Winning vs. Consistency: The Delicate Balance of the NASCAR Playoffs
How do you decide who the ‘best’ driver is? Is it the guy who wins the most races? Or the guy who runs up front the most? If it’s s combination, how do you weight each factor? And once you’ve decided that, how do you structure the playoffs to ensure that’s who actually wins?
NASCAR has modified their playoffs with the intent of creating a balance between winning and consistency. Have they succeeded?
Birth and Evolution of The Playoffs
The NASCAR championship points system has undergone 15 changes since 1949 (or so Wikipedia tells me.) Five of those changes have occurred since the institution of a playoff-type system in 2004.
Each iteration tried to make individual races more exciting by rewarding winning, but without diminishing the importance of consistency. You don’t want your champion to have ten wins and twenty-six finishes out of the top 30, right? On the other hand, you also don’t want drivers feeling like they have to play it safe so they have a “good points day”.
Has NASCAR Always Rewarded Winning?
In 2003, Matt Kenseth won one race and the championship. Ryan Newman won eight races, but seven DNFs left him in sixth place at the end of the year. But Kenseth’s championship wasn’t all that unusual.
The blue bars below show the number of wins the champion had each year from 1991 to 2017. The red bar appears only if someone other than the champion won the most races that year.
The green asterisks signify cases in which multiple drivers had the same number of wins as the champion. In addition to the 2003 season:
- In 1992, Alan Kulwicki won the Championship with two wins; Bill Elliott and Davey Allison each had five wins and finished second and third, respectively.
- In 1993, Dale Earnhardt was the Champion with 6 wins. Second-place finisher Rusty Wallace had 10 wins
- In 1996, Terry Labonte won with two wins, while Jeff Gordon, with ten wins, was second
Now look at the record after the 2004 introduction of the playoff system. Things didn’t look so promising in 2004, where
- Kurt Busch won the championship with three wins
- Jimmie Johnson (2nd) had 8 wins (+5)Dale Earnhardt, Jr. (5th) had 6 wins (+3)Jeff Gordon (3rd) had 5 wins. (+2)
I’m sure NASCAR was second-guessing themselves during the Winter of 2014–2015. But the driver with the most wins was the champion in:
- three out of the last four years
- 2015 was a difference of +1 win
- five out of the last seven years
- 2015 and 2013 were both differences of +1 win
- eight out of the last twelve years
- Two years had a difference of +2 winsTwo years had a difference of +1 win
There’s a clear shift to rewarding winning over the last decade and a half.
But What About Last Week?
The Kansas race eliminated four more drivers from championship contention. The graph below shows the number of wins after 32 races for the top 16 drivers.
The number of wins scales with the current standings — with one sort of big exception.
Regular readers know that drivers who get into the playoffs with a single win at either a restrictor-plate track or a road course don’t survive long in the playoffs.) Although Almirola is in eighth place, remember that he made this cut by winning at Talladega — a plate track.
- The four drivers with no wins are all out.
- Two of the three drivers who got into the playoffs with a single win at a restrictor plate track are out.
- Dillon and Jones are out
- Logano is still in.
- The one driver who got into the playoffs with a single win at a road course (Blaney, and the Charlotte Road Course) is out.
But Brad Keselowski has three wins and he’s also out. So how can you say the playoff system rewards wins when Logano, Kurt Busch and Almirola are in with one win and Keselowski is out?
I initially thought that this was a nasty consequence of the three-race elimination format, which I admit to having mixed feelings about. All you need is a mechanical failure or someone else wrecking you and you’re in jeopardy. Granted, Keselowski didn’t have a great three races (finishes of 14th, 27th and 6th), but he didn’t experience any mechanical failures or crashes, either.
So is his eliminate really reflective of how he’s run this year relative to other drivers?
Keselowski’s Year at a Glance
If we extend our graph to include top-5 finishes, Keselowski still looks like he should be in the round of eight. I’ve colored wins red and finishes in places 2–5 orangey.
He’s not up there with the “Big Three”, but Keselowski looked like he was a contender when he racked up three wins in a row, two before the playoffs started and one after. Unfortunately, that span is the highlight of an otherwise uneven season.
The graph below shows his finishes for all races thus far. Longer bars are worse finishes. And there are a lot of longer bars.
Just to give you an idea of how that compares to someone else’s year, let’s look at the same graph for Kyle Busch
You can really see the difference in terms of number of short bars. While Kyle Busch has three finishes out of the top 30, Keselowski has seven — the most out of any of the top 16.
And if we look at finishes out of the top-20, Keselowski has the most (9) out of anyone who made it into the Round of 12. Dillon (11), Johnson (10) and Bowman (8) were all eliminated in the last round of cuts.
It’s Not Just Top 5’s and 10's…
I’ve shown you a couple times (including a blog on the average age of the starting field at Daytona) that getting a complete picture of anything requires that you look at more than a single number or two. Distributions give us a much better picture than a simple number like an average, or the number of top-5s or wins.
A graph with a single peak would mean that the driver finishes in that range every race. A graph with a flat distribution would mean that the driver is equally likely to finish in those positions in a race.
A histogram shows the distribution of finishes in each range from 1 to 40. Here’s the histogram for Keselowski. He has three wins, seven finishes between 2–5, nine between 6–10, etc.
You can see the peak in the histogram — the most likely outcome of a race for him this year — is around 6–10 mark, with the peak size going down on either side.
