NASCAR: Putting a Number on a ‘Good Race’

It was the best of times, it was the worst of times.

That’s how I often feel listening to fans weigh in on NASCAR races.

Most people — me included — have a pretty strong feeling for whether a race was ‘good’ or ‘bad’. When pressed to explain that feeling, we often look to statistics.

It’s easier to compare cold, hard numbers. GPAs or rankings are so much easier to weigh than abstract concepts like ‘hard worker’ or ‘beat stronger opponents’. In NASCAR, people cite a record number of lead changes or more quality passing as signs of a ‘good race’.

With all the statistics available, it’s tempting to try to figure out which ones meaningfully determine an objective answer to the basic question: Was this a good race?

Zooming in on Passing Statistics

I would argue that passing for position — especially for the lead — is a necessary component of ‘a good race’.

A tweet from Todd Gordon (@ToddBGordon) about the 2019 Las Vegas race. He thought it was a good race.

In 2005, NASCAR introduced Loop Data. This data is collected from the myriad transponder loops around the circuit. If I remember right, some of the largest tracks have up to 18 loops. This data is useful for understanding individual drivers’ performances, but it might also tell something about the overall race. Mike Forde (the man in charge of loop data for NASCAR) had this to say about the 2019 Spring Las Vegas race.

A tweet from Mike Forde (@mforde) about stats from the 2019 Las Vegas Race: 47 green-flag passes and 2,245 total green-flag passes.

Green Flag Passes

Let’s look at the second bit of information first: This race had a whopping 3,345 green flag passes for the lead. Green Flag Passes measures how many times a driver passes for position while under green.

The weakness of this statistic is that you can be a few inches behind a driver at the first loop, then a few inches ahead at the second loop — a not-uncommon situation at plate tracks — and it counts as a pass. If that car gets ahead of you at loop #3, that counts as a pass, too. When cars are tightly bunched together on restarts (something we see more with the new rules package, and at plate tracks), there is a LOT of “passing” going on. This is more of an issue the more transponders there are at the track. This year’s Daytona 500 had 8,754 green-flag passes.

Green-flag pit stops artificially inflate the numbers for both overall green-flag passes and passes for the lead. If you’re the first person to pit under green, you are passed by every car that stays on track.

The 2019 Spring Las Vegas Race only had two cautions, both of which were for stage end. All other pit stops happened under green and they were staggered, which means that some drivers led by virtue of others cycling through pit stops.

A tweet from Mike Forde (@mforde) responding to the criticism that pit cycling is included in the statistics.

NASCAR pit stop data is on the web, but not in a form easily imported into a database or other analysis program. There are discrepancies between this data and the ‘official’ data. For Vegas, they cite a yellow-flag pit stop for Joey Logano on lap 176. The yellow flag didn’t officially start until lap 177. It’s important to be able to reconcile those differences before trusting the data.

Finally, anytime you change the number of transponder lines, you change the passing statistics. That has nothing to do with the racing, but everything to do with the measurement.

Lead Changes

The above stats covered all passes on track, but what if we just focus on the lead? I don’t have any doubt that the stat of 47 green-flag passes for the lead for Vegas is correct; however, I have no way of ferreting out that information from the statistics NASCAR makes available to fans.

Lead Changes is a race-level, lap-by-lap statistic. The driver crossing the start-finish line first is given credit for leading, even if he didn’t lead the entire lap.

The Lead Change variable (LC) doesn’t capture all the passes between the start-finish timing line. The 2019 Las Vegas race may have had 47 green-flag passes for the lead, but only had 19 lead changes.

How many of those 47 green-flag passes were the result of pit stops? There’s no way to tell from the data available to us.

So let’s look at the overall lead change variable, which is readily accessible to anyone. Here’s a graph of lead changes for the 2018 season.

A graph of the number of lead changes for every race in the 2018 season.

We expect the most lead changes at plate tracks. Sure enough, the maximum number of lead changes was 25 — at Talladega-1 and Daytona-2. The Daytona 500 had 24 lead changes.

But Atlanta and Chicagoland also had 24 lead changes. Mile-and-a-halfs can have large numbers of lead changes. They can also have few lead changes: The Charlotte Oval had only 9.

The smallest number of lead changes was 7 — at Darlington.

The Track Length Problem

Just as with caution numbers, we can’t use the raw numbers because races have varying lengths. There may be changes in the track layout or planned length, not to mention races cut short by rain or that went into triple overtime.

Perhaps the number of lead changes per 100 miles (LC100) will gives us a more objective view.

Lead Changes per 100 miles. The same data, but normalized to account for varying track lengths.

Darlington is still ranks the lowest, with 1.4 lead changes per 100 miles and Charlotte (the oval) is right behind it with 1.5 lead changes per 100 miles. But the next lowest lead change rate is the Vegas (Spring) with 2.75 lead changes per 100 miles.

If you want the most lead changes, the clear winner here is Bristol, with 6.75 (Spring) and 7.13 (Fall) lead changes per 100 miles. If the Fall Bristol race had been 500 miles, we would have seen 36 lead changes.

Now Can We Compare Years?

Okay, okay. Here you go. This graph compares LC100 for 2018 and 2019 on a track-to-track basis. Using this parameter allows us to compensate for things like the fact that Sonoma added the carousel this year.

Comparing lead changes per 100 miles for 2018 and 2019

To simplify the analysis, let’s also look at the values of the changes. In this graph, bars going up mean that there were MORE lead changes in 2019 and bars going down mean there were FEWER lead changes in 2019 than 2018.

