2018 Leader Statistics and Driver Ratings

by Sean Wrona

Even though I have not published a new Racermetrics column for the past eight months, I have a lot of new material that I will likely be releasing here over the next several months based on research I have been working on while I was inactive. I know I abruptly stopped before I finished repeating my NASCAR and IndyCar year-by-year lead change analyses for the CART era of Champ Car/Indy Car racing as I promised, but I will be starting that over the next series of articles. Since I am aware my articles tend to run way too long, I am probably going to split up this analyses into several articles per decade, one each for the 2000-2007, 1990-1999, and 1979-1989 decades of CART/Champ Car racing, but I can even one-up that now as I have been doing a massive amount of historical research to attempt to analyze every lead change back even to 1930. This is much more possible in terms of IndyCar racing than it is in terms of NASCAR, because the earlier years are much better recorded and were talked about much more in newspaper articles when IndyCar racing was the most popular form of racing in the US, while NASCAR was still considered niche and regional, and also because the races tended to have fewer lead changes, and they were nearly always recorded in contemporary newspaper articles. For NASCAR that would be much harder, because the only races that received national coverage were the superspeedway races and superspeedway races tended to have far too many lead changes, most of which were not mentioned in contemporary newspaper articles.

Similarly, I have collected lead changes for every Formula One race from 1967 to present with only a relatively small number of races unaccounted for. What is obviously frustrating about F1 (and the reason why I did not attempt to do this sort of analysis for F1 earlier) is the fact that video footage is hard to find for most past races. However, I was able to find footage for a lot of the races where I had questions as to whether the lead changes were natural or not, and the race descriptions I read through from grandprix.com were so descriptive I was usually able to deduce the reasons for every single lead change from them. Hence, while I think it would be almost impossible to go back much earlier than the mid-'80s for NASCAR racing, I am having much more success going much further back in time for both F1 and IndyCar racing, and those will likely be the next analyses I will post. For both series, I am going to start with a summary of drivers' natural races led and lead change records overall and perhaps splitting between various track types before I proceed with the year-to-year analyses as usual. I am only missing seven races (although I'm uncertain about a few others) for F1 and 58 IndyCar races (but that is still a very small percentage of the number of races held from 1930 to present.)

I am not going to release the entirety of those results now, but I will provide two noteworthy tidbits. Much to my surprise Lewis Hamilton is already leading in natural races led and has already overtaken Schumacher on the F1 list, and on the IndyCar list, I was very surprised to observe that Tony Kanaan and Hélio Castroneves were actually fourth and fifth overall in races led naturally. Clearly I did them a vast disservice in my top 100 IndyCar drivers ranking. In general, I can see now that while I invented a lot of great stats for it, I overrated drivers with short, dominant careers and underrated drivers with long, respectable careers but less noteworthy peaks. While I think rating them 4th or 5th on an overall list would be extremely wrong, the positions I chose for them were an equally extreme overreaction. I was too influenced by the analyses of folks like F1metrics, who ranked drivers exclusively based on teammate vs. teammate comparisons and three-year peaks, leading him to conclude for instance that Elio de Angelis is better than Nigel Mansell because of their teammate comparison, while ignoring that Mansell had a much longer career; obviously this sort of thinking is what allowed me to rank Gil de Ferran over Castroneves as well. de Angelis did outperform Mansell and de Ferran did outperform Castroneves when they were teammates, but ultimately they probably didn't have the longevity to justify higher rankings. Especially in de Ferran's case: Castroneves actually won one more race than de Ferran did when they were teammates, although de Ferran was substantially more consistent.

However, with this post, I am going to start by reviewing the completed 2018 seasons for NASCAR, IndyCar, and Formula One. I am going to do them in that order simply because lead change data are more interesting for NASCAR considering that there are many more lead changes to work with, while in many IndyCar road races and in most F1 races there aren't any on-track lead changes at all. I am also introducing a new method of rating drivers that includes both a reflection of their consistency and their on-track passing. In the past, I ranked drivers on these tables according to their number of lead shares, but as with any singular statistic, that in itself does not tell you who had the overall best season. Dale Earnhardt, Jr. had the most laps led, lead shares, and cumulative races led in the 2002 season but even his most ardent fans would not attempt to argue he had the best season that year.

However, consistency measures alone can obviously be equally flawed. Few people would actually try to argue that Terry Labonte in 1996 or Matt Kenseth in 2003 were the top performers in those years. Mark Martin had the best average finish in 1990, 1997, and 2002, but few would argue that he was the best driver in any particular season, unless you think based on his contemporary IROC results that his Roush cars were at a significant equipment deficit to the other premier teams. Even in series that reward top finishes more than NASCAR historically does like CART, you occasionally get similar results with Al Unser winning one race to Teo Fabi's four but winning the 1983 championship regardless. Having said that, consistency certainly has value and I wouldn't want to automatically say the driver who won the most races had the best season either. NASCAR again provides the best examples here, as I don't think too many people would argue Ryan Newman in 2003 or Kasey Kahne in 2006 were the best drivers in those years, but instead relied on pit strategy and an absurdly powerful package on intermediate tracks respectively.

What I wanted was a method of measuring seasons that reflected all of the above measures in aggregate rather than considering one aspect of driver performance alone. I did not want a statistic that rated Dale Earnhardt, Jr. as having the best season in 2002 when he was maddeningly inconsistent, nor did I want a statistic that rated Ryan Newman as having the best season in 2003 when his on-track passing was so weak relative to the rest of his profile, but the kind of points system I would want probably would have given Newman the 2003 title. The driver rating I have invented for this column combines both lead shares and a new statistic I introduce for this column called finish shares, where each driver is awarded a percentage of the sum total of points earned by the entire field. For each race in all three series, I awarded the top 75% finishers points where I adjusted x, the percentage of cars beaten in a race, according to the following rubric:

I already came up with a similar system to convert the percentage of cars beaten in a race to a points value in the introduction to my top 100 IndyCar drivers ranking but I have made adjustments to it since. This time I decided to award points for the entire top 75% of a field instead of the top 50% of a field. The reason for doing this is to ensure that basically every driver who competes regularly or semi-regularly should score points in every season, as opposed to for instance Formula One in the distant past when there were some teams that were so uncompetitive they couldn't even come close to scoring points. Using a system like this, pretty much everyone should except maybe a start 'n' park team in NASCAR that probably shouldn't be considered in such an analysis anyway. There will at least be some race where there is enough attrition that the backmarker teams should beat 25% of the field, which allows for fine distinctions to be made for pretty much the entire regular field.

