One NASCAR Rules Package for All Tracks?

One of NASCAR’s major goals is cutting costs for teams by developing a rules package that, with minimal tweaking, can provide good racing at any track. The 2019 rules package met with mixed results. Races were mostly better at 1.5-mile and longer tracks, while fan and drivers complained about the performance on short-tracks.

Adam Stern reported this week that NASCAR is negotiating with teams about modifying the 2020 rules package for short-tracks. Although the modifications will most likely be bolt-on parts (splitter and spoiler), teams are concerned about having to invest in testing two different configurations of a car in its last year of competition.

Was Short-Track Racing Really Worse Last Year?

This graph (from one of my two year-ending posts on tracks and drivers) shows the percent change in lead changes per 100 miles (LC100). A value of 200% means that the lead changes per 100 miles in 2019 were double those for the same race in 2018. Blue bars denote increases and red bars denote decreases.

A column chart showing the relative increase or decrease in lead changes per 100 miles for 2019 vs. 2018

The results reflect the mixed results, but they also raise two important issues in understanding the success (or not) of the 2019 rules package.

First: race with a large number of cautions often have a lot of lead changes. How many of those lead changes are green-flag, get-your-heart-pumping lead changes and how many are staying out while everyone else pits?

Second, no two races are the same, even at the same track. Some tracks have wildly varying number for metrics like lead changes and cautions, while other tracks are more consistent. Are we getting the real story comparing 2019 with 2018 without knowing if the 2018 races were exceptional?

Lead Changes vs. Green Flag Lead Changes

I separated green-flag and yellow-flag lead changes for all the races from 2010-2019 using a very basic metric: a driver listed on the race report as having taken the lead during a yellow-flag is credited for a yellow-flag lead change.

Last year’s Coca-Cola 600 at Charlotte Motor Speedway had a whopping 214% increase in lead changes per 100 miles. It also, however, featured 16 cautions for 80 caution laps. That’s the most cautions and the most caution laps in a Charlotte race in the last 10 years.

The graph below breaks out yellow-flag and green-flag lead changes for each race. The codes below each bar show the year and the number the race was that year. The ’12s’ and ’13s’ are Coca-Cola 600s and the ’30s’ and ’31s’ are fall races.

A column chart breaking down total lead changes at Charlotte into those happening under green flag and those under yellow flags.
The red boxes are the number of cautions/total caution laps per race.

I’ve shown the numbers of cautions and caution laps in red boxes for select races. This graph suggests that there’s a correlation between yellow-flag lead changes and cautions.

It’s not a simple because the length of each caution determines whether it’s possible to have more than one lead change as people pit. A quickie caution may not allow time for any lead changes. While you may one or two lead changes during a six-lap caution, that doesn’t mean you’ll have three or four during a 12-lap caution.

Cautions Breed Yellow-Flag Lead Changes

To test my hypothesis, I’ve plotted the number of yellow-flag lead changes as a function of the percentage of the race run under caution (which is essentially caution laps corrected for the track length). .

A scatter plot showing a correlation between the percentage of the race run under caution and the number of yellow-flag lead changes
Data are taken from races from 2010-2019 at tracks from Atlanta to Dover alphabetically, excluding Daytona.

I plotted data for racetracks alphabetically up to Dover (excluding Daytona) and stopped there because I think it’s clear that there is a trend. More cautions mean more yellow-flag lead changes. Yellow-flag lead changes aren’t what we’d call ‘good racing’, so what we ought to be comparing are green-flag changes.

Here’s the same plot I started with, but considering only green-flag lead changes.

A column chart comparing the 2019 green-flag lead changes to the 2018 green-flag lead changes.

It’s mostly the same, but there are a few differences.

