Do caution lengths increase at the ends of races in the hopes of causing green-white-checkers finishes? Let’s see what the data say.
I looked at caution lengths from 1980-2020. I purposely excluded 2021 because some types of tracks tend to have more cautions (hello superspeedways!) and it’s not fair to compare a partial season against a full season. That gives us 9,813 cautions.
I broke these into 10 categories. Of those, I am eliminating the 155 cautions caused by weather. I figured it was sort of unfair to hold NASCAR responsible for Mother Nature. That leaves us with 9,658 cautions in the following categories:
- Accidents (4,875)
- Spins (1,645)
- Debris (1,358)
- Oil or Water on Track (498)
- Other (405)
- Stage-end cautions (269)
- Engines (244)
- Competition cautions (183), which reliably average 4-4.5 laps
- Stalled Cars (181)
A Brief Tangent
Most of the ‘Other’ category are from the 80’s, when not everything got written down. But some did. My favorite entries in the ‘other’ category included:
- fire (2 infield, 2 on pit road, one on track)
- ambulance crossing track (7 – one we don’t (thankfully) see anymore)
- animals on track (3)
- security (1 – the guy who climbed the catchfence at Richmond
- umbrella on track (1 – Bristol, 1991)
Have Caution Lengths Changed Over The Years?
The first thing we’ll consider is whether caution lengths have changed with time. I’ll look at them by category because the frequency of the cautions is also sort of interesting.
Accidents and Spins
Let’s start with accidents because accidents are by far the most common cause of cautions. I’ve plotted the number of accidents on the top graph and the average length of cautions due to accidents on the lower graph for each year from 1980-2020.
The upper graph shows how things have changed since the 1980’s, when there were only about 50 accidents a season. We reached a maximum of 199 accidents in 2005. Since then, the trend has been mostly downward.
The lower graph tells us that the average caution length due to accidents has gone down over the years. The best way to see this is to look at decadal averages. NASCAR has cut about a lap and a half out of caution lengths due to accidents over the years.
|Decade||Average Length of Caution/Accident||Average Length of Caution/Spin|
Spins take a little less time to clean up because there are usually only one or two cars involved. The caution length due to spins has remained pretty constant over the decades. That makes for a boring graph, so I didn’t even bother showing it.
Over the last three years, the average time to clean up after an accident is 5.0 laps.
Debris cautions weren’t an issue until the 2000s. I suspect that is because cars tended to break down and retire more often than they wrecked out. The maximum number of debris cautions was in 2005 at almost 200. It started going up again in the mid 2010s — until the damaged vehicle policy was implemented. I think the DVP might be one of NASCAR’s most underappreciated changes.
As you can see from the lower graph, the caution lengths haven’t changed significantly over the years. The decadal averages varied from 4.67 to 4.84 from 1990-2010.
As for the other types of cautions:
- Stage-end caution lengths reliably average 6 laps
- Competition caution lengths reliably average 4-4.5 laps
- We hardly ever have engine cautions anymore
- Liquid on track caution lengths average right around 6 laps, but there are so few these days that we’re talking about very few data points.
Do Caution Lengths Increase Within a Race?
The next question is whether the lengths of cautions changes as you get closer to the end of the race.
Let’s start by dividing races into quartiles: The first 25% of the race, the second 25%, etc. If NASCAR is making cautions last longer toward the end of a race, then the average caution length in the last 25% of laps ought to be larger than the average caution length in the other quarters.
2020 as an Example
The plot below is for 2020. I’ve excluded competition cautions, stage ends, weather and engine cautions. This graph lets you see where cautions happen relative to the race end.
Of the 148 eligible cautions in 2020:
- 35 were in the first 25% of the race
- 41 in the second 25%
- 29 in the third 25%
- 43 in the last 25%.
This is consistent with the fact that cautions tend to happen later in races, when drivers get a little more willing to take risks.
But what about lengths? First, I’ll summarize a few numbers in this table. The number under each of the quarters is the average caution length in that quarter of the race.
There are few cases in which the average fourth-quarter caution is longer than those in the first (or first and second) quartiles. But there are no cases in which the average fourth-quartile caution length is greater than the third-quartile caution length.
It was too much data to put on one graph, so here are two, each containing 20 years worth. We’re looking for evidence that cautions get longer toward the end of the race — which means that the red bar should be larger than the yellow bar representing the third quarter of the race.
This happens exactly once in forty years — in 1986 when the average length of a fourth-quarter caution was 5.64 laps and the average length of a third-quarter caution was 5.52 laps. Not too much of a different.
Sure, there were cases in which the last-quarter lengths of cautions were longer than the first quarter, but no real evidence that would suggest a systematic process of increasing caution lengths at the ends of races.
What About Caution Lengths at the Very End of Races?
But then I got to thinking: What if I’m missing something by lumping together the last quarter? What if it’s only at the very end of the race that cautions start getting longer?
So I repeated the analysis, but this time, instead of quartiles, I used tenths. Unfortunately, that makes for really messy graphs. I’m simplifying things as follows: For all years but one, the average length of Q4 cautions are less than Q3 caution lengths. So I’ll just show you the last three tenths of the race.
The average caution length in the last tenth of the race is higher than the average caution rate in the tenth before it only three times in 40 years: in 1981, in 1992 and in 2015. But note that, in 2015, we’re talking about an average of 4.95-lap cautions vs. 5.05-lap cautions, so it’s not much of an increase.
With the exception of 2015, in the last twenty years most of the last 10% of the cautions have been anywhere from 70-90% the lengths of the earlier cautions.
In conclusion, this analysis shows no evidence that caution lengths are systematically increasing over the last 40 years, or over the course of an average race. One can no doubt find individual races that do exhibit this behavior, but the overwhelming majority of races do not.