Luck and Terminal Natural Lead Changes: 1990-2015

by Sean Wrona

Continuing with the ideas I introduced in December, I decided to calculate the terminal natural leaders for every Formula One, IndyCar, and NASCAR Cup race from 1990-present. Although one may see this as very similar to the How the Races Were Won series I did last year, there are a number of reasons that looking at these data in this way are superior. I have also changed the definition for what I consider a natural win to streamline it for greater consistency since last year. One of the main problems with calculating natural wins the way I was doing it before is that certain eras in each series seem to be more conducive to races being won naturally than others; this can even wildly vary from year to year. By calculating it based on the terminal natural leader (the last driver to take the lead under green flag conditions) I can have a driver for every single race. Since the idea ultimately is to measure which drivers took the lead on track more often and which drivers relied on taking the means through other means more often, it makes mores sense to have data for every single race.

I have redefined what a natural win is for this analysis. While I continue to include all races where a driver takes control of the race ultimately via an on-track natural pass for the lead (even if the eventual leaders are not in the lead at that moment and are off pit-sequence), I have now decided to be much more generous in what other passes I will include as well. First off, if a driver is the last driver to take the lead on track but then loses the lead not naturally and regains it unnaturally (for instance if the terminal natural leader is beaten out of the pits by another driver and then beats this driver out of the pits later, which does happen) I now count that as a terminal natural lead or natural win, which I was not doing before and that is quite silly. Additionally, I used to be very stringent about not counting passes where the leaders made any sort of contact, but I have since changed my mind. There are numerous incidents even in open wheel racing where drivers make slight incidental contact and it does no damage to either driver's car. I will now only discount such passes if the trailing driver spins out, wrecks, or causes damage to the leader's car. I now consider simple bump-and-run passes part of the game. I still don't like them very much, but it can be too, too arbitrary to determine for instance in a NASCAR race which incidents were bump-and-run passes and which were merely taking air off the leader's spoiler, which can look very similar, so Jeff Gordon bumping Rusty Wallace counts, Jeremy Mayfield bumping Dale Earnhardt counts, Terry Labonte bumping Dale Earnhardt on lap 499 at Bristol counts, but Dale Earnhardt spinning out and wrecking Labonte the next lap, Joey Logano spinning Matt Kenseth, or Robby Gordon wrecking Jeff Gordon at Loudon do not count as natural passes. I think it's a more consistent model than I had before. However, if the 3rd place driver or somebody further back inherits the lead when the leaders make contact (and was not part of the contact) I continue to count that as incidental (lucky). I also still count passes as not natural if the leader was penalized for a penalty (either during the race or post-race), had an illegal car, if the lead changed hands in the pits, if the leader has a mechanical problem (including the polesitter not making it to the grid) or if the leader slowed down to avoid a wreck ahead (because I largely see wreck avoidance as coming down to luck based on the positioning of the lead cars when a wreck among lapped cars occurs). If there are no natural on-track passes for position, I then default to the polesitter. While I did not check the video footage for every single race (indeed, for many of these races it doesn't exist online) as much as possible I judged terminal natural lead changes based on lead changes at any point of the lap even if they were not sustained at the start-finish line. For instance, in the 2008 NASCAR Kansas race, Carl Edwards passed Jimmie Johnson on the last lap in a banzai run to the checkered flag then hit the wall and gave the lead back to Johnson. Even though he did not actually lead the lap, I gave him credit for the pass, although this particular principle did not affect too many other races. Finally, if a driver intentionally lets another driver by for some reason, I now count those passes as natural as well unless there is significant evidence that team orders may have been invoked (this applies to several Formula One races and the 2005 IndyCar season finale at Fontana, among others). I also initially excluded various other races such as the Pocono race where Brad Keselowski let Dale Earnhardt, Jr. pass to clear off some debris on his grille, but since changed my mind. In most cases you can't necessarily read a driver's mind and know whether a driver let somebody past intentionally or not. In some cases it can be good strategy (such as saving fuel) and just from watching you can sometimes but not always read much into it. To be more consistent and fair, it makes sense to only include races where it seems there was evidence something was planned out in advance, and these are the races everybody knows about and still talks about (like those two 2002 Formula One Ferrari swap races). I think based on all these decisions, it becomes much more clear determining which passes were natural and which were not. The biggest problem I had by far was races where the leader was having a car break down slowly. Usually, I tended to judge these on whether the driver passed was still staying close to the leader or not or whether the former leader was losing time rapidly, but these are the hardest races to figure out and are the easiest to criticize. Finally, I now count strategic assist passes as natural passes. I didn't want to do it because I have been critical of Ryan Newman and Danica Patrick's fuel mileage wins where they each passed leaders who weren't really factors in the race, but the way I see it in most of the races I called strategic assist races, what happens is that for instance the former 3rd place driver passes the former 2nd place driver on track after both of them beat the leader out on a pit sequence. While there are some bizarre fuel mileage results and this system created a few new ones, I think there is too much data lost if I do not count these passes.