After I have calculated each driver's points total, I then normalize points totals for the entire field by dividing each driver's points total by the sum of all points scored by all drivers. This enables me to calculate finish shares, the percentage of points scored by each driver. Since they are now able to be added up in terms of races so that the total number of finish shares adds up to the total number of races in a season, they can theoretically correspond to lead shares, which also divvy up the entirety of a season's lead change activity between the complete set of drivers in exactly the same way. However, since finish shares are awarded to 75% of the field in every race, while lead shares are awarded to only those who make an on-track lead change, which is always significantly less, lead share totals tend to be consistently lower than finish share totals for the top drivers, and the gaps between lead shares tend to be much higher than the gaps between finish shares as well. This means that finish shares should definitely carry more weight and that makes sense. Finish shares already create a metric that rewards winning and consistency and top finishing in general by awarding the highest positions more. However, lead shares are certainly meaningful too as they indicate drivers' personal impacts on their own leading. After trying out a variety of methods for considering both, I decided that the most accurate driver rating weighted finish shares 80% and lead shares 20%. This came closest to matching my own personal rankings for driver performance in NASCAR. I do think I may need to use different weights for other series such as F1 and IndyCar, because I'm having some doubts about the idea that Josef Newgarden had the best IndyCar season in 2018 but he certainly did from a leading perspective and was unlucky, in a season when eventual champion Scott Dixon strangely did not have any natural races led at all by my definition.

As usual I'm going to use IndyCar as an example here when showing how this points system would work in practice because a 21-car field (a fairly typical field size for recent IndyCar) works well as an example of how points and finish shares would be awarded because the percentage of cars beaten are always multiples of five, since the winner would have beaten 20/20 cars (100%), the second place finisher would have beaten 19/20 cars (95%), and so on, meaning I would be dealing with mostly whole numbers in terms of points in this scenario.

Hence essentially as you can see here the finish shares reflect the theoretical percentage contribution of each driver towards the overall race results, with the winner here getting 22.1%, the podium getting 52.0%, and so on. One might find this schema a little too extreme in that most actual points systems now give the fourth place finisher more than half the winners' points, but in practice this works a lot better at reflecting performance than you think it's going to and usually it does closely match the season championship results even if it gives the top positions a bit more extreme weight than most modern points systems do. I definitely wanted a points system that was based on the number of starters instead of based on the actual finishing positions because I wanted to allow comparisons between different eras within the same series and even different series, and this method greatly enables that.

The one big flaw is that when converting from points to lead shares, there are simply more points available when there are more starters since I am awarding points to the top 75% of the field. This means each position earns fewer finish shares against larger starting fields, but that does make sense when you are awarding each driver's percentage of effect on a finish of a race. This can be adjusted by calculating the field strength of a race as I have done before when I measured each race's field strength based on weighted five-year averages of each driver's performance for Formula One and NASCAR. The deepest NASCAR race ever was the 82-car 1951 Southern 500, which I calculated to be 2.5 times deeper than the average NASCAR race, so multiplying each driver's finish shares by that amount should cancel out this effect. However, I have left the finish shares unadjusted for the time being both because the differences in field size between races are so small in all three series (and nonexistent in Formula One in 2018) that this makes up a minuscule difference. Also, it's an intensive and time-consuming procedure. The finish shares as they stand already measure performance pretty accurately I think without properly adjusting them, although I will eventually likely do this.

By adding the sum of 80% of finish shares and 20% of lead shares, I finally got what I wanted: a system that fairly accurately measured winning and consistency and dominance and the role drivers play in their own dominance all at once. While lead shares by themselves would argue Dale Earnhardt, Jr. was the best driver of 2002 and finish shares by themselves would argue Ryan Newman was the best driver of 2003, the average fan would know better. Junior drops to fourth place in the 2002 rankings behind Tony Stewart, Jeff Gordon, and Kurt Busch, which makes a lot more sense. Sure, that's substantially higher than the eleventh place in points where he actually finished, but it's worth noting that was the most competitive season in Cup history and the differences between any two drivers in points were usually very small, so I think Junior's on-track performance should carry some weight here, just not enough to carry him to first place. Similarly, Ryan Newman drops to fifth in 2003 when considering both finish shares and lead shares, behind Junior, Stewart, Gordon, and Johnson, which also sounds more sensible. When looking at most recent seasons in NASCAR history, I tend to like the results this produces more than almost any actual championship result, however I'll grant it might need some work for other series. What this obviously doesn't do is adjust for the quality of one's equipment, but it's a big start. Undertaking some model to adjust these driver ratings based on team and age in the manner of the earlier F1 models will likely be a long-term goal.

The tables you see below are of the same format as the tables I used for analyzing year-by-year leader tables for NASCAR and for IndyCar. However, I have added a few new categories to the tables this time. In addition to the categories I included on the 2018 tables (races led naturally, lead change record, wins, TNL, races where each driver led the most laps, lead shares, races with the most lead shares, and cumulative races led), I have now added finish shares, driver ratings, and TNL lead change record. The latter reflects each driver's lead change record on the last lead change of the race, what many people would consider perhaps the ultimate measure of clutch performance. As usual, I list the leader(s) in each category in red. Drivers will now be sorted based on driver rating and no longer by lead shares as I did before. If there is a tie in races leading the most laps, as with Denny Hamlin and Clint Bowyer at the Brickyard 400, I credited it to both drivers. However, if there is a tie for the most lead shares in the race, which happened a bit more frequently, I gave it to whichever driver made an on-track pass for the lead later. In terms of percentage-based statistics (lead change record and TNL record) I only included drivers that led five or more races naturally in Cup, three or more in IndyCar, or two or more in F1. I'm well aware oval races tend to have more lead changes than road races, and these adjustments adjust for that nicely. It would be a little bit goofy to award Aric Almirola both the highest lead change record and TNL lead change record when he only led four races naturally all season, but that odd fact certainly says something for the strength of the Stewart-Haas equipment last year.

