Can You Predict a Driver’s Success from their Rookie Year Stats?

Now that the season is over, it’s time to look through the statistics from the year.  One big story this year was Danica Patrick’s rookie season in Sprint Cup.  She didn’t set the world on fire:  One top ten, one pole, five laps led and only thirteen lead lap finishes put her in 27th place for the season.

Tony Stewart, when asked about Danica’s rookie year, said that he saw ‘potential’.  Potential tends to be an ephemeral thing and numbers geeks like me strive to put a quantitative slant on everything, so I thought I’d look at the numbers and see if there was anything we could glean from comparing different driver’s rookie year stats.  With thanks again to the folks at for the numbers, here goes.

I didn’t include stats like average starting position or poles because, frankly, you don’t get a championship that way.  I focused on how drivers finished.  The graph below shows the percentage wins, top five and top tens for a number of drivers, ranging from champions to drivers still proving themselves and, yes, two of our rookie-of-the-year competitors.


I think it is fair to conclude that the best drivers in the series – the multi-time champions like Johnson and Stewart – were just good from day one.  Finishing almost 60 percent of your races in the top 10 in your first year is pretty darned impressive.  The guys who looked like champions in their first year usually became champions (or came pretty close).

But the converse doesn’t hold.  A slow start doesn’t preclude the possibility of your becoming a champion.

If you look at the graph above to find a driver whose rookie year is comparable to Danica’s, you inevitably land on Brad Kezelowski.  His rookie stats weren’t that different than Danica’s.  She had one top-ten, he had two.  Neither had any top fives or wins.  They both had a pole.

So maybe there’s something that average statistics, like top tens and top fives don’t tell you.

Digging Deeper:  Histograms


A histogram plots the frequency of occurrence – for example, how many times a driver finished within each particular range of finishing positions.

I chose this kind of analysis because it’s the next step up from an average finishing position.  You can take two drivers with the same average, and one could finish 17th every race, while the other alternated between winning and crashing out in the first lap.    Someone once told me that ‘averages hide a multitude of sins’ .

I broke the finishing positions into five-position segments, but I broke out wins separately.  The histogram to the right shows Jimmie Johnson’s finishes for his rookie (2002) season.  The bar corresponding to ’35’ on the chart at right shows that Jimmie Johnson had four finishes between 35th and 39th that year.  The bar corresponding to ‘5’ shows that Jimmie had 14 finishes between P5 and P9.   Johnson had an average finish of 13.5 and finished 5th in the standings that year.

Stewart1999The histogram for Tony Stewart’s rookie season (1999) looks a little different than Johnson’s in the details, but (like Jimmies) it’s skewed to the right-hand side of the plot (meaning more good finishes).   Stewart had an average finish of 10.3 and finished 4th in the standings.

And then we come to Kezelowski, who had an average finish of 22.4 and finished 25th in the standings.  Unlike Johnson and Stewart, who have significant bars for top fives and wins, there are no bars for Kezelowski.  He had a few bad finishes, but he consistently ran between 25th and 10th, with more of the finishes being on the higher end of that range.


Which brings us to Danica.  Even though she finished in 27th place (only two behind Brad in his rookie year), the statistical distribution of her finishes looks much different than Keselowski’s.


The mode of a group of numbers is the number that appears most often.  It’s the most probable outcome – and the highest bar in these histograms.  Brad has a pretty even spread between 30-10, but Danica has a pretty significant peak at the 25-29th place finishing positions.  She had a lot of finishes in the 25-29, while Brad was almost equally likely to finish 10th as 30th.

How Drivers Change

The big question, of course, isn’t how anyone did last year – it’s how they’re going to do in the future.  If there were some magical algorithm that allowed an owner to predict that, if they put three years of development into a driver, he or she would pay off, they’d have no problems deciding who to keep and who to let go.

I was curious about how drivers changed.  If you start off good, there isn’t much room for improvement, right?


This histogram compares Jimmie Johnson’s finishes in 2002 with those of this year.  He’s got fewer bad finishes and increased the number of wins, top 5s and top 10s.  Bad finishes are tricky because in a sport like this, you can get a bad finish through no fault of your own if you get caught up in someone else’s accident.  I haven’t looked at it in detail, but my first glance suggests many of the bad finishes for drivers who don’t have a lot of bad finishes are at the short tracks and the superspeedways, or they’re due to equipment malfunctioning, like a motor or transmission giving out.

What about Brad, whose rookie year wasn’t quite as promising?

