Sunday, September 29, 2013

How did my Dover Picks do?

Reviewing my Dover picks, as featured in the latest BSports video:

Dover Favorites

  • Jimmie Johnson (WON the race and led 243 laps)
  • Jeff Gordon (finished 4th and led 3 laps)
  • Kyle Busch (finished 5th and led 30 laps)

Threats for First Dover Win

  • Joey Logano (finished 3rd)
  • Kevin Harvick (finished 6th)
  • Clint Bowyer (finished 10th and led 1 lap)

Going 1-3-4-5-6-10 is not bad.

Wednesday, September 25, 2013

Matt Kenseth + Jimmie Johnson + Kyle Busch = 90%. Everybody else = 10%

For the second week in a row, Matt Kenseth and Kyle Busch finished 1-2.
For the second week in a row, Jimmie Johnson got a top 5.
These are the only drivers to start this year's Chase with two straight top-5s.

And just like that, the 13-driver Chase has been cut down to a 3-man battle.

Everybody else has some serious catching up to do. The latest numbers from our Chase prediction model give our newest expectations:

Matt Kenseth is again the big winner, with his title chances now sitting at 41%. His increased lead in the points standings gives him a bigger cushion in the model, which will come in handy in case he falters in an upcoming race.

Johnson, with two consecutive top-5 finishes, has kept his title chances steady at 27%.  He has been hovering in the 25-27% range since the Chase began. Even without winning races, he could still win a title on the strength of numerous top 5s.

Kyle Busch, with another second-place finish, crept forward to 22%, up from 20% last week.

Carl Edwards and Kevin Harvick were the big losers in the wake of Kenseth’s 16% gain (from 25% to 41%). Outside our top 3, Edwards and Harvick were the only two drivers last week with chances above 5%, but now the model shows them each at 3%.

It has very quickly become bad for the rest of the field: the ten drivers in the back have a combined title chance of only 10%. Think about that: we are only two races in, but there is a 90% chance of the title going to one of our top 3 drivers. Put another way: our three top drivers each have an average 30% chance of winning the title, while the ten others only have a miniscule 1% average chance.

If anybody were going to break away from the crowd, it's not surprising these would be the three to do it. Before we started the Chase, our prediction model had these same three drivers up top. No matter what, everybody else needs to step up soon, before this thing is over. For many drivers, it may already be too late.

Finally, in predicting how many points the champion will have, we are now looking at a 2380-2400 range, in line with our last two champions (2403 and 2400).

Consider using 2400 as a round number for what any driver will need to be champion. Take the example of 13th place Kasey Kahne, with 2040 points right now. For him to win the title, he’d need 360 points in the final 8 races, an average of 45 points per race. Since the winner of a race gets 47 points and second place gets 42, Kahne would need to finish top-2 in the final 8 races to be a threat to win. That would take a miracle, which is why the model puts Kahne's title chances effectively at 0 percent (0.042% to be exact). He should start preparing for next year.

Check in next week, after Dover, for the latest analysis.

Sunday, September 22, 2013

How Did My New Hampshire Picks Do?

Following up on this week's latest video, where I made my New Hampshire picks, let's see how I did:

New Hampshire Favorites
  • Jeff Gordon (led 36 laps and finished 15th)
  • Ryan Newman (won the pole, led 2 laps, finished 16th)
  • Clint Bowyer (led 5 laps and finished 17th)

Threats for First New Hampshire Win
  • Matt Kenseth (WON the race and led 106 laps)
  • Martin Truex, Jr. (led 98 laps but finished 10th)
  • Brad Keselowski (led 2 laps and finished 11th)

Thursday, September 19, 2013

Predicting the Chase Champion: These Numbers Still Say Jimmie Johnson is Leading the Chase

Predicting the Chase Champion: These Numbers Still Say Jimmie Johnson is Leading the Chase

Only one question matters now: “Who will win the Chase?”

We can accurately predict the likelihood of each driver winning the title, based on two factors:
(1) The current points standings
(2) Each driver’s full set of finishes this season.

We can also predict how many points the eventual champion will have.

More interestingly, we can update these predictions after each Chase race, based on the latest results and updated standings. This will show us exactly how much a driver’s title hopes changed, in percentage terms, from race to race.

Here were the chances for each driver to win the title, before the Chase started:

Jimmie Johnson, entering the Chase, was the model's favorite driver, due to the strength of so many upfront finishes in 2013. Remember, the percentage likelihood of winning the title may not match up exactly with the points standings. That’s because this model combines the points standings with the expectation of each driver’s performance.

Ryan Newman and Jeff Gordon, the two drivers who were added to the Chase after the fact, have very low odds of winning a title – notice them near the bottom of the chart. NASCAR did not significantly reduce the odds of the main contenders by including these two in the Chase.