2018 Monster Energy NASCAR Cup Series season analysis

The standard media narrative for the 2018 season was that Kevin Harvick, Kyle Busch, and Martin Truex, Jr. so dominated the rest of the NASCAR field that they were the "Big Three", but the data seem to indicate that "Big Two" would be more accurate for this season. From a leading perspective, no one was close to Harvick, who dominated most categories, often by large margins, but also clearly had the fastest cars. For the fourth time in five years (all years excluding 2017) Harvick has posted the best lead change percentage in Cup, placing him in a tie with Jimmie Johnson for the most seasons in which he has led in lead change percentage since 1985. He also nearly tied Johnson for the second highest overall lead change percentage among major drivers this year with Harvick now at 56.59% and Johnson dropping to 56.67% even though Harvick entered the year over 1.5% behind. Considering how long Harvick's career has been, this is a particularly massive jump and also noteworthy because he and Johnson are the same age. Harvick continues to thoroughly dominate all his teammates, while Johnson has now been eclipsed by one of his. Harvick also matched Johnson's 2009 in becoming only the second driver ever to lead in races led naturally, lead change record, wins, TNL, lead shares, races with the most lead shares, and cumulative races led all in one year. However, I have to say Johnson's advantage in 2009 was a lot bigger than Harvick's this year. Regardless, even though Harvick did not score the most points either using the chase system or a full-season system, he was significantly more dominant than Busch while Busch's consistency advantage was nowhere near as wide, so although Busch led in full-season points and also in finish shares, Harvick is easily the highest rated driver of the season.

But just because Harvick had the highest rating does not necessarily mean he was the best driver if I adjust for equipment strength. In fact, I would tend to argue that Busch was in fact the more impressive because I think his cars were a lot slower than Harvick's. What can be said for both of them is that they thoroughly dominated their teammates. While Harvick himself had 20 races led naturally and a lead change record of 44-18, his three teammates combined for 19 races led naturally and a lead change record of 27-25. While Busch had 13 races led naturally and a lead change record of 24-21, his teammates combined for 10 races led naturally and a lead change record of 15-21. Similarly, if you aggregate almost any of these statistics, both Busch and Harvick outscored the entire rest of their teams, and that is pretty rare. However, I think I like Busch's season better not just because he was more consistent but because he also matched Harvick in wins despite what seemed to be a clear equipment deficit. For that matter, Busch was also more clutch as he was 7-2 in the final lead change of the race, while Harvick was 9-3. While on the surface Harvick's season looks flashier, it seems Busch was elevating himself more particularly at the ends of races (even though he was slower in the first half of the race most of the time.) Furthermore, he did not manage to get outdueled by his inferior teammate like Harvick was when Clint Bowyer beat him in the rain-shortened race at Michigan. Indeed, Bowyer won twice after an extremely long winless streak while Kyle Busch's lead teammate Denny Hamlin went winless for the first season in his career, and most people would argue Hamlin has been better than Bowyer for their entire careers. All of these factors lead me to conclude that Busch was the better driver when adjusting for equipment, even if Harvick did have the best season on the surface.

As for the third driver in the "Big Three", I do not think Truex was close to either Harvick and Busch, even though he looks close to Busch on the surface. As you can see below, both Truex and Busch had the same number of races led naturally (13) and nearly identical lead change records (24-21 for Busch and 23-20 for Truex) and Furniture Row Racing was often considered by many to be the fifth Joe Gibbs car. I now think this is inaccurate actually and I think the team was better than JGR. What leads me to this conclusion is the differences in their battles for the wins. While Kyle Busch had a 7-2 lead change record on the final lead change of the race, the drivers he passed in those races were William Byron, Chase Elliott, Erik Jones twice, Brad Keselowski, Kyle Larson, and Joey Logano. Busch passed Logano in the Coca-Cola 600 in a period when the Penske cars were much slower than they were later in the season, and Keselowski at Richmond in a possible contender for the best race in Busch's career. The only two drivers to pass Busch were Harvick and Truex. This implies strongly to me that both of them had faster cars than him. I actually think Truex was a disappointment when you consider how unclutch he was in the races. Harvick passed him for the TNL in the spring Kansas race; nothing wrong there. But the other three times he were passed were all fairly embarrassing. In the Coke Zero 400, Truex and Erik Jones were the only major-team drivers left after basically all the other relevant drivers in the field had already crashed, and yet Truex somehow let himself get passed when he had next to no competition left, extending his career winlessness on plate tracks another year. Then he decides to clear Logano at Martinsville, which allowed Logano to bump him out of the way in the final corner (when he probably would have had a guaranteed win had he stayed side-by-side with him.) Then he threatened to spin out Logano at Homestead in response to that and just let him go for the championship. While it's admirable that he didn't cause an intentional wreck for the championship, it seemed to show he's becoming a driver that all the other drivers know they can bully around, which may be part of the reason (beyond just absurd bad luck) that he hasn't won a short track race yet. I realize Truex's cars did seem to get slower later in the season especially after it was announced that Furniture Row was shutting down, but I still think he actually had a fairly disappointing season for the cars he had. Overall, I think his performance was probably on par with Harvick and Busch's during the first 2/3 of a race, but his unclutch performance and lack of consistency late means he really wasn't on par with them.