I did have to make judgment calls especially for the early years of NASCAR, which are the worst recorded of the three series I am covering here. While I was able to come up with some combination of video footage and online and/or newspaper reporting (either contemporary or modern) to identify which passes were on-track or off-track sometimes there weren't adequate data available and I had to make a guess at the terminal natural leader based on the lap leader data available, but I think I usually did a good job of this. There are a few more clarifications I want to address. If the leaders pitted and there was a lead change among the leaders who went off pit sequence and it was the final lead change of the race (this only happened a few times), I would count it as the terminal natural lead change if the driver who led the previous stint on the main pit sequence and the driver who led the following stint on the main pit sequence were different, but would not count it if a natural leader took the lead on track then pitted and another lead change happened before the leader got back to the front. This has been a fairly common kind of thing in recent IndyCar road races. In the second Detroit race in 2013, Mike Conway pitted, then Tristan Vautier passed Tony Kanaan on-track while they were off pit sequence before pitting a lap later and hanging the lead back to Conway. In that race, I gave the terminal natural lead to Conway, even though Vautier technically made the last on-track pass for the lead in the race. However, in the 2003 IndyCar Nashville oval race, I gave it to Tora Takagi even though he wasn't a huge race factor because the leader on the previous stint (Scott Dixon) and the leader on the final stint (Gil de Ferran) were different. Similarly, in the most recent IndyCar race at Sonoma, Tony Kanaan passed Sebastian Saavedra in the only on-track lead change of the race and I gave it to Kanaan rather than Will Power (who led the first stint from pole) or Scott Dixon (the eventual winner who really took control of the race by beating Power out of the pits). I initially gave it to Power in my earlier column, which may make more sense, but changed my mind because on-track passing is what I want to reward and because in so many IndyCar races on road and street courses, they often come down to pit timing and dumb luck, and if a caution had come out at just the right moment for Kanaan, that could have been a winning pass. There's no way of knowing how a race will ultimately unfold.

What does this measure? I believe it does a better job of who had the most clutch run in a race, independent of the actual race results which include team-related factors and luck. While I fully acknowledge that an attempt to measure luck may be somewhat foolhardy because one can argue some of the things that are considered luck (avoiding crashes, conserving equipment, not running out of fuel, benefiting from other drivers' mistakes/spinouts/off-track excursions, beating someone else out of the pits) are part of driving skill. I agree. I am not saying in all or even most cases where the terminal natural leader and the winner differ that the terminal natural leader "should have won" (although that is definitely certainly the case sometimes to fairly often, and I'll add that the vast majority of these instances come down to passes in the pits, where the driver can play some role especially in races with long green-flag runs, but often doesn't, especially in races where the field is all bunched together). However, while I don't necessarily agree with any particular individual race outcome difference here, I do feel that in aggregate the difference between drivers' terminal natural leads and race wins measures something. To some degree it is the strength of the overall team. To some degree it is capitalizing on others' mistakes (or not). To some degree it is luck. I think I can summarize most of the discrepancies in general as coming down to luck, and the remaining debate is whether you believe drivers make their own luck or if teams play a factor. The answer is clearly both, but while there were some wacky results that came out as the terminal natural leader (drivers I am almost certain would not have won those races) I do think it is a good measure of who had the pace and the speed to win even if something happened, and who did not have quite the same pace and won anyway. I have a personal bias towards hard chargers (drivers who would more likely have more TNL than race wins), but I honestly think they matter equally, but it depends on the series. In F1, most unnatural wins occur when there is a lead change in the pits on a long green-flag run, where the driver may play most of the role (by fuel-saving and making really fast outlaps), but in NASCAR and IndyCar, I think luck is much more important, so I would tend to prefer the win list in Formula One but the TNL list for NASCAR and IndyCar where the more frequent cautions can shake up a field too much.

