On-Track Passing in NASCAR

by Sean Wrona

I realize I have not posted any new articles in over a year and a half, but I have still been undertaking many racing statistics projects. Over the past year, especially over the last half of 2017, I have been going through footage of every archived NASCAR and IndyCar race on YouTube and collecting all the on-track lead changes for each race in an attempt to determine which drivers were the best duelists. Most racing analyses focus entirely too much on the final result and fail to consider performance throughout the race, especially in instances when drivers lose races due to circumstances arguably outside their control (particularly slow pit stops and mechanical failures.) One of the best ways to evaluate drivers that has been little explored to this point is one of the most overlooked: on-track passing, since that nearly entirely depends on drivers and their equipment and removes most team-related factors except for equipment strength (which I have also been beginning to attempt to measure, although I have not come up with anything that I consider publishable in that regard yet.) Many people have made the distinction between being a great driver and a great racer and I did not originally understand the difference myself, but now I do. Some drivers are highly mistake-prone but exceptional at making on-track passes while others are consistently strong at getting results but are rather too conservative in battling on the track. I generally believe drivers who are stronger in racing ability than their results may indicate tend to be undervalued because drivers likely have more impact on their passing ability than their ultimate finish. Passing solely depends on the driver's ability and equipment, while finishing results involve numerous other factors, especially considering many, many lead changes in every racing series happen in the pits. When watching a race - particularly watching an oval race, where many would argue racing the other drivers is more important than racing the track, unlike in road racing, where many would argue the reverse - fans tend to remember and celebrate drivers based on their ability to battle other drivers throughout an entire race rather than merely their finishing results, but this is something that is not always well-reflected in the eventual finishing statistics.

A proper appraisal of a driver's career should consider both factors (driving ability and racing ability, or alternately, racing the track and racing the other drivers), in much the same way that for instance most of Jeff Sagarin's sports ratings systems are based on an average of a team's Elo score (based on their won-loss performance) and their Predictor score (based solely on the level of the team's dominance.) Since auto racing is not a 1-on-1 sport I do feel it works a little differently. While almost every points system NASCAR has ever had was a complete joke that rewarded consistency too much and top finishes not enough, the NASCAR points system results for each given season (at least the pre-Chase seasons) will usually have a strong correlation with consistent performance, even if I think it inevitably overrates drivers who don't take risks. If I were to calculate something similar to Sagarin's Elo score (and I may eventually) it would certainly weight the top finishers in a race more and the bottom finishers less, but that still would only reflect one perspective: that being the attempt to measure the best drivers. It does nothing to consider dueling ability. While most people would measure dominance based solely on laps led (or better, cumulative races led), that's still a little simplistic as the driver who leads the most laps in a race need not be the best driver in a race any more than the driver who actually wins the race. For example, Brian Vickers led the most laps of the 2005 spring Pocono race but never took the lead on track, as he beat polesitter Michael Waltrip out of the pits then got passed by Carl Edwards three separate times after Vickers repeatedly beat Edwards out of the pits. Obviously you can argue Vickers dominated the race solely because he led the most laps, but a smarter perspective holds that Edwards actually was the dominant driver in the race because he managed to pass the driver who led the most laps three times while Vickers passed no one to take the lead. However of course there are many other races I could cite as examples where the winner was not dominant but won largely due to pit strategy or someone else's mechanical failure. Leading in and of itself is not necessarily impressive, nor is winning in and of itself necessarily impressive. Ultimately, to properly evaluate drivers throughout a race, just like you have to consider how the races were won, it is essential to determine how each lead change took place. I already started this sort of analysis in previous years when I invented concepts such as the terminal natural leader, reflecting the last driver to take the lead on track. Just as I think drivers with more TNLs than wins in general are underrated, and those with more wins than TNLs are typically underrated, I think drivers with a better ability to take the lead on track than their statistics would otherwise indicate are underrated, and those who seem to have a level of performance greater than their dueling ability would indicate are overrated. This analysis extends the earlier "How the Races were Won" and TNL analyses to consider every lead change for every NASCAR race on YouTube from 1985-2017. I determined all the lead changes myself except for 2000, 2001, and the first half of 2002, which were determined by Football Perspective's Adam Steele. In subsequent columns, I intend to break this down by track type, by individual track, by year, and by comparing each driver's overall lead change percentage to their lead change percentage on final passes only (to determine which drivers are the most clutch.) I also intend to undertake these same analyses for IndyCar subsequently, and I intend to go further back than 1985 there because there is much better coverage for '80s IndyCar races on YouTube than there is for NASCAR. While I have compiled most of the results for both series already, it will take some time to publish the articles, especially with a day job and my having worsening chronic health issues. However, I spent so much work on this project that I do intend to complete it, although maybe not at the pace I worked at in 2015 or 2016. I eventually want to publish this and some of my other analyses in book form (along with a few future projects I would like to do but have not yet revealed.) I haven't decided whether to try to self-publish or seek out a traditional publisher. I would still provide the main results of analyses on this site, but go into greater detail in a book. For instance, in a book I would likely list every single lead change in each race, but on a webpage that would be considerably more annoying to read through. My columns tend to be far long enough as it is.