Okay, I admit it. I have egg on my face. Less than a year after I close a column with the snarkier-than-I-wish-it-had-been wham line "Mark Martin may have thought Logano could become one of the greatest drivers ever in NASCAR, but he was wrong", not to mention calling him the most overrated driver of the lat 30 years, he goes on to win the championship. This is not the only time I had egg on my face this season. However, there really isn't a case that his performance was on par with Harvick, Busch, or Truex even though he lit it up late. It's not even clear to me he outperformed Brad Keselowski. Logano was as usual more consistent than Keselowski, but Keselowski had a quite underrated season with 3.6 lead shares to 2.4 cumulative races led, and Keselowski ended up being closer to Truex in lead shares than he was to Logano, even though Keselowski never even led the most laps in a race all season, which means he was probably the most clutch driver in the field outside of Harvick and Busch. He also took the lead on track in three more races than Logano did. According to my new ratings system, Keselowski actually outperformed him, although I don't know if I personally agree with that. I would probably take Logano's season simply because he elevated himself most at the end, and did have to outduel the three more dominant title contenders in the final race, which he managed in aplomb. While I may have been wrong to say Logano wouldn't go on to become one of the greatest drivers, I certainly don't think he's stopped being overrated now that he has turned a fourth-best season (at best, and there are strong arguments Brad Keselowski, Chase Elliott, and Kyle Larson outperformed him when if you take into account equipment strength) into a championship.

Honestly, when you compare drivers to the level of their equipment, I was more impressed with both Chase Elliott and Kyle Larson than either Truex, Logano, or Keselowski. Both of them dominated their teammates (and even all the other Chevies) to as great a degree as Harvick was dominating the other Fords or Busch was dominating the other Toyotas. Just how badly did Elliott dominate the other Hendrick drivers? He led eight races naturally while his three Hendrick teammates combined only led on plate tracks. Elliott did so on an extremely diverse set of racetracks also: besides the two Daytona races, he also led Loudon, Pocono, Watkins Glen, Bristol, Richmond, and Phoenix naturally. All those tracks with the exception of Daytona and Pocono are places where horsepower is not as important (and it's clear Hendrick did have really fast cars on the plate tracks even if they did nowhere else) so he was clearly elevating himself almost to Kyle Busch levels. While he arguably lucked into all three races he won (he was not the TNL in any of them), that merely more or less makes up for his previous seasons when he collected two TNLs without winning, and while he may not have deserved any of the specific races he won, he clearly did deserve several wins for how relevant he was on every non-horsepower track in the second half of the season. He'll be particularly scary once Hendrick actually gets a good superspeedway package again. It's hard to say how much it is due to Elliott's ascent vs. Johnson's decline, but his utter domination of Jimmie Johnson this year is something heretofore unforeseen as Johnson had never even had close to a bad season until now, and particularly striking Johnson is still the same age as Harvick and has not suffered a major injury. I would probably rate his season third best, when adjusting for equipment.

Kyle Larson likewise was competing for best in class with Elliott quite often but the major difference is that Larson seemed to have a lot more speed on the intermediate superspeedways as three of his races led were at Chicagoland, Las Vegas, and Homestead, which is not counting the Kansas race where he actually at one point had all the other Chevies a lap down before it was revealed his car was illegal after the race, so I had to throw those passes out. That's why I'm a bit less impressed with Larson's season than Elliott's, even though Larson did tend to beat Elliott in most of the lead change statistics. A lot of Larson's dominance came on superspeedways while most of Elliott's came on short tracks, road courses, and one-mile ovals and Elliott tended to be out to lunch on the superspeedways in comparison (which bizarrely is an almost exact inverse of Chase's father's reputation.) This is not to say Larson was not versatile in his own right. Four of his races led were Phoenix, the two Bristol races, and the Charlotte roval, and he was extremely unlucky not to win this year. He had the most lead shares and cumulative races led of a non-winner since Jeff Gordon in 2010, and no driver in my lifetime has led the most laps in four races in a season without winning until Larson in 2018. It was hard not to feel for Larson at the Southern 500 where he was responsible for all three of the on-track lead changes in the race and then got beaten out of the pits by Keselowski, who ended up winning. As with Elliott, he'll certainly improve a lot when he has faster cars, as both of them dominated the other Chevies by a long shot, with none of their teammates even coming close. I'd definitely rate his season ahead of Truex's, Keselowski's, and Logano's when adjusting for equipment too, but Elliott seemed to do just as much with slower cars and was more clutch in the end.

Besides those six drivers, only six others were even remotely competitive throughout the season, as the twelve competitive drivers pretty much dominated all categories amongst themselves, with none of the other drivers really coming close in any of them. Although Denny Hamlin's season got an awful lot of criticism, I actually think Kurt Busch's was worse. His 9-16 lead change record is incredibly disappointing when you consider that Harvick had barely more lead changes against than that and five times as many lead changes for, not to mention that he's older than Kurt Busch. Ten years ago, I was pretty certain that I thought Kurt was a better driver than Harvick, but now I don't think he was ever close. Kurt Busch seems to have been overrated all along (I suppose it's telling that he lost the truck title to Greg Biffle and then Harvick beat Biffle to win the Busch title the very next year...) The only time Kurt really looked great were his 2002-2005 seasons when he was with Roush at their actual peak and had Jimmy Fennig as his crew chief, who gave all seven of his full-time drivers from 1989-2009 their best career points finishes: Dick Trickle, Jimmy Spencer, Hut Stricklin, Derrike Cope, Mark Martin, Kurt Busch, and David Ragan. I'm not going as far to say that Kurt Busch was a Fennig creation, but I really think he was essentially a Ricky Rudd/Terry Labonte-style driver for his entire career who just got lucky to have a really strong crew chief and the strongest cars in the field at his peak. It's also telling that Clint Bowyer, who is also younger than Busch, won more often even outdueling Harvick once and posting a much better lead change record despite a much longer winless streak. While Kurt Busch is still better than Daniel Suárez, it's still kind of understanding he's the driver they let go when you consider this. While Busch certainly outperformed Aric Almirola at least, it was a lot closer than you'd think. Busch was certainly good at collecting consistent top tens while Almirola was feast and famine, either fighting for the win or uncompetitive for the most part, Almirola's dueling record of 6-1 and 2 TNL is really amazing proving that he was elevating himself at the end of any races where he was competitive, but he wasn't competitive in enough races to take him over his teammates. He still led half as many races naturally as Busch did though and posted a way better lead change record, so I actually do think they were close, but it's more that I found Kurt to have probably the most disappointing season for any star driver (except for Jimmie Johnson) than my being impressed with Aric's. What I found especially odd is that Hamlin's season seemed to get more criticism than Kurt's even though their overall profiles here seem to look pretty similar, and everyone knows the SHR cars were faster than the JGR cars this year. If Hamlin's year was disappointing, Kurt's was even more. Besides those eleven drivers, Ryan Blaney was the only driver to be competitive at all, but I don't have any particularly interesting insights about his year, other than he won a race that he shouldn't have won but lost a couple of races he should have. He was nowhere near as unlucky as Larson though.