Showing the gains and losses for lead changes per 100 miles in 2018 vs. 2019
Bars going up mean there were more lead changes in 2019 than 2018. Bars going down mean there were fewer lead changes in 2019 than 2018

After 17 races:

  • 10 races had more lead changes in 2019 and 7 had fewer.
  • The average gain in lead changes per 100 miles was 1.68.
  • The average loss in lead changes per 100 miles was 1.68.

That’s right. The gains equal the losses. There were 270 lead changes to this point last year and 270 lead changes this year.

Putting it in Context

Comparing two years isn’t really fair because the variation in numbers of lead changes can be considerable. Just looking at 2018 and comparing the same track in two different races, we see big swings in lead changes.

  • Las Vegas: 23 (Fall) vs. 11 (Spring)
  • Talledega: 25 (Spring) vs. 15 (Fall)
  • Michigan: 9 (June) vs. 15 (August).

How much do those numbers change over time? I chose the Coca-Cola 600 for this analysis because it’s always been 600 miles. I accounted for the few cases that had overtime or rainouts.

By 1963, there were reliably 40-43 cars per race. The lead change varies from 9 per 100 miles in 1973 to 1.5 in 2016 and 2017. (That corresponds to 54 lead changes and 9 lead changes respectively.)

The historical data for lead changes per 100 miles for the Coca Cola 600 from 1963-2019

But there is a lot more variation in the early years, so let’s look at a smaller range of data. From 1995-2019

  • The most number of lead changes is 6.3 per 100 miles (38)
  • The smallest number of lead changes is 1.5 per 100 miles (9).

Below, I’ve plotted a histogram for LC100 values from 1995-2019 to determine the most likely number of lead changes.

A histogram showing the wide range of possible lead changes for the Coca-Cola 600

This is by no means a normal (or binomial) distribution. This graph tells us that there’s no ‘most likely’ number of lead changes. It’s rare to have a number below 3 (18 lead changes over 600 miles), but you can’t say that there’s much of a difference in probabilities for the numbers above.

Consequently, you can’t say ‘if it’s above this number, it was good’.

The Caution Problem

The NASCAR stats that go into determining lead changes don’t differentiate between laps led under green and laps led under yellow. Some lead changes happen during pit stops. They have everything to do with strategy and less to do with who’s faster.

For example, let’s look at last year’s Indy race. There were 14 lead changes, but I’ve shown the leaders of the race, separating yellow and green-flag laps.

A schematic showing the drivers who led laps at the 2018 Indy race that shows how many of the lead changes happened at cautions.

A segment that starts with a yellow block indicates a lead change under caution. Eight of the 14 lead changes happened under caution. The end is a little confusing. Hamlin took the lead under caution, kept it through the next caution, then Keselowski took the lead on the restart and went on to win.

A high number of lead changes could indicate competitive racing, or pit strategy and a lot of cautions. Without delving deeper, there’s no way to tell.

Something You CAN Tell From Numbers

Even though we can’t use a high number of lead changes to unequivocally support the idea of a ‘good race’, I think we can all agree that a race is less excited when there are a lot number lead changes. Here’s a graph of laps led for the 2018 Coca Cola 600, which ended up on a lot of people’s ‘bad race‘ list.

A graph of the lap leaders for the 2018 Coca Cola 600 that shows what a race looks like graphically when one driver dominates.

This race, despite being 400 laps, had only 9 lead changes. Two of those lead changes are when Kyle Busch pitted and Brad Keselowski led two laps under caution, then Kyle Busch took the lead again. There were really only three drivers who led and seven lead changes. Kyle Busch led four times for 377 of 400 laps (331 of those laps led under green). Fifty four laps were run under caution.

But you didn’t really need the numbers to know that Kyle Busch ‘stunk up the show’, did you?

It’s Not Really About the Numbers

I know: Wouldn’t expect me to say that, would you? But let’s look at two examples. On those same lists of best and worst of the year for 2018, the Roval was uniformly cited as one of the best, if not the best, race of the year.

That race had only 10 lead changes (4.02 lead changes per 100 miles). Four of those 10 due to yellow flag pit stops. What everyone remembers is the end: Jimmie Johnson wreaking himself and Martin Truex, Jr., allowing Ryan Blaney to win.

And let’s talk about Darlington, which had the smallest number of lead changes of any race in 2018. What did people talk about on social media and comment boards? How Brad Keselowski pulled past Kyle Larson to win and how cool the throwback weekend was.

Not all races are going to be be exciting. If that’s what you want, then you’re looking for entertainment, not sports. Sports are defined by not having predictable (or predetermined) outcomes.

The truth is that the races we deem to be ‘good races’ either have exciting moments (often endings) or hold personal meaning. We all celebrate a driver’s first win, an underdog performing well, or a veteran who hasn’t won for awhile making what might be his last trip to Victory Lane. Most fans would call any race their driver wins a ‘good race’, while the fan whose driver crashed out on Lap 50 is going to have a far less sunny view of the race.

And isn’t that what racing is all about, anyway? Passion? My favorite thing about NASCAR fans is their passion. Let’s not lose sight of that amidst the avalanche of numbers available to us.

One thought on “NASCAR: Putting a Number on a ‘Good Race’”

  1. Let’s get real and do like THE UFC. THEY count meaningful strikes not pitty patty. Nascar is trying hard to sell mediocrity. Following the lead car with a faster car and not able to pass us why I have fast forward on my car. Still a fan but………

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: