fittipaldi, and so on). Nor should it be a surprise to most fans that competitive depth took a fairly stiff downturn in the late '80s/early '90s (in this period only the roughly six or so Chevrolet cars were competitive until 1992 when Ford joined the fray but Michael Andretti won eight races, so this largely left only the Newman-Haas Fords and Penske Chevrolets competitive for the next few years). Competitive depth rapidly increased in the years leading up to the split as the washed-up American legends were replaced by fresher international talent as from 1993-2001 an amazing string of rookies entered almost every year (1993: Nigel Mansell, 1994: Jacques Villeneuve, 1995: Gil de Ferran, 1996: Alex Zanardi and Greg Moore, 1997: Dario Franchitti, 1998: Tony Kanaan and Hélio Castroneves, 1999: Juan Pablo Montoya and Cristiano da Matta, 2000: Kenny Bräck, and 2001: Scott Dixon and Bruno Junqueira), and that is even leaving some good mid-pack drivers like Bryan Herta, Mauricio Gugelmin, Mark Blundell, Patrick Carpentier, and Christian Fittipaldi out. Few people want to argue that CART peaked talent-wise during the split, but it clearly did, as despite the split, it continued to have deep sponsorship and massive funding from three or four engine manufacturers (Ford, Honda, Toyota, and in most years Mercedes). In addition to the influx of talent in that period, the loss of many of the weaker drivers and teams to the IRL also improved the average depth of the field. It's no wonder the head-to-head battles between CART and IRL in the 2000 and 2001 Indianapolis 500s were so unbalanced - 2001 was by far the deepest season in CART history, and that is hardly a surprise. Every driver who ran the majority of the races in that season won a race except for Memo Gidley, Tora Takagi, Max Wilson, and Shinji Nakano - a staggering 21 of the top 23 points finishers won a race, very similar to the 1978 Formula One season, and an amazing eleven of those won races in the 20-race CART season. Eleven drivers from that field also won either a CART title, IRL title, and/or an Indy 500, also extremely impressive. The entry of Scott Dixon and the re-entry of Alex Zanardi were enough to compensate for the loss of Juan Pablo Montoya. That field's depth was so incredible that despite Penske defecting to the IRL after that season and the decline in cigarette and high-tech sponsorship, 2002 was STILL deeper than any season before 1995, because in most of the pre-split CART seasons, only a maximum of three teams threatened for multiple wins in a season. However, after most of the talented drivers left also left the series after 2002 (da Matta to F1, Fittipaldi to NASCAR, and Andretti, Bräck, Dixon, Franchitti, Kanaan to IRL), that was too much to not be felt, and the remaining Champ Car seasons were indeed among the shallowest, as is shown here. I think this matches up well to the reality. Because of the intense manufacturer competition (which admittedly eventually helped lead to CART's demise) and the fact that each manufacturer had multiple competitive teams, many more drivers had the resources to win during the split than before it, and I actually think CART was more enjoyable after the split than before it, even if the overall peak in driving talent was likely shortly before it in 1994 when there four past/future Formula One champions competing (despite one team winning nearly all the races). This largely matched my earlier opinions (I already thought the early split, especially 2001, was the peak of competitive depth in CART), but it was nice to see further evidence for this opinion. Since this likely almost exactly matches what I previously thought, I don't think the split years need to be adjusted for CART.

Champ Car seasons by depth

Despite the staggering amount of attrition in Champ Car during the split, the attrition was clearly felt in the competitive depth ratings. However, the IRL has a different issue because ratings are on the same scale despite the fact that winning, consistency, and dominance were all easier to do in the IRL than in CART, which leads the same drivers to have much higher season scores in IRL in 1996 than in CART in 1995. Arie Luyendyk had a score of 13 in the IRL in 1996, as did Buddy Lazier. In 1995, Luyendyk scored a 1 while Lazier scored a 0. Most other drivers' improvements were not that extreme but still considerable. I ultimately decided the correct way to compare competitive depth would be to compile a list of all the drivers who started half or more of the races in each IRL season who had ten or more CART starts between three years before that IRL season and three years afterward. For instance, for IRL drivers' 1996 seasons, I concocted multi-year career scores for those drivers' achievements in CART from 1993-1999, and so on for the remaining years, obtaining the chart below. I made one change, counting the 1996 season as all races within the calendar year 1996, which included the first two races of the 1996-97 season. I did this because it allows the 1996 IRL season scores to be based on more data and because the IRL teams in 1996 (including the first two races of the 1996-97 season) used CART equipment which means dividing by calendar years instead of seasons makes sense since it allows comparisons between ALL the IRL races those drivers drove in CART-spec equipment with their CART results, rather than having their races in CART equipment split between two seasons.

