Will the Gen-6 Car Affect the Number of Cautions?

 Cautions, Edwards Carl, math, Racing, Tony Stewart  4 Responses »
Jan 252013
 

I love the Gen-6 car.  Not as much as I love the Nationwide cars (but that’s got more to do with what I drive than it does the cars).  The big question is whether the decrease in cautions is going to be changed because of the new car.Let’s start (as we usually do) with the new car.

Graph4Let’s start (as we usually do here) with the data.  I’ve tabulated the data for cautions for the last twelve seasons and found that cautions have been decreasing since 2005, as shown,  for both the Nationwide and the Sprint Cup series.

In order to compare the two series and to compare between seasons within a single series, I’ve plotted the number of cautions per 100 miles.

In 2012, the Sprint Cup series had 1.57377 cautions per 100 miles.  They drove 13725 miles total, so that was 216 cautions total.

In 2012, the Nationwide series had 2.23969 cautions per 100 miles.  They drove 8240 miles, with comes out to 189 cautions — essentially the same number of cautions per mile they had last year.

Conclusion #1.  If the Nationwide drivers had driven the same number of miles as the Sprint Cup drivers, they would have had 307 cautions.

You’ll notice that I’ve drawn lines through each set of data.  They aren’t just a best fit by eye – I actually did a non-linear least-squares fit that determines the line that goes closest to all the points.  The data are decidedly linear and, more importantly, there aren’t any bumps or jump in, say, 2008, when the COT (which I guess is now the Gen-5 car) was introduced, or in 2011 when the Nationwide car was changed.  The data remained pretty consistent.

Conclusion #2.  Cautions are not affected much by the car that’s being driven.  Sure, I expect there to be some driver errors when a car doesn’t handle the way the driver expects it to behave; however, these guys catch on really quickly, so that’s going to be maybe 5 cautions.  Five out of 216 is like 2.3 percent, which is well within the error in the fit parameters.

Why are the cautions decreasing?  I’ve gone into this before, but I believe it is essentially because the drivers have a lot more experience now than they did in previous years.  There are a lot of veteran drivers in the Cup series right now, and I calculated that if you add up all the races run by the current crop of drivers, they have run a total of about 1000 more races in 2011 than they did in 2005.  That’s a whole lot of experience, and it’s distributed amongst the drivers.   Compare just two drivers:  Tony Stewart had run 248 races in  2005 and at the end of 2012, had run 500.  Carl Edwards had only run 49 races in 2005 – compare that to the 301 races he’d run as of today.  (I am only counting points paying races.  If you could somehow quantify the number of practice laps, time testing, etc., I think that would only make my argument stronger.)

So, in short, I don’t expect there will be any significant change in cautions because of the new car — up or down.  What do you think?

 

Cautions: A New Low for NASCAR

 Cautions, NASCAR, Racing, Statistics  2 Responses »
Nov 212012
 

At the start of the season, the big news was that cautions were remarkably down from last year.  As I showed, this isn’t a new trend – it’s a continuing trend since 2007.  Since the season’s data are now complete, I thought it was time to revisit the data.

Plotted at right are the caution data from 2012 compared to the data from the previous six years.  I’ve normalized the cautions to cautions per 100 miles to account for the changing lengths of some races over the years.

Note how the data jump around pretty wildly until about race 10.  That is because averages are only meaningful when you’re averaging enough information.  When you average a small number of measurements, the average fluctuates until you get enough data that the numbers mean something.  You can’t predict anything from the first five or ten races.  Please remember that next year when the prognosticators tell you someone’s season is over just after first Martinsville.

The 2012 data are the green triangles – a whopping decrease in cautions from 2011!  In absolute numbers, there were a total of 218 cautions this year compared with 278 in 2011 and 265 in 2010.  Where were the biggest numbers of cautions?

Highest Lowest
Race Month Cautions/100 miles Race Month Cautions/100 miles
Bristol August 4.87  Fontana  March  0.388
Martinsville October 4.18  Texas  April  0.400
Martinsville April 2.58  KansasHomestead  AprilNovember  0.750

The Nationwide Series didn’t experience the same drop in cautions – in fact, they had just about the same numbers of cautions this year as they did last year.  The tracks with the largest caution rates per 100 miles for Nationwide are:

Highest Lowest
Race Month Cautions
/100 miles		 Race Month Cautions/100 miles
Bristol August 6.77  Fontana  March 1.33
Phoenix November 4.90  Iowa  September  1.37
Kansas OCtober 3.88  Dover  September  1.50

Here’s the updated graph, showing cautions per 100 miles since 2001.  You’ll notice that 2012 marks a new low for the Cup Series.