  • Ignore superspeedways. Pack racing means a lot of lead changes that are less meaningful than lead changes at other tracks. Talladega was coming off a 2018 fall race that had very few lead changes, so the 2019 race looks huge in comparison.
  • Larger tracks still did well
    • Most mile-and-a-half tracks had more green-flag lead changes than they did in 2018.
    • One Pocono race had a 25% increase; the other had a 25% decrease.
    • Both Michigan races were up, the first one by more than 200%
  • Not all short tracks were down
    • Spring Bristol was flat, while fall Bristol was up 133% – but again, we’ll have to see whether 2018 was an exceptionally static race.
    • Martinsville, Phoenix and Richmond showed significant decreases in both races.
    • Dover looks better when we only compare green-flag lead changes. Dover had fewer yellow-flag passes in 2019 than 2018, but more green-flag passes.
    • New Hampshire experienced an overall 40% increase in lead changes, but a 33% decrease in green-flag lead changes.

But we are still simply comparing 2019 with 2018. What happens if we broaden our view?

Ten Years of Track History

A few posts ago, I compared the race histories of Phoenix and Homestead to try to understand if we had to worry about Phoenix now being the last race. To do this, I had to introduce box plots. So, in case you’ve forgotten…

Box Plot Reminder

  • The red (or yellow) line represents the median (the middle value).
  • Fifty percent of all races fall within the box.
  • The whiskers show you the full range of ‘typical’ values.
  • Values far outside the norm are represented by dots.
  • I’ve put the average (or mean, which is different than the median) in red at the bottom of each box plot.

The Metrics

  • Average Green Flag Passes per Lap are passes for position, averaged throughout a race
  • Green-Flag Lead Changes and Margin of Victory are self-explanatory
  • Number of Leaders is the number of drivers who led one or more laps during the race
  • Quality Leaders excludes drivers who led only one or two laps at a time — which is usually because they stayed out while people were pitting.
  • Most Laps Led is the most laps led by a single driver throughout the course of the race. The larger this number is, the more dominant one driver was.

An Example: Michigan

Michigan had a 200%+ increase in green-flag lead changes in 2019 relative to 2018. That makes the new rules package look amazing. However, the 2018 race was way below average in terms of lead changes. Compare 2019 to the last decade at Michigan.

Box plots showing the range of values for average green-flag passes per lap, green flag lead changes and margin of victory for Michigan 2010-2019
The green arrows show the values for the first (spring) race and the red arrows show the values for the second (fall) race.

It’s clear from these metrics that 2019 was indeed an improvement not just over 2018, but over the typical Michigan race from the last ten years. The average green-flag passes per lap high and the green-flag lead changes are high. One margin of victory is just about at the median, while the other is among the smallest.

Let’s re-examine our results now.

1.5-Mile Tracks

All of the mile-and-a-half tracks showed improvements, although some were larger than others. Most of the improvements are significant enough that it’s fair to credit the rules package.

  • Texas
    • The average number of green-flag lead changes at Texas is 14.8. The two 2019 races were 22 and 21.
    • The average green-flag passes per lap is 9.5. 2019’s were 10.6 and 11.4 (The 2018s were in the 6’s, making 2019 look really good compared to 2018, but just ‘better’ compared to the average.)
  • Las Vegas
    • The average number of green-flag lead changes at Texas is 14.1. The two 2019 races were 17 and 21.
    • The average green-flag passes per lap at Vegas is 10.7. 2019’s were 13.1 and 14.3.
    • In its history, Las Vegas had never seen more than 7 quality leaders. The spring race had 8 and the fall race had 10.
  • Kentucky
    • Kentucky had a record of 6 quality leaders prior to 2019. In 2019, there were 9 quality leaders
    • The average green-flag lead change is 8.7 per race, whereas the 2019 race had 11.
    • The average green-flag passes per lap is 8.9. The 2019 race saw 12.9 green-flag passes per lap on average.

Short Tracks

The news for the short tracks is still mixed. Let’s look at the three tracks of concern: Martinsville, Richmond and Phoenix.

Martinsville

Here are the box plots for Martinsville covering 2010-2019

Box plots showing the range of values for average green-flag passes per lap, green flag lead changes and margin of victory for Martinsville 2010-2019
Again, the green arrows show the values for the spring race, and the red arrows show the values for the fall races. The red number at the bottom is the average.