KezTimeThis histogram compares Brad’s rookie year, 2010 with his championship year, 2012.  Look at the difference in finishes in just two years.  With the exception of a very few bad finishes (one was Daytona) every finish was 19th place and up.  What happened in those two years?  Some experience.  A new crew chief with whom Kezelowski seems to have really bonded.

What Does it All Mean?

If a driver starts off with a slew of top 10s in his first year, the chances are very high that driver will continue to run at the top as his or her career progresses.   You cannot, however, make the converse argument:  a driver with a weak rookie year may — or may not — become competitive.  We’ve seen some good drivers from other series come to NASCAR and struggle.

There are many factors that go into winning:  you have to have most of them to run well.  Sometimes something like changing crew chiefs, owners or even being in the final year of a contract will radically change how a driver runs.  Look at Joey Logano when he moved to Penske, or Matt Kenseth when he moved to Gibbs.

Now… that isn’t to say that there aren’t some clues that might tell you whether a driver is destined for greatness or mid-pack purgatory.  We’ll dive deeper into the data next time.



The Science of Speed: Week Ending 11/14/13

Hard to believe it’s winter, but we got our first snow already. Here’s the news of the week.

  • It’s holiday shopping time – great idea for your gearhead: magazine subscriptions.
  • Despite hoping for a down-to-the-last-lap Homestead finale, all the stats geeks cite a pretty high probability Johnson takes it all.
    •  36 Races notes that Johnson has a 91% chance of winning the championship, but also that Harvick has a 3% chance of taking home the title.
    • Focusing more on who wins the Homestead race, PitRho has an interesting first choice for your fantasy team:  Parker Kligerman.
  • @nascarnomics looks NPR’s “esteemed commentator” Frank DeFord’s assertion that NASCAR fan numbers rose as speed increased and now that we’re only getting marginal increases in speed, that explains the declining attendance.  He analyzes attendance vs. pole speed and the results are very interesting.
  • F1’s all aflutter about the ugly noses on the 2014 car.
  • Have you ever seen a string theorist interviewed by a cat?  If not, you really should.

And in my news, I’m busy getting things ready for the end of the NASCAR season.  Whole lot of interesting stats I’ve found looking at the year as a whole.  Those’ll be coming up soon…


NASCAR’s Concussion Policy

Concussions were big news in a week where no one actually got one.

NASCAR announced a new policy on concussions :  Starting in 2014, all drivers will be required to have a baseline test at the start of the season.  NASCAR will be using the ImPACT (Immediate Post-Concussion Assessment Test) , which I discussed back when Dale Jr had his concussion and had to sit out a couple of races.  The ImPACT test has been widely used already in NASCAR and NASCAR did quite a bit of education to the drivers and teams about the test, how it originated, and why they are doing this.  Unlike the NFL, which has consistently denied that concussions were serious or needed attention, NASCAR is taking this seriously.

Brad Keselowski doesn’t like the new policy. And, as is his style, he made no bones about his unhappiness with the new policy.

“Doctors don’t understand our sport. They never have and they never will. Doctors aren’t risk-takers. We are. That’s what makes our sport what it is and when you get doctors involved, it waters down our sport.” (via USA Today)

One aspect Keselowski didn’t like was the lack of a firm number coming out of the test.  Without one, he said, the test was “a waste of time.”

“It’s just another subjective field for doctors that don’t understand our sport… This is not the field for doctors. Let them play in their arena and I’ll play in mine.” (via USA Today)

Sigh.  I remember being twenty nine and thinking I knew everything.

Let’s just review for a moment.  ‘Concussion’ denotes a range of damage to the brain that results from blunt trauma.  Your brain sits on your spine, but it’s not rigidly held in place.  Cerebrospinal fluid surrounds the brain and cushions it as it moves.  There’s only so much cushioning that fluid can do.  If you get hit hard enough, the brain hits the skull and the cells that make up the brain can be damaged.  Brain cells are essentially electrical components.  If you throw a radio or a television and damage the components inside, it won’t work.  Same thing goes for your brain.

We don’t have a good way of figuring out how much your brain is damaged at present because we can’t open up someone’s head and look for the equivalent of a bruise.  Much like psychological illnesses, we have to piece together symptoms and deduce the cause from the symptoms.  Someone who is concussed may or may not lose consciousness and may or may not suffer from symptoms that include ringing in the ears, headaches ranging from mild to blinding, intolerance to light and sound, and a host of other things.