I’m still working on the analysis, but I think my original theory holds for why the number of cautions has been decreasing:  we simply have more drivers with more experience.  There are a lot of veterans and fewer rookies.  The average time each driver has been driving in the series is higher, which means that the drivers are simply better.

The Six-Year Downward Caution Trend: In the Nationwide Series, too!

 Cautions, Consipracies  1 Response »
May 242012
 

Just out of curiosity, I decided to do a similar analysis on the Nationwide Series caution record as I did on the Cup record.  My intent was that if there was something specific to the Cup series – the new car, the Chase, etc.), it would show up because the Nationwide would follow a different trend.  Not at all!  Remember that I’m plotting number of cautions per 100 miles run.

No too surprisingly, the Nationwide Series has more cautions than the Cup Series; however, the cautions are decreasing much faster in the Nationwide Series than in the Cup Series.

Taking the linear trend to its logical extreme, we can expect the first caution-free season in 2018 for the Cup Series and 2030 for the Nationwide Series!

Cautions: A Historical Downward Trend Over the Last Six Years

 Cautions, math, NASCAR  4 Responses »
Apr 242012
 

OK, I know I promised the next post was on engines, but I got sidetracked…

Being the data geek that I am, I was really curious if the decreasing number of cautions was specific to this year.  It’s not:  Cautions have been decreasing since 2005,as the graph below shows.  The squares are the cumulative number of cautions per 100 miles, obtained by adding up all the cautions in a season and dividing by the total number of miles in the races.  (This is a more accurate number than total cautions, given rainouts, shortening races and different venues from year to year.)

The straight line is a linear regression, with a R-squared of 0.87, which is pretty good.  The grey box in the lower right hand corner is what the fit predicts the number of cautions should be if the trend continues.

Of course, someone is wondering what happened before 2005. The trend was totally opposite.

My rationale for going back to 2001 is that this was the first 36-race season.  Not a great reason, but that was pretty much it.  The peak number of cautions was in 2005.  What happened in 2006 that sent the number of cautions down?

 

 

Are Cautions Really Going Down?

 Atlanta Motor Speedway, Cautions, Martinsville, math, New Hampshire, News, Talladega, Texas Motor Speedway  No Responses »
Apr 182012
 

I honestly cannot help it – scientists are naturally skeptical.  If you make an assertion, I will have to question you on what data you have that supports it.  This is second nature to the people I work with, but I realize it is damned irritating to non-scientists (aka “normal”) people.

So when I started reading everywhere that “cautions were down 35%”, I had to go look into it.  This is a preliminary post – more detailed analysis will follow as soon as I’ve read my students’ final projects and gotten comments back to them.

First, let’s talk statistics.  Reliable statistics require large numbers.  It drives me nuts when people extrapolate from the first few races of the year.  You can’t claim much on the basis of five data points.  Even the top quark required seven (if I remember right – they did get more after they announced they’d found it).

The stock market fluctuates up and down.  Everyone except people who are thinking about retiring ignore the short-term fluctuations and focus on the long-term trends.  What do the data say about cautions in NASCAR?

I picked five tracks to analyze in this first round:  Martinsville, Texas, Talladega, New Hampshire and Atlanta.  The first four represent a range of track types, while the last was chosen to see whether the cautions were “cookie cutter-like”.  I first plotted the number of cautions as a function of year for all the tracks together.  If cautions are decreasing, we should see a general trend downward.  Here’s what I got:

Not much of a clear trend, huh?  If anything, it looks like the overall trend (since 1950) is going up.

Thinking it might be unfair to use really old data, I decided to focus on 1997-2012.  I plotted all five tracks on their own graph for just those years.  I’m sorry for the color – those are the defaults on Origin.  I will change them when I do a full post.

What do you think?  I might buy a downward trend for Texas, but it’s hard to make that argument for the other tracks. Martinsville went from 18 in the last race of 2011 to 7 this year – that’s a 61% drop right there — but if you compare it to the Spring race (and an argument can be made for comparing Spring to Spring and Fall to Fall), that race had only 11 cautions.  That’s a drop of 36.3%.  there is a wide gap between those two figures.

Just for fun, I took the historical data for the three tracks with long records.  Here they are:

As I said, I’ll follow this up with more extensive analysis, but I wanted to get the data out there ASAP.