Obviously I had to make judgment calls on which lead changes should count and which should not count towards this analysis. The object of this analysis is to determine which drivers were the best duelists, and as such I wanted to make sure I only counted passes where the two drivers seemed to be on a relatively level playing field. By that I don't mean that both drivers had the same equipment, and I would obviously rate drivers who manage to pass other drivers with faster cars very highly indeed. What I mean instead is simply that the drivers battling for the lead are reasonably close in speed and neither is significantly off the pace of the field in general. Typically, all on-track lead changes under green flag conditions were counted as lead changes for the sake of this analysis unless the leader was passed on a pit stop exchange, had a penalty, crashed, hit the wall significantly, spun out, went off track on a road course, or had a significant mechanical failure that led to considerably slowing on the track, even if the failure could be argued to have been driver-induced (including an engine failure, cutting a tire, running out of fuel, having a fuel pickup problem, missing a shift, and the like.) I counted all passes on the start and restarts unless again there was an obvious mechanical problem (for instance, Jeff Gordon missing a shift on the final restart at the 1995 spring Pocono race), although there were obviously some judgment calls especially in the double-file restart era particularly when one driver spun their tires. If the polesitter did not start in first place and the driver who was elevated to the pole got passed by the outside front row starter, I counted the lead change against the driver who actually started on the pole, not the polesitter. I judged all passes based on the leader at the line. If the leader lost the lead before having a problem (for instance, if somebody had the lead and got passed before spinning out - like John Paul, Jr.'s pass of Rick Mears in the 1983 Michigan 500, or several different NASCAR restrictor plate races, where this sort of thing was more common than you'd think) I counted the pass, but if the leader lost the lead because of the problem I did not. If the leader let a car past intentionally (either to conserve tires, conserve fuel, clean debris off their grille, or to give the other car bonus points for leading) I counted it as a pass, albeit not enthusiastically. While that certainly inflates the number of natural lead changes in some races (especially the mid-2000s era Roush Fenway Racing drivers who traded the lead to give each other bonus points a bit too often), usually both drivers would swap the lead with each other multiple times anyway so the passing records more or less evened out in the end. I tried as much as possible to avoid judging the driver's intent. Obviously in many cases it can be good strategy to let someone pass (such as Martin Truex, Jr. letting Brad Keselowski past in the 2017 Watkins Glen race because Truex needed to save fuel, which is actually the only time Truex was ever passed for the lead on a road course) but those instances will still be reflected in the actual results. On-track passes throughout the race however are tremendously undervalued and can give you a totally different perspective of the level of drivers' dominance.