Below I list all the races where the terminal natural leader and winner differ for Formula One, IndyCar, and NASCAR from 1990 to present, followed by a table indicating how many race wins each driver gains and/or loses when switching from wins to terminal natural leads based on on-track passes.

Formula One

As I stated above, I am much more inclined to accept the Formula One win list as meaningful relative to the terminal natural lead list because drivers do play a role in their luck obviously. I have Michael Schumacher losing 21 wins based on this model which seems a little excessive. Yes, he did have the overwhelmingly best team in his Ferrari years, but he also made that team, which had been struggling for about 15 years before he got there. Quite often he stayed out of the pits a lap or two later than the driver he was chasing and made a pass based on a really fast in lap or out lap entering or leaving the pits, which certainly takes skill. I would say most of his discrepancy is from that alone, but I still don't think we can say all of it was. I do think the quality of Schumacher's pit crews did play a big role, while other drivers may not have had quite as strong teams overall. While most of the other legendary drivers of the past 25 years gain wins when converting to terminal races led, Fernando Alonso does not and I think that comes down to the quality of his teams. There are so few on-track passes for the lead in most Formula One races the TNL often comes down to who wins the pole or leads at the start, and a driver who has had and won with slower equipment like Alonso does not have the speed to qualify as quickly so in order for him to win at all he has to win by beating people out of the pits. I would not criticize Schumacher (who still has a big lead in this category regardless) or Alonso for having more wins than terminal races led. I do think there is room to criticize the drivers who aren't quite as legendary as those two who have fewer terminal natural leads than wins, particularly Kimi Räikkönen, Jacques Villeneuve, Mark Webber, and especially Jenson Button, who loses seven of his fifteen wins this way. Of all the major drivers in Formula One over the top 25 years, Button would seem to be the luckiest. As for the unluckiest, that would be Juan Pablo Montoya, who had 13 terminal natural leads compared to his 7 overall wins, with Felipe Massa's difference of 18-11 not far behind. Massa was usually beaten out of the pits when he lost, while Montoya brought on a lot of his problems himself by crashes (although he did also have a lot of mechanical problems that ruined his day as well...had he just not ha two engine failures while leading in 2003, he could have won the World Drivers' Championship that year). Four drivers who had wins in actuality would have been winless without benefiting from lucky off-track passes (Johnny Herbert, Heikki Kovalainen, Pastor Maldonado, and Olivier Panis), while two drivers who are still winless (Timo Glock and Nico Hülkenberg) would have had wins according to this model. I suspect more people would pick the latter group over the former group.

IndyCar

I think the difference between win totals and terminal natural leads makes a much clearer difference here than on the Formula One list. Many fans have speculated that the reason Chip Ganassi has usually come out with the championship in recent years despite one Roger Penske driver or another leading the points standings for most of the season most years is because Ganassi has a much better pit crew and makes much better pit calls and tremendously benefits on pit stop exchanges, while the Penske drivers, who usually qualify better on road and street courses, particularly Will Power, consistently get burned while Ganassi drivers like Scott Dixon who qualify slightly worse can take advantage by taking chances in terms of pit strategy. Indeed, the results fit this model. Most of the Ganassi stars (Scott Dixon, Dario Franchitti, Alex Zanardi, Jimmy Vasser) lose races according to this model indicating they got lucky or benefited from strategy more than other drivers did, but Juan Pablo Montoya, who also had many more terminal natural leads in Formula One than wins, was a big exception, indicating he was much more responsible for his dominance in his day than the other Ganassi drivers were theirs. I think this may explain why Montoya succeeded in Formula One and Zanardi failed. Meanwhile, most drivers associated with Penske in recent yeras had more terminal natural leads than wins, including Will Power, Paul Tracy, Sam Hornish, Gil de Ferran, Ryan Briscoe, and again Montoya. The consistency within each of these teams seems to indicate that Ganassi's advantage does seem to be from its strategists more than its drivers (which isn't to say that Dixon and Franchitti aren't necessarily all-time talents). To be fair, some of this comes down to style. Dixon and Franchitti have very conservative styles and again like to beat people out primarily on blistering outlaps while driving steadily and conservatively the rest of the time, as opposed to a Power, Tracy, or old-school Montoya who were all aggression, all of the time, and probably cost themselves wins as a result. Among the drivers with many wins, the luckiest by far was Adrián Fernández who won 11 races but was only the terminal natural leader five times. I think that sounds right too. Of all the IndyCar drivers to score ten or more wins in this period he does seem to have the lowest reputation by most. In most of his prime years, he drove for Patrick Racing, who had Jim McGee, one of the most legendary pit strategists, every bit as much as Mike Hull for Ganassi today, and he probably benefited considerably as a result. Others who have overall been very lucky (luckier than Dixon) include Ryan Hunter-Reay, Emerson Fittipaldi, Simon Pagenaud, and Danny Sullivan, who in this period had even smaller ratios between their terminal natural win total and their actual win total, but at least in the case of Hunter-Reay and Pagenaud that may be because in many of their peak years they didn't have the equipment to have killer speed and needed to find strategic ways to even compete. While I think Dixon, Franchitti, Sébastien Bourdais, and Al Unser, Jr. have been legendary enough to not be criticized too much here (they did benefit from the strength of their teams, but have enough success to be above criticism), I don't think this applies to Hélio Castroneves. All his full-time teammates except for Simon Pagenaud (Will Power, Gil de Ferran, Sam Hornish, Ryan Briscoe, and JPM) have more terminal natural leads than wins implying the Penske team in general has more pace than its eventual results, but Castroneves has seven fewer wins than terminal natural leads in spite of that. All his teammates would seem to be unlucky while he is lucky. Is he doing something to make his luck or is he just getting lucky? I think it might be the latter, because that difference is huge. While the Castroneves and Franchitti marks do support my image of them as two of the more overrated drivers in IndyCar history, I may have to consider Dixon and Bourdais as well. What is neat is that most of the drivers I did think were undervalued (Gil de Ferran, Justin Wilson, Kenny Bräck, Juan Pablo Montoya, Bryan Herta, and so on) did all have more terminal natural leads than wins, indicating they all did have more pace than their results would indicate. This is particularly impressive again for the late Wilson who did not have stellar equipment. The wins Montoya gains put him in a wildly different perspective considering how short his career is, while best buds Tony Kanaan and Castroneves end up tied even though they have wildly different win totals (and Kanaan has had generally worse equipment). I think this does a fine job of measuring race pace and other drivers who got kind of hurt by my previous metrics (such as Jacques Villeneuve who only had one natural win before but now actually gains a race losing TNL) also improve. This isn't how I'd rank these drivers either and I think Power and Tracy's penchant for choking make them not the best two IndyCar drivers of the period, but I do think in many respects it is better than the win list. Quite a few drivers who actually won races become winless using this model (Christian Fittipaldi, Jim Guthrie, Carlos Huertas, John Andretti, Carlos Muñoz, Eliseo Salazar, and Oriol Servià), while Vitor Meira, Affonso Giaffone, Raul Boesel, Memo Gidley, Tristan Gommendy, Jimmy Kite, and Tora Takagi become winners. Unfortunately, unlike the NASCAR list, I think I clearly view the first list in general as better than the second in general, indicating the value of terminal natural leads and wins may not be as clear here.