Besides the ten relevant SHR, JGR, Penske, and Furniture Row drivers, and Elliott and Larson, who made their teams relevant almost single-handedly, almost nobody else even factored except on restrictor plate tracks. The few exceptions were Ricky Stenhouse, who passed Clint Bowyer at Dover, Ryan Newman, who shockingly passed Kyle Busch at Atlanta (perhaps another sign that maybe Busch's cars weren't as fast as what he was doing in them?), and in the upset of the year, Darrell Wallace, Jr. passing Brad Keselowski at Bristol, while his fellow rookie in the superior car William Byron only led naturally on plate tracks. It's a pretty sad season when only twelve drivers are responsible for all the on-track lead changes except for three outside plate tracks. However, it's certainly possible that the new 2019 rules package will compress the field on the intermediates again, which could help drivers like Larson and Elliott and probably hurt drivers like Truex who are great duelists by themselves but somewhat struggle in a pack.

2018 Verizon IndyCar Series season analysis

Just as with Joey Logano winning the title in a year I sharply criticized him, I also have to acknowledge the egg on my face here as well as I previously predicted that Scott Dixon would go winless in 2018 and then he went on to win the championship. But despite whiffing that badly on the surface, in a way I was on to something. Despite Dixon's championship, he actually never made an on-track pass for the lead becoming the first champion to do that in an IndyCar season since at least 1973 (Roger McCluskey may have won the championship without an on-track pass for the lead, but I have several races missing for that season, so I cannot confirm that for sure.) For Dixon personally, it is the first time he has failed to lead a race naturally since 2002. Even in 2004 and 2005, when Dixon had very slow lame-duck Toyota engines, he still made on-track passes for the lead. Although he still went on to win the championship, it is hard not to think that he is also simultaneously in decline. Dixon is still a master at passing people in the pits, as he proved at Texas, when he beat Robert Wickens on a pit cycle exchange, and at Detroit, where he did likewise against surprise polesitter Marco Andretti. He also inherited the lead when polesitter Josef Newgarden hit the wall on a restart at Toronto and won the pole at Gateway and dominated most of the race, but made no on-track pass there either. To Dixon's credit, I really don't think the Ganassi team is as good as it used to be, particularly after they lost Target sponsorship (I would guess NTT Data is not paying as much), and it's telling that Ed Jones didn't really improve on his rookie season driving for Dale Coyne while a sophomore driving for Ganassi (as Jones basically was a complete nonfactor.) Perhaps Dixon is not in decline and it's only that the Ganassi team is, but it's definitely weird that he failed to make an on-track pass for the lead even when he did in other years when he had theoretically slower cars.

Early in the season, after Power had crashed in three of the last seven races, while Newgarden had at one point won 5 of the last 10 races, I was actually starting to doubt Power's future at Penske, because I thought he had a chance of being "triangulated" with Newgarden beating him on dominance/on-track passing and Simon Pagenaud beating him on consistency. I did say that I thought Power might be in trouble if he didn't win one of the next two Indy 500s, and he did. Now he will certainly be with Penske for the rest of his career, but as it turns out, even Marshall Pruett reported in some articles that he was possibly close to being on the hot seat before that, so my argument may not have been unfounded. Power certainly redeemed himself winning three races including an Indy sweep to more or less complete his career, but even though he won the Indy 500, matched Newgarden in wins and TNL lead change record, and beat Newgarden in points, I still think Newgarden's the team leader. In fact, I don't think it's very close.

While the vets may have stolen most of the headlines with Dixon winning the title, Power winning the Indy 500, and Ryan Hunter-Reay winning the double points final race, a different and much younger triumvirate combined to deliver the vast majority of the excitement in 2018. If you look at his statline (particularly his lack of consistency), Newgarden was especially disappointing compared to his 2017, but I think he is still performing at just as elite a level as he was then; he was just a bit lucky in '17 and absurdly unlucky in 2018. While the Toronto wall contact was his fault, most of the other races he lost were due to sheer dumb luck. If you counted the two double-points races only once, he would have finished third in points behind only Dixon and Alexander Rossi, for one thing, but that was not all. At the Indy 500, he was actually leading the three Penske cars and running in 3rd place when Tim Cindric decided to take him off pit sequence under caution for reasons that were not entirely clear. Subsequently, Tony Kanaan cut a tire and Power beat Ed Carpenter out of the pits. Newgarden could have easily been in the position to win (and collect double points, and extend his points lead) had Cindric not made that call. Newgarden made two daring passes for the lead on Rossi at Mid-Ohio and Portland, but he never actually led at Mid-Ohio because he pitted later that lap after taking the lead (I did decide to count it as a lead change and TNL for Newgarden, while Rossi ended up winning because he pitted twice while everybody else including Newgarden pitted three times.) At Portland, the caution during the middle of a green-flag pit cycle destroyed the hopes of all the serious contenders and allowed Takuma Sato to back into the win (although I still have to give Sato props for that, since he, not his team, called the strategy, and that is fairly impressive.) At Iowa, although he just got beat by James Hinchcliffe (and I was very surprised to see Newgarden, the best duelist in IndyCar, get snookered at arguably his best track) he lost 2nd place because he pitted under caution to get fresh tires and the race never restarted. Combining all the points he lost in those races, I daresay he would have been pretty close to Dixon otherwise. What was a mediocre season otherwise actually looks like a dominant one if you look at it slightly differently. He was certainly lethal at Barber and Road America, where he led the entire races start to finish without once being passed, and although he was somewhat lucky at Phoenix to gain a lot of spotss in the pits, he still had to outduel a feisty Wickens to win in a race that had almost no passing. It's quite telling that for the fourth year in a row Newgarden has now led in lead change percentage, which nobody back to 1930 has done in IndyCar history, and even more so that he managed to do so despite tying Power for the most poles in 2018. Normally we see drivers that win a lot of poles tend to have low lead change records (as I have observed throughout Hélio Castroneves's entire career and Ryan Newman's early career, and also for instance Kurt Busch this year) but not in Newgarden's case. Despite winning four poles, he was not once passed at Barber, Road America, or Toronto (unless you think him hitting the wall and giving up the lead to Dixon should count, which I have not been doing.) He was only passed by teammate Simon Pagenaud at Texas and by Hinchcliffe for the win at Iowa, admittedly in both cases those drivers' only passes for the entire year. Although Newgarden lost to Power in the championship, the first time he has ever been beaten by a teammate in his career, he still pretty clearly outperformed him. He factored in lots of races he didn't win, while the only races Power factored in to the same degree were the races he won.