However, there’s a pretty big peak at the far right. About 19% of all Keselowski’s finishes were in the top 5, but almost 16% of his finishes were between 31 and 35. Almost 25% of Keselowski’s finishes were out of the top 25.
We can compare Keselowski’s histogram with other drivers we are in the round of eight. I’ve plotted all the histograms on the same scale for easy comparison. They all show the same number of races: what’s different is the distribution of finishes. Here’s the Histogram for Joey Logano.
- While Logano’s histogram peaks in the 6–10th place, he has 14 finishes in that range, whereas Keselowski has 9.
- If you look at the weight on the left side, they’re actually about equal: Both drivers have the same number of top 5s, although Logano has 1 win and Keselowski three.
- 81% of Logano’s finishes were Top 15, whereas only 65% of Keselowski’s finishes were Top 15.
- The really big difference is to the right of the peak.
- Less than 10% of Logano’s finishes were outside the top 25,Keselowski has 25% of his finishes outside the top 25.
Here’s the Histogram for Almirola, who currently sits at 8th.
You can see that the peak is again in the 6–10 range, but Almirola has more weight toward the right side: almost as many 11–15 finishes as 6–10. Only 7% of his finishes are outside the top 25, but there’s a very large peak in the 21–25: 25% of Almirola’s finishes are outside the top 20. You can see from these distributions how these drivers compare. (I’m rooting for Almirola, but I don’t give him a strong chance of making to the Round of 4)
A histogram helps us look at the top four as well. Here’s Kyle Busch’s histogram:
The block at the right side are the three DNFs he had this year. He’s got one win in each of the 11–30 bins, but everything else is in the first three bins. Only 4% of Kyle’s finishes are outside the top 25. Compare with Kevin Harvick’s:
This looks very similar to Kyle Busch’s on the left side, but there’s a little more weight to the right. 12.5% of Harvick’s finishes are outside the top 25.
Martin Truex, Jr.’s is quite odd. As I noted earlier, he has no 6–10th place wins.
He’s got a lot more wins in the 11–30 range than the first two drivers. 16% of his wins are outside the top 25, including 5 DNFs. His consistency in getting top-5 finishes, even if they aren’t wins, offsets his lower-place finishes.
I almost forgot about Chase, but here is his data:
You can see that Elliott has a much broader range of finishes. The weight is much more equally spread over the first five bins. About 15% of his finishes are outside the top 25. Let’s circle around and compare Elliott to Keselowski.
They both have three wins, but Elliott peaks more to the left (higher finishes) than Keselowski.
Finally, we can put all this information for all 16 drivers on a single graph.
You see why I went through the individual graphs first, right? I’ve used a ROYGBIV + grey and black color scheme, showing wins in red at the bottom, followed by 2nd to 5th-place finishes in orange, 6th-10th-place finishes in yellow, etc. I ran out of rainbow and used grey and black at the end.
What does a champion look like on this graph?
- Lots of red: A champion has to win races
- The top-10s (red + orange) are more than 50% of the races (the Big Three)
- The top-15s (red + orange + yellow) are more than 50% of the races. This is true for Kyle Busch, Harvick, Truex, Jr., Elliott, Logano, Kurt Busch, Keselowski, Hamlin and Jones
- Outside the top-25 finishes (both purples, grey and black) are small. This is what killed Keselowski and (along with lack of wins) Johnson.
I think this analysis shows that NASCAR has actually balanced consistency and winning much better than many people (including me) thought. If you look at the entirety of a driver’s finishes, it shows that having a few wins can be offset by a lot of bad finishes. Drivers with lots of wins can afford a few bad finishes, but drivers without a lot of wins can’t.
Bonus Playoffs Data
As long as I had to go to all the work to pull this data together, I thought I’d look at some other things I found.
What Happened to Johnson and Larson?
Chase Elliott has been the bright spot for Chevy this year. The histograms for Larson and Johnson show the situation.
Johnson’s peak is in the 6–10 range, with almost nothing to the left. Johnson only finished in the top-5 6.25% of the time. He only had 12.5% of finishes outside the top 25, which is the same as Harvick; however, Johnson doesn’t have the wins and top-5s to offset those low finishes.
If Larson would’ve had a few wins, he’s be a strong competitor. The bulk of his finishes are in the 2–5 range and only 12.5% of his finishes are out of the top 25. But there’s a long tail on the right side of the distribution and nothing to offset it. If he’d won the two races he finished second in, you’d be looking at a very different outcome.
Leading the Most Laps Isn’t That Important
Havick has led a lot more laps than anyone else (1682 vs. 1231 for Kyle Busch and 970 for Truex, Jr. — but it didn’t put him in first place. Some of that is problem in the pits or accidents, but with stage points, it’s more important to be leading particular laps, even if you don’t lead the majority of them.
But here’s something interesting. I’ve compared the percentage of laps led in the regular season to % laps led in the first six races of the playoffs.
Harvick has led 30% of the laps during the playoffs, compared to about 18% during the normal season.
DNFs Do Not Matter — up to a point
- The top five drivers have 3, 4, 5, 4 and 5 DNFs. They have enough high finishes to offset those.
- The drivers at the bottom of the group of eight have the fewest DNFs: they have to, because they don’t have enough high finishes to compensate.
- Note that Keselowski also has the highest number of DNFs.
Originally published at The Building Speed Blog.