The 2019 Martinsville races are both below average when it comes to green-flag passing and well below average in green-flag lead changes. On the positive side, the margins of victory weren’t too far from the median.

Box plots showing the range of values for number of leaders, number of quality leaders and most laps led by a single driver for Martinsville 2010-2019

Martinsville doesn’t look any better if we look at this set of metrics, either. We had low numbers of leaders and quality leaders and one dominant driver in each race. In case you’re thinking maybe one team figured out Martinsville, there were two different winners from two different manufacturers in those two races.

The 2019 package is the culmination of a number of experiments over the last few years, but if you look at the trends over time, they’re going the wrong way. This is a pretty big piece of evidence as to why NASCAR is going to make a separate rules package for short tracks in 2020.

Total lead changes and yellow-flag lead changes from 2010-2019 for Martinsville.

The number of lead changes is down in 2019, but they were down in 2018 relative to 2014-2015. So the argument that there really are problems with racing at Martinsville holds, no matter how you look at it.

Richmond

Sadly, we also can’t write off Richmond as looking bad compared to awesome 2018 race: The metrics are below average when compared to the whole decade, including the largest margin of victory in the period covered.

Box plots showing the range of values for average green-flag passes per lap, green flag lead changes and margin of victory for Richmond 2010-2019

The second set of metrics isn’t any better, although there were more drivers leading laps here than at Martinsville,

Box plots showing the range of values for number of leaders, number of quality leaders and most laps led by a single driver for Richmond 2010-2019

Phoenix

As I pointed out in the earlier post, Phoenix wasn’t horrible. It was just on the low end of average for a Phoenix race.

Box plots showing the range of values for average green-flag passes per lap, green flag lead changes and margin of victory for Phoenix 2010-2019
Box plots showing the range of values for number of leaders, number of quality leaders and most laps led by a single driver for Phoenix 2010-2019

Hopefully, the tweaks NASCAR makes to help the tracks that decidedly need it will all help tracks like Phoenix.

A Few Other Tracks

Bristol was average to better-than-average in both races when compared to the last ten years of races. Both races had higher than average numbers of leaders, less domination by one driver, and higher-than average green-flag passes per lap.

New Hampshire, like Phoenix, was simply on the lower end of average for its races.

Homestead was just weird. In the last four races, an average of 40% of the cars finished on the lead lap. In 2019, only 25% of the cars finished on the lead lap. The metrics are low-to-average, except the margin of victory was the highest we’ve seen in ten years. Given that the rules package performed well on the other 1.5-mile tracks, 2019 might just have been an anomaly. Of course, there’s no way to know for sure until 2020.

Box plots showing the range of values for average green-flag passes per lap, green flag lead changes and margin of victory for Homestead 2010-2019
Box plots showing the range of values for number of leaders, number of quality leaders and most laps led by a single driver for Homestead 2010-2019

Conclusions

If there were any way to avoid introducing a separate rules package for short tracks in 2020, NASCAR would have found it. It will be extra work for everyone, but it’s to their credit that they didn’t decide to just suffer through the year with the hope that 2020 would be different. And this way, they can collect more data that can be used to improve the 2020 rules package.

2 Comments

  1. I hate to say it, but, Great Races are in the eye of the beholder, not necessarily analytics. I’d like to see more Full Unbroken Races, and Wire to Wire and 1 or More Lap Victories. I’d don’t know any other way than the High Porch, but, NASCAR has historically struggled with Rules Packages (Daytona-Talladega).

    • I agree with you totally that ‘great races’ are in the eye of the beholder — but I see a lot of people supporting their arguments with observations and impressions that often aren’t correct. If a person uses analytics to support his or her argument, I suspect we both agree that he or she should at least use the right analytics.

      I find that how the drivers you pull for do in the race plays a big role in whether you liked the race.

      Thanks for taking the time to comment.

Leave a Reply

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