In most illnesses, the patient wants to help the doctor figure out how to make him or her better.  In a concussion, patients sometimes hide symptoms because they are afraid (as I suspect Mr. Keselowski is) that they are going to be prohibited from doing what they love. Racing IS different than stick-and-ball sports:  if the quarterback on a football team gets a concussion, the team can still make the Superbowl.  If the driver gets concussed and has to miss races, his or her season is over.  It is nearly impossible to make the championship if you miss a race.  Imagine if Matt Kenseth or Jimmie Johnson were told mistakenly that they had a bad concussion and they couldn’t race.  You have no guarantee you’ll ever get back to that position again.  I get that.

Having said that, the idea that racers are risk takers and therefore the rules shouldn’t apply to them is pretty absurd.  Take a look at the articles Matt Crossman wrote about concussions in the NFL for Sporting News.  There are men in their forties and fifties who can’t read because they can’t focus for more than a few minutes, men with sleep disorders that cause them to wake up and find they are choking their wife, men who say that, if they knew what playing football was doing to their brains, might have decided not to play, or to retire earlier.

Think of a concussion as analogous to a broken bone.  The bone needs time to mend.  If you get in another accident and the bone is still weak, you stand a high probability of damaging your body in a way that can’t by fixed.  There is nothing physical that prevents Brian Vickers (who is taking blood thinners to deal with blood clots) from driving.  He said “I can drive fine.  I just can’t crash.”  The blood thinners are doing their job, but if he were to have an accident and start bleeding, the blood thinners would decrease the blood’s ability to clot.  It’s the same thing with a concussion. It might not impair your ability to drive; however, it puts you in a very bad position if you have another accident.

This is why doctors have spent years developing ways to determine whether someone is concussed that is more objective than asking the patient how they feel.  ImPACT tests six areas:  verbal and visual memory, visual motor speed, reaction time, impulse control and symptom score.  It’s a half-hour computer-based test that flashes numbers and letters and asks you to remember colors or sequences.

And the reason they don’t use numbers is that you don’t measure people against some ideal standard.  Everyone’s brain works differently.  Some people have better reflexes than others.  Some people are better are remembering numbers than others.  What is important is how your brain changes after a concussion.  Without a baseline test, there’s nothing to compare the test to if you do have an accident. Likewise, there’s no magical ‘cut off’ where a doctor is going to say “You can’t race”.  They’re going to be able to show you your scores in various areas at the start of the season and then how those scores have changed after an accident.  Then the doctor is going to make a recommendation.  The driver, the owner and NASCAR then have to come to a decision about whether the driver races or not.

We all take risks.  Drinking alcohol has risks.  You decide whether those risks are worth the rewards.  The same should hold with concussions.  A person has the right to know what the risks are – then they have to make decision about what risks they are or aren’t willing to take.  Leaving aside the obvious cases in which a driver’s condition might endanger other people, there’s a broad range of possibilities left.

Keselowski also argued that we simply don’t know enough about concussions to have tests.  That’s also remarkably ill-informed.  Yes, we don’t know everything about concussions, just like we don’t know everything about depression, heart attacks and why we have appendixes (appendices?).  But the evidence overwhelmingly suggests that concussions have much farther ranging impacts than we thought.  A brand new study out this week shows that NFL players benched due to concussions have altered brain connections.  They used functional magnetic resonance (fMRI) imaging to study the brains of people while they did tasks.  fMRI detects oxygen.  Since your brain draws blood to the parts that are being used the most, you can tell what areas of the brain are active using MRI.

The study compared football players who had been benched due to at least one concussion with age-matched college graduates and gave them tasks, like given two different arrangements of colored balls, how many moves would it take to change the arrangement from one to the other.

They found very small differences between the two groups.  Only when the number of moves became large did the football players perform significantly below the non-football players.  But – they found that the areas of the brain that were being used were different between the two groups.  The different areas of football players’ brains didn’t interact as much as the non-football players.  This was particularly strong in the frontal lobe, which is the part of the brain being used for planning, organization and attention.  In essence, this study showed that although the two groups performed the same, it took the football players much more brain power to do it.

It’s a limited study and needs follow up, but it sure is interesting because the harder the task, the harder the brain is going to have to compensate.  And these were people who had one concussion.

The other reason I didn’t want to let Mr. Keselowski’s comments go is that we do have evidence that concussions in younger people could be more impactful (sorry, couldn’t think of a better word) than in older people.  The Institute of Medicine released a report this week on the science of youth concussions.  Thousands of young people play football, soccer, hockey and serve in the military – all places with a higher risk of concussion.

Concussions are potentially serious and everyone’s tolerance for risk is different.  But we need to make sure people have enough information so that they make wise decisions.