As indicated from the above table, comparing drivers' IRL season scores to their average CART season scores over a seven year period centered around the current IRL season, IRL season scores added up to 167, while CART scores added up to 41, implying that CART was 4.07 times stronger than IRL for the 1996-2001 period (I grouped all these seasons together because I did not believe there to be any substantive difference in IRL field strength in this period that was not already covered by the differences in IRL season scores). However, if you look more carefully you will notice that over half the CART season score points (24) were scored by one driver, Kenny Bräck, whose CART and IRL profiles were rather similar (to the point that he scored the same number of combined season score points in these comparisons), which clearly appears to skew these data and make the fields look closer than they were, since pretty much all other drivers except Bräck had significantly higher performance in the IRL than CART. While it shouldn't be ignored that Bräck contended for championships in both series and didn't really do any worse in CART than he had in IRL, it also shouldn't be ignored that almost every other early IRL driver was unable to replicate their IRL results in CART. Therefore, I also calculated the ratio of season scores excluding Bräck, reducing 24 points from both sides, and this creates a much steeper field difference; the ratio of 143 to 17 without Bräck implies that CART was in fact 8.41 times stronger than the IRL without Bräck. However, just as you can't ignore that almost every driver improved in the IRL, you also can't ignore Bräck's presence, so I have decided to split the difference and weight the 1996 CART season as 6.24 times stronger than IRL. Since almost all my previously attempts to do this converged to about a ratio at or near seven, this seems to make more sense than a ratio of 4 (which seemed too biased towards IRL) or a ratio of 8 (which seemed too biased towards CART). Even though I compared all data from common drivers in the 1996-2001 period who made enough starts, the goal was to primarily get an estimate for IRL field strength in 1996. The regular IRL field did not change that significantly from 1996 to 2001 and most major driver entries and departures were felt somewhat in the season scores (there was a somewhat sizable drop after 1998 when Luyendyk and Stewart retired from full-time open wheel competition and another after 1999 when Bräck moved to CART), so since these driver changes were already measured in the data, there was no reason other than that to not treat the 1996-2001 period as static, with one exception. Since I compared each year to the period in CART/Champ Car beginning three years before and ending three years after that year in IRL, each seven-year period in CART did not have static levels of competitive depth, so I adjusted these years slightly to reflect the changes of competitive depth in CART. IRL was given a default weight of .160 for this period, but this was slightly adjusted based on the strength of surrounding CART seasons.

During the second half of the split, there were many more full-time crossovers making the comparisons easier than they were around the turn of the millennium when Bräck and Unser, Jr. were providing almost the only significant data. Since the 2002 CART field was significantly weaker and the IRL field significantly stronger with the Penske team switching to the IRL, and the 2003 changes were even more stark after the Andretti Green and Ganassi teams followed suit, I decided I had to judge all seasons after 2001 individually. The five crossover drivers in 2002 (Hélio Castroneves, Gil de Ferran, Alex Barron, Al Unser, Jr., and Richie Hearn) scored 21 points in CART over the 1999-2005 period and 39 points in IRL in 2002, implying the IRL deserved a .538 weight for that season. In 2003, the seven crossover drivers (Bräck, de Ferran, Castroneves, Tony Kanaan, Scott Dixon, Tora Takagi, and Bryan Herta) scored 40 points in CART and 56 in IRL, implying a ratio of 40/56 or a weight of .714 for that IRL season. 2004 and 2005 were calculated similarly with weights of .654 and .857. The drop from 2003 to 2004 is a result of de Ferran's retirement and Bräck's near-fatal injury that ended his full-time career; as these were two of the most successful CART drivers of the period, the strength of the IRL drivers in the field based on CART success diminishes. For the 2006 and 2007 seasons, there were no longer any eligible drivers, as no post-2002 Champ Car drivers competed in the majority of IRL races in either 2006 or 2007 so I continued to assume 2005 field weight for those seasons since there were no particularly major entries or departures in the IRL in that season, with the exception of Ryan Hunter-Reay's late 2007 entry, but my original decision to count only drivers who competed in half of the IRL races in a season unfortunately excluded him from the analysis. For all seasons after 2007, I assumed full field strength again, although this is certainly arguable. Many would argue in any situation as debilitating as the IndyCar split, any kind of unity would certainly be weaker than what came before the event happened. However, the IRL's full season score in 2008 was comparable to 1989 (probably a slightly below average CART season in depth when only the Chevrolet teams were competitive) and its second merged season in 2009 was lower than any CART or Champ Car season except 1979 and the five Champ Car seasons from 2003-07, which does seem to match the reality. In the years since, the field rapidly improved in strength to be comparable to the early split seasons, and I believe there is a good argument for that. It does seem like the pre-split and post-split seasons match the reality well, but the early IRL seasons did not, and making these adjustments is a good way of estimating field depth, although we'll never know for sure. The main drawback is that the three categories I used to calculate season scores were not entirely linear so perhaps simply dividing them does not have great validity, but almost every time I have attempted to estimate CART vs. IRL field strength for the years 1996-2001, I've gotten a result that CART was about seven times stronger and that does feel right. It's possible my estimate of 6.24 is a little generous towards the IRL, but I think taking an average of the ratio with Bräck (which makes the IRL look better) and the ratio without (which makes CART looks better) is about the most balanced thing I could have done given the situation. Below I provide the unadjusted scores for IRL seasons, followed by the original weights I multiplied to determine the adjusted field strength scores for IRL, the adjusted weights (which were slightly different to reflect the different average field strength of the surrounding CART seasons), and finally the overall field strength statistics.

Below I provide a summary of season scores for both series ordered by year instead of in order of field strength. I believe this largely reflects how the field has changed over time. 1979 is certainly underrated as this season's scores are based on fewer data leading to a larger standard deviation denominator (if/when I enter USAC 1956-78 results onto race-database, which I am interested in doing, that may change the 1979 result considerably), but I think the other data check out. Many would argue that the IRL surpassed CART in competitive depth in 2003, while my data still have the latter CCWS seasons in the lead, albeit only barely. However, I think there is justification for this. In recent decades, road racing specialists have done much better at learning IndyCar oval racing than vice versa. Even drivers who proved much more proficient on the ovals than the road courses (such as Sam Hornish, Jr., Dan Wheldon, Arie Luyendyk, and Tony Kanaan) primarily road raced in their earlier careers. It stands to reason that a series with stronger road racers might be slightly stronger given this precedent, even if Sébastien Bourdais did win four titles in a row while there was more championship parity on the other side. Furthermore, in the IRL, Penske, Ganassi, and Andretti's advantage was arguably even stronger than the advantage Newman-Haas and Forsythe held in Champ Car. Additionally, the IRL introduced only two major rookies who didn't race in Champ Car first after 2002 (Dan Wheldon and Ryan Briscoe) while Champ Car introduced several more (Sébastien Bourdais, Ryan Hunter-Reay, A.J. Allmendinger, Justin Wilson, Will Power, and Simon Pagenaud). These factors indicate that the result I got indicating the 2003-07 Champ Car seasons were stronger may not be as false as it seems. A lot of the data seems to check out, as the amount of CART field strength lost between 2001 and 2002 and between 2002 and 2003 is quite similar to the amount IRL gained in both years, and the 2008 figure also checks out as the merged field was definitely stronger than every split field, but nowhere near twice as strong, as many of the Champ Car drivers and teams were lost and others had their impact minimized as the incoming Champ Car teams took a while to get competitive and many did not survive for long. Regardless, I think this is perhaps about as good as can be done to estimate the talent levels of both series during the split. The 2001 figure is particularly striking as CART was at its strongest (competition-wise, surely not management-wise) and IRL was actually at its weakest, and I think there is a strong case for both of those being the case. The on-track competitive gap in the Indianapolis 500 of that year (which was a more extreme example of CART's dominance than any earlier or later example in the split) where five of the six CART drivers entered and ex-IRL driver Tony Stewart lapped all the full-time IRL drivers in the field should come as little surprise, especially since the the only two IRL drivers who won multiple races after the CART drivers arrived full-time, Sam Hornish, Jr. and Scott Sharp both crashed. I always believed Robin Miller was wrong when he said that this was the most competitive IndyCar field in history. I always thought it was the most competitive field since CART 2001 but generally paling compared to the early split years, and this is backed up by the data.