NASCAR Cup

The order of the NASCAR win list doesn't change a whole lot when you switch to terminal natural leads instead of wins. Most of the drivers are more or less in the same general range. However, some drivers were extremely lucky or unlucky and that still must be noted. Most people go on and on about how lucky Jeff Gordon and jimmie Johnson are. No, not really. They were both on the lucky side with a few more wins than terminal natural leads, but not by much. Better examples of lucky would be Kyle Busch and Bobby Labonte, who lose seven wins each, Joey Logano and Mark Martin who lose five, Ricky Rudd, who loses four, and Davey Allison and Ryan Newman who lose three. The unluckiest drivers were Ernie Irvan, who led the way with eight more terminal natural leads than wins, for a corresponding total of 23 TNL, Sterling Marlin, who had six more TNL than wins for a total of sixteen, ranking him over Bobby Labonte and Ryan Newman. A few years ago I said sincerely to someone in private that I thought I rated Marlin higher than Labonte because at least he was arguably the best driver at his specialty rather than 6th-10th everywhere as I felt Labonte was in his prime, not to mention that I think Marlin would have been more likely to do what Labonte did in the #18 than Labonte would have done what Marlin did in the #4 or #40. It's nice to see some backup for htat. The most conservative drivers (Mark Martin, Jeff Burton, and Bobby Labonte) did end up scoring more wins than terminal natural leads as a result of their conservative, along with Gordon and Johnson, while more aggressive drivers like Tony Stewart, Rusty Wallace, and Ernie Irvan usually gained. Interestingly, defying their general stereotypes, conservative Terry Labonte had one more terminal natural lead than win while hard charger Kyle Busch lost as many wins as anyone (I guess career-wise his reputation of being the worst choker in NASCAR is correct, even though he was definitely the best closer in NASCAR in 2015). Dale Earnhardt ended up losing three wins in this regard because in the early '90s he did have the best pit crew (or second to Rusty Wallace's) and he also had a very conservative style for the most part which also probably cost him some race wins in exchange for more consistent championship runs (I'm sure if you count Earnhardt's '80s he would have more terminal natural leads). His son, much to my surprise, gained a lot and is clearly proving that I have been underrating him considerably for years. What's interesting is apart from Earnhardt, almost every other driver who was regarded as a restrictor plate specialist (Stewart, Irvan, Marlin, Junior, Michael Waltrip, although this is certainly unfair to Stewart and Irvan who could win anywhere) gained a lot, indicating ability to win plate races and make passes there may have something to do with passing ability everywhere else as well. The most unlucky driver by far was Martin Truex, Jr. who only has three wins but has been the terminal natural leader in eight races. Truex has 8.63 cumulative races led, which along with his only three wins, seems to indicate him as one of the biggest chokers in NASCAR history. Not really. When you consider he was in position to win eight times (before usually being beaten out of the pits by stronger teams) that lines up very nicely with his actual level of dominance, and I believe you will find this list a bit more consistent with dominance lists as well. Ernie Irvan with 19.02 cumulative races led looks like a choker with only 15 wins, but if you consider his 22 terminal natural leads, he does not. Bobby Labonte looks like one of the best closers in history with 21 wins versus 13.09 cumulative races led, which is behind the likes of Marlin, Newman, and Kahne, and he is right behind them or in the same league as them on this list too. Kyle Busch with 36.31 cumulative races led and only 27 terminal natural leads does look like the worst choker in today's NASCAR (again, not last year), especially considering he is tied with Carl Edwards who only has 17.09 cumulative races led and is way less dominant, and maybe the best closer in NASCAR, in spite of not being able to (yet) close on a title. This analysis also can be used to shape very different impressions of individual seasons. In 2003, Ryan Newman won 8 races to Dale Earnhardt, Jr.'s two but Newman lost three races and Junior gained three races in terms of TNL so they end up tied in that category. More dramatically, in 1997, Dale Jarrett gains two races while Gordon loses four, implying maybe Jarrett should have been the driver to win ten races that season. Considering how much faster the Fords were and how Jarrett was the most dominant driver that season, this kind of makes sense. I think considering this, he has a much stronger case of having been "robbed" by the officials than Mark Martin in 1990, considering he was penalized for tire smoke that turned out to be nothing at Watkins Glen costing him about twenty positions when he only lost the title by fourteen points. Looking at it this way, he may have legitimately had a stronger season than Gordon. Of course Gordon should have had 1996, so it evens out in the end. In 2002, Stewart had a rather ambiguous and ugly-looking title based on his mere three wins, but he had a total of seven terminal natural leads, easily the best of that season, making him a much more obvious champion from that standpoint. Again, I repeat that I do not think the terminal natural leader should always be the winner, but I do think in many cases (certainly not all of them), this does a fairly good job of spotlighting drivers who are overrated/underrated or who have been lucky or not, but some drivers do gain or lose races solely due to their driving styles which can cause the lucky outcomes, so it is fairly complicated but I believe I have got something here. Four drivers who actually won NASCAR races become winless by this model (Brett Bodine, Trevor Bayne, Casey Mears, and Regan Smith) while four who never won earn wins (Chad Little, Ted Musgrave, Mike Skinner, and Ricky Stenhouse). I do believe that second group is stronger.

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.

---