However, in any project like this there will be any number of judgment calls you have to work through. There were many races especially as you go into the 1990s or earlier (not to mention a lot of Talladega races since) where there were lead changes that were not shown during commercial breaks or races that did not have complete footage and I had to use my own best judgment based on where the cars were running before or after as to whether an on-track lead change had occurred or not. While almost all races from the 1990s or later had footage, some races did not, and for a few of the 1980s races I did include, I solely made judgment calls about whether passes were made based on lap leader data and what laps were under green versus what laps were under caution (at tracks a mile or shorter, particularly those that have many cautions, this is very possible) and I did include some races like that from the late '80s that likely have slightly faulty data but that won't make a huge difference in the overall results. There were all sorts of more minor judgment calls that certain fans could disagree on. Although I wasn't thrilled to do this, I counted bump-and-run passes that did not result in spinouts or crashes as legitimate passes for the sake of this study, accepting that most NASCAR drivers in this period pretty much consider it part of the game. Also, if a driver broke loose and slid out of the main racing groove and got passed but did not spin out, I counted it as a pass, while there are certain instances of this sort of scenario that I imagine some fans would not count. If two drivers bumped while side-to-side and one passed the other, or both got passed by the third driver, I counted it as a pass in most cases. There were a few exceptions, such as the last lap pass in the 2013 Fontana race, where the Hamlin/Logano crash did not actually begin until after Kyle Busch had passed them both, but it seemed to me that this would be how most people would judge a situation like that. If a driver had a blatantly illegal car and made passes with it, I did not count those passes as legitimate, but did count the passes against that driver, because passing an illegal car is seriously impressive (indeed, Graham Rahal's pass of James Hinchcliffe on the last lap of the 2016 IndyCar Texas race definitely made me take notice once it came out that Hinchcliffe's car was illegal, for example.) I judged blatantly illegal as a car that received a points penalty of 100 points or more based on the Latford points system scale, or 25 points or more based on the current scale. The hardest situations to judge were usually when a driver cut a tire or had a slowly worsening mechanical problem and got passed but still had more speed than you would expect in spite of that. If a driver managed to pass other drivers for the lead even with a mechanical failure, I did count the passes against. For instance, Darrell Waltrip managed to take the lead at Martinsville in 1989 despite being down a cylinder. If he was fast enough to take the lead in spite of that, I did also count the passes against him, but a situation like that is extremely rare. There were also several instances where somebody slightly bounced off the wall and lost the lead despite not really crashing. Every year, there were probably about ten or so significant judgment calls I had to make but I tried to do so in a way consistent with how I think most other fans would judge those passes. One area I may have not been quite as consistent is in terms of on-track passes for the lead during green flag pit sequences. If there were two drivers off pit sequence and one passed the other, I counted it as a pass, even if it was the sort of scenario where drivers were pitting every couple laps. If driver A pitted, handing driver B the lead, and driver C took the lead from driver B on track before both of them pitted to hand the lead back to driver A, I counted that sort of pass. Where I was less consistent was in determining what to do with instances where a driver passed another driver on track on in or out laps, or instances when a driver passed another driver for the eventual lead. If driver X led at the start of a green-flag pit sequence and still led the drivers who had pitted after pitting, but driver Y passed driver X before everyone had pitted, I counted those sort of passes in most seasons, but changed my mind in the later seasons. Hence, in 2000-mid-2002 and 2014-2017 passes for the eventual lead did not count, but in most other seasons they did. Similarly, I generally counted passes on in laps such as Jeff Gordon passing Kevin Harvick at the 2014 Brickyard 400 before they both pitted, but never leading at the line because Harvick beat him out of the pits, although that was actually the race that changed my mind about considering passes not at the line. I acknowledge these inconsistencies, but to correct for this, I would probably have to completely rerun the study and I don't have time for it now. When if ever I rerun this study, I will adjust this so I will count the pass if the driver takes the lead entering the pits and remains the leader after the pit cycle is completed (as for instance Nigel Mansell did in the 1993 Indy 500) but I will no longer count the pass if the driver takes the lead entering the pits but does not maintain it (as Gordon failed to do.) I am interested in expanding this study in more depth in a potential book, and at that point I would make the analysis more consistent and polished. However, these sorts of scenarios make up a very small percentage of the actual number of lead changes and thereby play a quite minor role in determining the overall results regardless. However, generally on outlaps I was more consistent. Since it takes more than a lap for drivers to fully get up to speed, I don't consider passes for the lead on outlaps to be fair game here since usually the "passer" and "passee" will not be going at the same rate of speed. I did the same sort of analysis (not counting passes for the lead on outlaps) for IndyCar as well. I realize many people would argue many of the most impressive passes are when a driver passes somebody who has just left the pits on cold tires (at least on road courses) and I would somewhat agree with that perspective. However, obviously the outlap relies heavily on the pit stop and the pit stop relies heavily on the team. What decided this for me was looking at old newspaper reports, which I frequently did for '80s races. There were many times in IndyCar road course races where there were passes on outlaps but even the newspaper articles at the time viewed those as passes in the pits even if there technically was an on-track pass of a leader with cold tires. If the experts of that time tended to view passes like that as passes in the pits as it seems like they usually did, I decided to follow the same tradition, though I don't blame you if you disagree. This is not and cannot be an exact science, but that doesn't mean it isn't worth doing. I think a lot can be and will be revealed about certain drivers' careers in undertaking this analysis as I have uncovered more information about drivers' dueling ability and going further back than I think I have seen in any previous article. While no one will agree with anyone else on whether every single pass should count or not, the vast majority of passes for the lead in NASCAR races are not disputable, so the minor judgment calls really won't make a difference in most drivers' results in the end, particularly those who have led races often.