While Josef Newgarden may be the most exciting driver while battling for the lead, Alexander Rossi is clearly the most exciting driver to watch while battling in mid-pack. Rossi at his most dominant was every bit as dominant as Newgarden, as he was unchallenged at Long Beach and Pocono to the same degree that Newgarden was unchallenged at Barber and Road America. He is probably more polished and well-rounded in his skills than Newgarden is as well, as he seems to be better able to adapt between different driving styles than Newgarden is. At Phoenix, he fell a lap down due to a pit penalty and nearly unlapped the entire field without the aid of yellow by himself, and at Indianapolis, he drove from the last row to 4th place in a race that had minimal passing, but at Mid-Ohio and Gateway, he did the exact opposite, intentionally slowing down (especially at Gateway) to save fuel. He seems to have more tools in his toolkit than any other driver as he is able to switch from aggressive to conservative on the drop of a hat in a way no other current IndyCar driver is (Dixon and Pagenaud are at this point entirely conservative, while Newgarden seems to rely mostly on his speed and passing ability and hasn't had many great "strategic" races). I think Rossi was greatly aided by the strength of the Andretti equipment this year, as I think Andretti definitely had faster cars than Penske and Ganassi. Marco won his first pole in years, Hunter-Reay broke a three-year winless streak, and rookie Zach Veach looked to be about 5th-7th on the ovals in speed late in the season. Considering Newgarden made two on-track passes of Rossi while Rossi never passed him, and considering I think the Penske cars were slower than the Andretti cars (especially if you take a look at Simon Pagenaud's season... despite winning the title two years earlier he was only 13th in cumulative races led, behind even the likes of Marco and Ed Carpenter and Sato) I think Newgarden was more impressive. Another difference is that Newgarden tends to pull off his aggressive moves while Rossi tends to make contact with other cars when he tries to do similar things, most notably when he wrecked Robert Wickens on the final restart in the season-opener at St. Petersburg. Rossi was definitely very skilled at leading by himself, and although Newgarden led most other lead change statistics, Rossi led the most laps in the most races and also had the most cumulative races led (barely, although Newgarden had more laps led, as the Iowa race Newgarden dominated had more laps than the Pocono race where Rossi dominated.) Regardless, while Dixon and Power stole the headlines, I think Newgarden and Rossi brought the substance and are better-suited to dominate in years to come, while to some degree it felt like a last hurrah for both Dixon and Power, but I could be wrong. After all, I was last year.

Poor Robert Wickens. Even though he had one of the most talked-about rookie seasons in IndyCar in many years, it was still underrated and better than you think. Despite missing the last three races of the season due to his crash at Pocono where he suffered paralysis in both of his legs, he still led four races naturally, which was only second to Newgarden, and ahead of Rossi, Dixon, or Power. He did this despite being a rookie and driving for a team that is usually considered second-tier relative to the Penske, Ganassi, and Andretti teams. No Schmidt driver ever led four races naturally in a year before; indeed, future champion Simon Pagenaud only led three races led naturally combined in his three full seasons for the team, less than Wickens did in one partial rookie season. No rookie had led four races naturally in a year since Juan Pablo Montoya in 1999, who was driving for a Ganassi team that had won three consecutive championships. Although his teammate Hinchcliffe got a win and he did not, Wickens regularly outran him despite Hinch having much more vast experience. He earned the pole in his first race at St. Petersburg, becoming the first driver to do so since Sébastien Bourdais in 2003, and his 2nd at Phoenix in his first oval start was the best oval debut since Bourdais won his that same year. However, despite how fast and polished he was from the beginning, he was extremely unlucky in that he failed to win despite being the TNL at St. Petersburg, where he dominated before Rossi wrecked him on the final restart, and at Texas, where Dixon beat him out of the pits shortly before he eventually crashed. If he is unable to return to IndyCar, I think I might already pick him as the best-ever IndyCar driver to never win a race. Obviously I most hope he can manage to walk again more than anything else. That is clearly more important than whether he ever races again, and he was fortunate enough to survive the wreck. However, it's unfair by any measure that he already should have had a win or two and now may never get the chance.

The biggest disappointment of the year was clearly Simon Pagenaud. In a year when his teammates Newgarden and Power both tied for the most poles and the most wins, and Newgarden dominated most metrics by a leading perspective, while Power won glory at Indianapolis, Pagenaud was entirely invisible all season long. When you consider he also went winless in 2015 and led zero races naturally in 2017 (winning both of his races on pit strategy), his Penske career has been entirely disappointing except for his championship season, and now if any Penske driver is on the hot seat, it certainly would be him. The one big difference between his championship season and his others is that he was winning a ton of poles in 2016 and he wasn't so much the other years. When he wins the pole on a road course, he's fast enough to stay there but he seems to be too conservative to come from behind. His conservatism does pay dividends sometimes - he is probably the least crash-prone driver in the series and if you add his raw speed to his crash avoidance, he can be difficult to eliminate. However, when both of his teammates are regularly out-qualifying him and outdueling him, he has to rely only on strategy to win, and it wasn't there for him in 2018; indeed, Scott Dixon has always been best known for his strategy and lately, Alexander Rossi has surpassed him as a strategic driver in addition to surpassing him as a duelist. While I seemed to have been right that one of the Penske drivers would be "triangulated", it ended up being Pagenaud, not Power. For a Penske driver to finish tied for 9th in lead shares with Jordan King, a part-time rookie now demoted to Indy only, and 13th in CRL, is fairly astonishing.

After Newgarden and Pagenaud both overachieved with smaller teams, I thought it would be interesting to see who would come on top when they became Penske teammates, as they clearly had different styles even in earlier years. Pagenaud earned three straight top five points finishes and four wins for the Schmidt team in 2012-14, while Newgarden earned only one top five points finish and three wins for the Ed Carpenter team in 2015-16, which were considered roughly the same caliber. However, as earlier mentioned, Pagenaud relied on not making mistakes and getting good finishes more than domination, as he only led three races naturally. Newgarden meanwhile posted the best lead change record in both of his Carpenter seasons, led a combined six races, and actually scored more laps led than any other driver in 2015 (although Power and Dixon beat him in CRL), when Pagenaud never came close to doing that for Schmidt. I had no idea whether consistency or domination would end up coming on top, because Newgarden definitely takes more chances than Pagenaud, but also makes more mistakes. My instinct was that it's easier to overachieve in a second-tier car in terms of consistency than it is in terms of domination. You can rely on pit strategy and attrition to carry you far as long as you have a certain level of speed, but passing for the lead takes more. These two teammates who have become almost polar opposites in styles do mark one of the best examples of this I can think of. Now that Newgarden has won 7 races to Pagenaud's 2, and led 11 races naturally to Pagenaud's 1 in the time period they have been teammates, this seems to provide strong evidence that leading more often than you should for a second-tier team is a better predictor of eventual success than finishing better than you should. Pagenaud is totally capable of dominance - when he wins the pole on a road course. However, when Newgarden and Power are both usually outqualifying him on them, it's a lot harder for him.

2018 Formula One season analysis

Even though I have now collected all the lead change data I could for Formula One back to 1967, I have to say it isn't usually as interesting (particularly lately) as the NASCAR or IndyCar data is. Since most Formula One races do not have on-track lead changes, and since most passes generally are done in the pits, qualifying and strategy become more important and passing somewhat less. To give a weight of the disparity in this year, there were 226 lead changes in NASCAR (6.28 per race), 20 in IndyCar (1.18 per race), and 8 in Formula One (0.38 per race.) It's almost too small to draw any conclusions for this season, although a lot of earlier seasons particularly in the '70s and '80s are a lot more interesting and more like present-day IndyCar seasons (admittedly, that was when F1 was probably at its closest to spec, when most drivers and teams had the Ford Cosworth DFV engine. Furthermore, there were only three teams: Mercedes, Ferrari, and Red Bull, that even sniffed the lead all season, and you're never going to get some fluke like Jordan King taking the lead on his debut from Robert Wickens also on his debut, or somebody in the equivalent of the equipment Darrell Wallace, Jr. had taking the lead on track (even in a rain race.)

Since qualifying is probably one of the biggest predictors of race results in Formula One, much more than it is in oval racing, there tends to be a trend in F1 where a lot of the dominant drivers end up actually having negative lead change records in many seasons simply because their dominance was also reflected in qualifying. Despite running away with the championship and winning eleven races, Lewis Hamilton had only a 1-4 lead change record while Sebastian Vettel had a 3-0 lead change record. Does that really make a case that Vettel outperformed him, as I'd be more inclined to argue in NASCAR or IndyCar? No, not really. It just means that Hamilton was faster in qualifying, while possibly Ferrari was faster in the races than Mercedes was, especially considering Valtteri Bottas went winless while Kimi Raikkönen ended a long winless streak by passing Hamilton at the start, no less. While Hamilton was passed more than anybody else just because he won most of the poles, it's hard to argue he didn't outperform Vettel, especially because he beat Bottas by a far worse margin than Vettel beat Raikkönen (and many might argue especially based on 2017 that Bottas is better than recent Raikkönen.)

Vettel was a bit unlucky at Baku where he lost the lead to Bottas on pit strategy, felt like he had to pit with him on a safety car period, and then overshot a corner trying to make a pass, although that demonstrates the flaw in Vettel's season, as a lot of Vettel's bad luck was self-inflicted. That was one of the two races where Vettel was TNL but Hamilton won, and the other was at Hockenheim, where Hamilton inherited the lead when Vettel crashed. However, in a way one can say Vettel was luckier than Hamilton because he also had a race where he won and Hamilton was TNL after he beat Hamilton out of the pits. In the two races Vettel was TNL and Hamilton won, Vettel made mistakes, while in the race Hamilton was TNL and Vettel won, Hamilton did not make a mistake, so even though they ended up closer in terms of TNL than in terms of wins, it's hard to make any case Vettel really outperformed Hamilton.

Hamilton did certainly have good fortune elsewhere as he won at Interlagos after leader Max Verstappen, who had passed him earlier in the race, made contact with the lapped car of Esteban Ocon, and at Sochi, where he was given the win on team orders. But ultimately, the championship difference was so massive that even if all those factors were combined, nothing would have likely changed. Since the Mercedes drivers had lower lead change records and TNL records than the Ferrari drivers, there's a case to be made the cars were slower in race conditions, which would only make Hamilton's season more impressive rather than less. Although one could argue that Bottas probably should have won at Sochi, this was the first season that a driver won the championship while his teammate went winless since Vettel in 2013, which is actually a bit more recently than I expected.