While I ultimately agreed with IndyCar to equally include all Champ Car and IndyCar wins equally on the official American open wheel win list, that doesn't mean they should be considered equally when measuring competitive depth. In order to adjust IndyCar drivers' win totals for the widely divergent field strengths across different series and eras, I have divided each race's field strength score by the average field strength score from 1979-present (regardless of sanctioning body). That average field strength was 22.585, so drivers who won most of their races in the CART period or the post-split IndyCar field for the most part will all increase in adjusted wins while drivers who peaked in the post-CART Champ Car or IRL split period will drop, and that is what most people would consider the reality. Just as in the case of Formula One and NASCAR, I do not rate all races within a season based on each season score, but rate each race individually. Since the Indy 500 has had a deeper field every season than the average series race, Indy 500 winners do get a bump when adjusting for field strength, which I think most fans would agree with (this doesn't mean that the 1996-2001 Indy 500 winners should get as much credit as the winners of any CART race in this period, which they don't). However, this does not mean that I support double points for the Indy 500 or for any race. The financial and historical rewards should be more than enough. Below I provide the adjusted win list for the entire 1979-2015 period.

Looking at the above results of all the Champ Car and IndyCar winners from this period, I believe the results make sense and that it's easy to understand why all the drivers appear where they do. Mind you, this would hardly be my overall rating of all these drivers. There are drivers who are relatively low on this list whom I have more respect for than drivers who are higher, and a more perfect adjusted win list would also include equipment strength factors, adjustments based on whether races were won naturally or not (if I offered a complete win to drivers who won a race naturally and a half win to drivers who did not, it would adjust some of the goofier rankings like very fluky winners being ranked ahead of a number of drivers that were clearly superior), some measure of versatility, and perhaps other factors, but I do think this does an excellent job of correcting the win list to adjust for field strength. It does match some of my previously held opinions of split-era drivers. I always felt Tony Kanaan and Dan Wheldon were slightly superior to Sam Hornish, Jr. because they were all about equally good oval drivers, but Kanaan was a little better than Wheldon and Wheldon was a little better than Hornish on road courses, not that road courses were any of their specialty, and that is exactly how they are ranked on this list. Even though Hornish and Scott Sharp won several races against very shallow IRL fields, the fact that both of them were the only two drivers to win multiple races after the CART drivers entered full-time places them higher than any of the early IRL drivers who did not race in CART, and I think that is a solid argument. Similarly, even though Juan Pablo Montoya has fewer wins than Kanaan, Wheldon, or Hornish, nobody in their right mind would rate him lower than those three, because he, like Scott Dixon, was a rather rare all-around driver who was equally strong everywhere and didn't really specialize on any type of circuit, while Kanaan, Wheldon, and Hornish were definitely all oval specialists. More to the point, Montoya competed against some of the strongest fields ever in the early split before the CART teams and drivers defected, and right now is competing against some of the strongest fields since. Hornish left IndyCar before it reunited, Kanaan fell out of regular championship contention shortly after it reunited, and Wheldon unfortunately didn't really get a chance to test his mettle against a united field except for his one year at Ganassi in 2008. Montoya contending against stronger fields than those three faced at their peak says a lot for him and his higher ranking I believe is merited despite fewer wins and only four full-time IndyCar seasons. There are a few split-era drivers I believe this ranking overrates, but those drivers would tend to be the ones who did not win their races naturally with as great a frequency and/or had extremely dominant equipment. Such drivers would be just as overrated on an unadjusted win list, so I believe this ranking does what it does rather well. Adding natural win, equipment strength, and versatility factors to this list would make it just about perfect, but I think this is a good start.

The beauty of this sort of analysis is it now allows me to begin making the comparisons fans have been making for years and I may repeat this kind of analysis several times in the future. What is the relative strength of Formula One vs. IndyCar? F1, IndyCar, or NASCAR vs. their feeder series? My personal instinct for instance is that while IndyCar's competitive depth has recovered to close to what it was at its peak (as backed up by these data), Indy Lights hasn't come close to doing so, and particularly that in the last decade or so, the NASCAR Xfinity and especially the Camping World Truck Series have lost considerable depth. While I may not undertake all these sort of analyses in the near future, I think this worked pretty well in comparing the overall strength of CART and IRL and matched the reality pretty closely, even as crude as some of my decisions were to do this. This sort of analysis could work in any series that have many common drivers, although I'd still stay away from comparing different disciplines of racing in so doing. Trying to judge the Tudor United SportsCar Championship based on its IndyCar or NASCAR cameos in the 24 Hours of Daytona may be a stretch. However, there's enough crossover between that series, the World Endurance Championship (at least Le Mans), and the European Le Mans Series that I think I could at least make significant comparisons between those series (at least certain classes within those series) based on only a few years' data (much fewer years worth of data than the seven year periods I used to compare CART and IRL). Although many fans enjoy arguing about which premier series have the most driving talent, there have been surprisingly few previous attempts to quantify it where such quantification is possible. I hope I have contributed positively to this dialogue to begin sorting some of this out.

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.

---