Below I provide a table listing all drivers sorted by the number of races they made an on-track pass for the lead. If multiple drivers had the same number of races led naturally, I then sort drivers by their overall lead change percentage (the ratio of their number of lead changes for to the sum of the number of lead changes for and against.) There were four races that had no on track lead changes (spring Atlanta in 1991, fall Rockingham in 1992, spring Richmond in 2008, and fall Phoenix in 2015.) For these races I gave the TNLs one race led naturally each but did not award them any lead changes. However, if the polesitter never made a pass for the lead on track, I did not count it as a race led naturally. All races from 1994-2017 are included in the analysis, and almost all races from 1990 onward are included, however there is weaker coverage the earlier you go prior to that, so drivers who peaked in the '80s or earlier will be underrated by this, but I think it is still clear that the results have meaning.

It should come as little surprise that Jeff Gordon has taken the lead on track more times than any other driver of this era, considering he holds almost all counting records in this time period. What may come as a bit more of a surprise is how large his lead over Dale Earnhardt and Jimmie Johnson is, considering how both of them aren't that far behind him in wins. To be fair, Earnhardt does have several of his wins missing, but not too many as apart from his 1980 championship season, he did not win many races in a season prior to the 1985 season, which was the first season included in the study. What may be more telling here is that the two seven-time champions have significantly higher lead change percentages than Gordon does. In much of Gordon's heyday in the late '90s he had by far and away the best pit crew, which may have overrated his passing ability to some degree. Johnson and Earnhardt are definitely (and not surprisingly) both standouts when it comes to dueling ability, as there are not many drivers with a lead change percentage of greater than 54%, and Johnson never quite had the pit crew advantage to the degree that the Rainbow Warriors did. Earnhardt certainly did with the Flying Aces in the early '90s, but obviously with a lead change percentage that high he was hardly taking the lead only on pit stops. This would indicate that while Gordon had a longer run of success in this period than either Johnson or Earnhardt did, Johnson and Earnhardt definitely had greater peak performance at least when it came to dueling ability. Considering Johnson and Gordon drove for the same team, Johnson's 56.76% lead change percentage and Gordon's 51.20% is a staggeringly huge difference. That alone may explain why Johnson and Earnhardt ended up with more titles, and it may also explain why their win totals were so much closer to Gordon's than their number of races led. Johnson and Earnhardt were simply better clutch performers, and that definitely does seem to reflect the realities of their careers.

Almost all the drivers generally regarded as great do have lead change percentages greater than 50% indicating that being a great driver in NASCAR almost necessitates being a great duelist. Those drivers that have lead change percentages below .500, such as Mark Martin, Matt Kenseth, Alan Kulwicki, Terry Labonte, and Darrell Waltrip are all drivers who had a reputation for conservatism, which definitely matches perceptions. I realize Waltrip was much more aggressive earlier in his career, but in the time period we are talking about from the mid-'80s onward he was not, so he is not a surprise either. Besides those five, almost every other major or near-major driver did pass others more than they were passed (and even those other five came quite close.) However, the vast majority of major drivers do fall within an extremely narrow range between 47% and 53% indicating that we are generally talking extremely minute differences here between the elite drivers. Almost all of them pretty much pass other drivers as much as they are passed, which makes the drivers who fall outside this range considerably more interesting.

Sorting by percentage, the best duelist in this era of NASCAR was Tony Stewart, which makes a great deal of sense. Although he did not really have a reputation as a clutch performer since he seemed to cost himself a lot of race wins, and didn't win many marquee races (which is one thing many people care about much more than I do) he did win the championships all three times he strongly contended for it (I don't really count 2009 because the Chase nullified his advantage.) Although other drivers were more clutch than Stewart at the end of races, and I will reveal shortly that Jimmie Johnson, not Kevin Harvick, was the most clutch driver, Stewart more reliably and consistently passed people throughout the race. He took the lead very often during long green-flag runs before losing it under caution and was the hardest driver to pass. This makes a lot of sense when you think about it particularly considering some of the summer Daytona races he dominated start to finish. Almost nobody else dominated restrictor plate races to that degree without a stronger equipment advantage than it seemed like he had (in the races Dale Earnhardt, Jr. dominated his DEI or Hendrick teammates were usually also fighting for the win, but that was not generally the case amongst Stewart's Gibbs or Stewart-Haas teammates.) However, Stewart's dueling dominance did not merely come on plate tracks; this was a trend throughout the vast majority of his career and on all track types. He was the most consistent and versatile duelist in the sport for a very long time and had a .500 or greater dueling record every season of his career until his decline began. Stewart's percentage of over 59% is so significantly higher than anyone else's among drivers with a large number of starts that it begins to become apparent why Jeff Gordon touted Stewart as a greater driver than even he was.

On the flip side, I have revealed that Stewart's onetime teammate Ryan Newman's reputation as supposedly being the hardest driver to pass is one of the biggest myths in NASCAR history. Of all the major drivers of the last 30 years, Newman is by far the worst duelist in recent NASCAR history, and no one else is even particularly close. This really shouldn't be surprising when you think about it harder. If Newman really was the hardest driver to pass like he is so often hyped, he would have converted his 51 poles to a lot more than 18 wins, especially when you consider how he lucked into a greater percentage of his wins than most other drivers with his number of wins have done. To be fair, Newman is obviously hurt by the fact that he led at the start of many, many races before losing the lead and I do not count races he won the pole to his benefit. However, not every driver who won the pole got passed on track (many held onto the lead until being beaten out of the pits on a caution, for instance), so it still seems to be a valid criticism. It makes one wonder where Newman's reputation that seems to be completely wrong came from. Obviously, no one seems to have done this research before so that's a part of it, but I think part of it must be that Newman goes off-strategy so much to give himself the lead but because he tends to lead on strategy more than on speed, he therefore usually faces a fresh tire deficit and is slower than the cars behind him, thereby holding them up. He probably has to lead on strategy so much in the first place because he isn't as strong as other drivers at passing. In his 2003 season, when he did make passes for the lead, he often passed drivers who were also off-strategy but did not pass the dominant cars in the race. Even in his heyday, when he did lead lots of races, he consistently got passed more than he passed other drivers. Much like Stewart almost never had a negative lead change ratio, Newman never had a positive lead change ratio in his entire career. Somehow the easiest major driver to pass has become misunderstood as the hardest. This is probably the biggest surprise of the entire analysis, but when you think about it harder (especially considering his pole/win difference), it's really not as surprising as it may seem. Considering Stewart and Newman were actually teammates, this possibly makes Stewart look even better and Newman even worse, since they theoretically had the same kind of equipment.

While the drivers with low lead change percentages are generally the drivers you would expect (either those that are not major drivers or major drivers with extremely conservative driving styles, although some might be surprised by Newman) the drivers with very high lead change percentages also seem to be drivers people would expect. Johnson, Earnhardt, and Stewart are probably considered the three most clutch drivers in championship battles of this era and they all fare extremely well. So does Kevin Harvick at over 55% who did score the most overall points in a season three times but only won one championship. Although Harvick's "closer" reputation is overblown, especially after he went to Stewart-Haas, he is a great duelist, so unlike Newman's hard to pass reputation, Harvick's closer reputation obviously has some validity. When more and more races in recent years had cautions and restarts late in a race, not to mention double-file restarts even in the closing laps of a race, that will obviously give the best duelists such as Stewart, Johnson, and Harvick a major advantage, and maybe it also explains why Gordon suddenly started winning a lot less after the double-file restarts were introduced (since he wasn't the best duelist, and Gordon's dueling strength was usually best on long runs, he was hurt immensely by that.) With regard to Harvick, it's more that he was excellent at passing other drivers at any stage of a race, but it was more noticed because of how often he did so on late-race restarts (although Johnson was a better duelist throughout the entire race as well as at the end of races, not to mention his many more championships, so he really deserves the closer title, not Harvick.) Among drivers with shorter careers, there are four noteworthy standouts. Tim Richmond and Ernie Irvan should be no surprise to anyone, as they were known especially for their dueling ability against anyone of their era, and they were especially good duelists on all track types. However, Richmond and especially Irvan were obviously more mistake-prone as a result of the risks they and few others would take, so they can't quite be rated with those other drivers, as somebody like Harvick had a similar lead change percentage despite having a very conservative driving style (and one of the lowest crash DNF percentages in Cup history.) Regardless, everybody knows drivers like Richmond, Irvan, and Davey Allison (who I now admit I was fairly wrong about in earlier columns - I felt Allison was overrated because he didn't have enough TNLs, but Allison was stronger at taking the lead earlier in the race than he was at the end, much like Kyle Busch) were exciting to watch, and that clearly shows up in these data as all three have extremely strong lead change percentages. Martin Truex, Jr. also has a percentage over 54%, largely due to the fact that his equipment in 2017 seriously may have been the most dominant of any driver in this period (Truex in 2017 was tied with Jeff Gordon in 1997 and Jimmie Johnson in 2008 for the most races led naturally in a given season, but Truex had a far higher percentage this year than either of them did.) However, the other standout is a bit more surprising. Bobby Hamilton's lead change percentage of 54% pops off the screen compared to the other drivers with similar numbers of lead changes to him. He was a strong duelist on all track types except for unrestricted superspeedways, which makes sense because he had an equipment disadvantage for most of his career. On every other type of track - road courses, short tracks, flat tracks, and plate tracks, he took the lead on track at least as often as he was passed. Hamilton was hyped by some such as Michael Daly as the most underrated driver in NASCAR history, and these data prove there is definitely an argument for that. Given the equipment he had, which usually wasn't as strong as most of the other drivers who battled for the lead, it's astonishing he managed to take the lead on track 54% of the time when a lot of drivers with faster cars and seemingly more success did not. I always suspected that Hamilton deserved to be hyped more than other contemporaries who the fans seem to hype more: Ward Burton, Ricky Craven, and Johnny Benson, who never impressed me as much as Hamilton, and this seems to be very convincing proof.

Overall looking at these data can give you a very different and perhaps more accurate way of looking at drivers' careers. If you believe the finish is the only thing that matters as many fans do, you're welcome to ignore this. However, if you believe as I do that drivers' battles with other drivers are what we tend to remember more than passes in the pits and that dueling ability plays a significant predictive role in wins and even clutch performance and championships (particularly in an era when there are so many late-race double-file restarts, which is definitely different from NASCAR ten years ago or more) it's something to strongly consider when evaluating driver talent that has been heretofore overlooked. Obviously performance at the end of a race matters more than performance early on, and I will discuss clutch lead change performance in an upcoming article, but these results obviously capture something if the drivers we think of as conservative generally do have losing lead change records and the drivers we think of as aggressive do tend to have the highest ones. Just looking at old-school statistics such as wins and laps led, this may not be as quantifiable, but this obviously does begin to capture a lot of what we consider the intangibles of certain drivers: what excites people and draws them in as fans. The fact that this does generally tend to match the fan consensus on drivers based on the perceptions obtained from just watching the races rather than looking at stat sheets (admittedly with a few surprise exceptions like Ryan Newman and Bobby Hamilton) does indicate this may have a lot of value. I will continue this analysis over the next few weeks prior to the start of the NASCAR season before beginning my similar IndyCar analysis prior to the start of the IndyCar season. I'd certainly love to do Formula One too, but obviously you can't obtain past video footage from F1 so easily.

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.

---