The Red Bull drivers were a bit more evenly matched, although it seems that Max Verstappen had the edge in terms of both consistency and on-track passing, although Daniel Ricciardo's luck was certainly absurdly poor. Verstappen ended up with 1/3 more lead share than Ricciardo because one of the races Ricciardo led he was the second driver to take the lead, while Verstappen was the only driver to lead naturally in both of his. Additionally, Verstappen's lead change came against Hamilton while Ricciardo's came against Bottas. With Vettel, Raikkönen, and Verstappen all passing Hamilton and Ricciardo not, that further makes the case for Verstappen, but it's worth noting that Ricciardo did outscore Verstappen in points in the races both of them finished.

Doing a lead change analysis really only matters for the three teams that are capable of leading, so to analyze the rest of the field, points probably would be the better bet, although there were some slight deviations between points and the finish shares statistic I invented. Only three drivers moved more than one position away from their championship finish when considering finish shares versus the actual championship points. Kimi Raikkönen fell two positions from 3rd to 5th (but he, Verstappen, and Bottas were all virtually tied in the points, separated by a mere four points in the actual standings), while Carlos Sainz, Jr. gained two positions and Nico Hülkenberg. This probably merely means that drivers who retired a lot are hurt more in my system than the actual points system, which makes sense since I am awarding points to the top 75% of the field and not the top 50% of the field as F1 does. In this situation, you'd probably just be better off comparing point totals for the marginal teams than using finish shares. I still think finish shares are valuable because they can break down the percentage of value each driver contributed to the race, and they can be compared across series, but there are probably other series where it would work better than F1, particularly because F1 awards points to a smaller percentage of the field than I do.

Conclusions

While I do think analyzing lead changes has value in almost any racing series, as I'm more impressed with drivers who are able to make passes for the lead than those who don't, I think this particular kind of analysis has more value when there is a larger volume of lead changes, which tends to happen much more often in oval racing. I think passing is a strong measure of oval performance but it's admittedly probably not as vital in assessing road racing performance, particularly in series like Formula One where the polesitter is more likely to be the favorite to win. Having already calculated lead change totals for F1 drivers over the last fifty years, many of the drivers considered among the best all time have negative lead change records, while some of the drivers who have very strong lead change records are kind of odd and surprising (such as Riccardo Patrese.) Obviously, qualifying matters more than passing in Formula One and perhaps in most road racing series, so while I am going to continue the analysis there, I am not probably going to put as much emphasis on it in evaluating talent as I would for NASCAR and IndyCar, where there are a lot more field reshuffles to bring the field closer, and a lot more parity in the field, which results in more passing. Drivers who make lots of passes in F1 are much more likely to be drivers who qualify poorly than they might be in oval racing, and qualifying poorly can be associated with not being as fast. I think it's something to consider for sure, as overtakes excite F1 fans just as they do everywhere else, but they're probably not as essential as plenty of great drivers focus on overtaking in the pits. This is why Lewis Hamilton is already ahead of Michael Schumacher in races led naturally: because Schumacher took the lead in the pits so often. Does this mean Hamilton is necessary better than Schumacher at this point? I don't know.

As for my new ratings system, I think I like it very much for NASCAR but I'm not sure how well it works for other series. The F1 results seemed pretty fine as well but the IndyCar results seemed like they may have been very messed up. While I do think Newgarden and Rossi brought more excitement than any other drivers to IndyCar in 2018 (along with Wickens to a lesser degree), it's possible I may need to adjust the weights for that series so Dixon's consistency is worth even more, because while I do think Newgarden had one hell of an underrated season, I'm not sure it was that underrated. There's also the debate about what passes should count. Many people would not count Newgarden's pass of Rossi at Mid-Ohio because he never actually led the race at the line even though it was an impressive pass; in that case Rossi would be the TNL and would take the lead in driver rating. If I did not count that Newgarden pass, and also counted Newgarden hitting the wall as a pass for Dixon, that would make Wickens, not Newgarden, the top duelist of the year. This is definitely not a science, but that doesn't mean it's not worth attempting to evaluate things that many other analysts have not.

While I don't think it's all that effective in evaluating F1 talent today, lead change analysis is likely very effective in evaluating F1 dueling talent in the past. Drivers who fans rate highly and models do not like Gilles Villeneuve and Nigel Mansell shine much more when you consider their on-track passing records. Overtakes are what the fans like to see and no doubt that is why drivers like that are overrated relative to what computer models come up with, because taking more passing chances also means you're usually more likely to make mistakes. I do think a composite model that considers both consistency and dueling ability, both driving and racing ability, would be an improvement over past models that considered nothing other than finishing positions, but there's obviously a ton of subjectivity here in determining which passes count and which do not, and for F1 particularly this will be hard since there is so little video content available of past races. Eventually, which might be a year or two from now, since I am primarily focused on writing a typing book more than this, I think I'm going to try to use my driver rating model and adjust it for equipment strength to do an all-time ranking for each series similar to what F1metrics did for Formula One alone, which could be the start of an ambitious all-time ranking project, although any list I would make would ultimately be a subjective one, that uses various objective statistics, but is not bound by them. And I'd definitely focus on longevity more, which was my big mistake on the IndyCar ranking list. However, all of that would be some time in the future. Over the next months, I'll probably be periodically publishing decade reviews of IndyCar and F1 lead changes in the past. I don't think I can go much further back in the past for NASCAR, but for those other two series, I can go back many decades, and there will be definitely many interesting results.

Most of all, I'm going to try to do better with regard to sticking my foot in mouth with poor bombastic predictions. After predicting Scott Dixon would go winless and then he won the title, declaring Joey Logano the most overrated driver of the last thirty years and then he won the title, and suggesting Will Power might be on the hotseat at Penske immediately before the Indy 500, I did not have a good batting record this year. I'm going to try to stick to the stats more and the punditry less. Few people like a pundit.

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.

---