Using the same prediction model as above, we can now shift our focus from the winning driver to the number of points needed to win.  It’s all calculated as part of the same process, simulating thousands of possible race results. Here is the expectation, prior to the Chase starting, of how many points the champion will have:

As the chart shows, we should expect the champion to end with a points total around 2360-2380. Compare that with what happened in the past two seasons: Tony Stewart finished with 2403 points in 2011 and Brad Keselowksi had 2400 points in 2012. If somebody gets on a hot streak during the Chase, you will see this chart shift to the upside.

We will be updating both charts weekly so you can see how the field fares race by race.

Now that the Chase has begun, here are the updated title chances after Chicago:

The winner, on the race track and in the model, is Matt Kenseth. His title chances leaped to 25% from 14% due to his win and increased points lead. Jimmie Johnson, at 26%, still has a slight edge over Kenseth though. Remember, despite Kenseth's points lead, the model currently prefers Johnson due to Kenseth's high number of poor finishes this year. Kenseth can change this prediction by consistently finishing up front, breaking the pattern of his early-season troubles.
Joey Logano and Dale Earnhardt, Jr. hold down the bottom of the chart, because of DNFs at Chicago. Their equipment failures and mediocre 2013 performance may have already sealed their fates: their title chances are currently below 0.1%.

Finally, our expected championship point total predicts the winner to be near 2360-2400 points.

Check in after this weekend’s race to see updated numbers.

Monday, September 16, 2013

How Did My Chicago Picks Do?

In last week's BSports Video, I gave my six top-ranked favorites to win at Chicago.

Here's how I did, listed in rank order for how likely they were to win:

  1. Jimmie Johnson (finished 5th after leading 40 laps)
  2. Matt Kenseth (WON the race and led 89 laps)
  3. Kevin Harvick (finished 3rd and led 2 laps)
  4. Kyle Busch (finished 2nd and led 67 laps)
  5. Brad Keselowski (finished 7th and led 2 laps)
  6. Ryan Newman (finished 10th and led 1 lap)
Not bad this week.  All six of these drivers led laps and finished in the top 10.  My top 4 choices all got top-5 finishes, and my top 5 choices all finished in the top 7. Can't do much better than this.

Wednesday, September 11, 2013

How Important is Richmond to Building Momentum for the Chase?

Among drivers, teams and the media, there is always talk about building momentum to start the Chase. Specifically, Richmond is the place for teams to demonstrate they are primed for contention. In fact, if you listened carefully to the ABC telecast on Saturday night, the announcers described how Dave Rogers, Kyle Busch’s crew chief, wanted to do well that night, so his team could enter the Chase “with momentum”. Also note these comments by Joey Logano after the race: “We’re gonna take this momentum we have getting into the Chase and let ‘er rip and have some fun.”

Today we examine how real this momentum is. Here are the questions we are trying to answer:
  • Does the winner at Richmond do significantly better in the Chase?
  • Does the eventual Chase champion come from being a top finisher at Richmond?

Here is the raw data:

The table shows us the answers to our two questions.
  • The winner at Richmond actually does very well in the Chase, with a median championship finish of 3 throughout the entire history of the Chase. In the 7 most recent years, the Richmond winner has a median championship finish of 2. This is a great sign for Carl Edwards, who won the race this past weekend.
  • As far as eventual champions go, they entered the Chase with a median 7th place finish at Richmond. Only once did a driver finishing outside the top 15 at Richmond go on to win the title (Jimmie Johnson in 2006, after finishing 23rd at Richmond).
Let's see how this applies to 2013: here is where the 12 Chase contenders finished this past weekend:

Recall that no driver has won the chase after finishing worse than 23rd at Richmond.  This would presumably diminish the chances for Clint Bowyer (25th) and Jimmie Johnson (40th). However, both drivers had uniquely rare incidents which caused poor finishes: Johnson missed all of practice and qualifying because of his baby’s birth, and Bowyer spun out his car (on purpose according to many analysts) late in the race. That would suggest all 12 drivers have a shot of winning the title.

In nine years of the Chase, six of the champions had top-7 Richmond Finishes. This should be good news for Carl Edwards (1st), Kurt Busch (2nd), Ryan Newman (3rd), and Matt Kenseth (6th). 

Overall, the data suggests that yes, momentum in fact does appear to be a factor in bridging Richmond to the final 10 races. Richmond winners are generally Chase threats, and the eventual Champions usually did well at Richmond. With his Richmond win, Carl Edwards should be a factor in the Chase.

Sunday, September 8, 2013

Video of Richmond Preview

Richmond Favorites
  • Kevin Harvick (finished 11th)
  • Kyle Busch (finished 19th)
  • Denny Hamlin (finished 21st)

Threats for First Richmond Win
  • Carl Edwards (WON the race and led 46 laps)
  • Ricky Stenhouse, Jr. (finished 10th)
  • Greg Biffle (finished 12th)

Wednesday, September 4, 2013

Why Dale Jr should have a 100% Chance of Making the Chase (and Keselowski 0%)

With one race to go before the Chase starts, only 6 spots in the top 10 have been clinched. That leaves 4 more spots still to be filled, with these 10 drivers mathematically eligible for them:

In today’s post, we look at these 10 eligible drivers, and using a simple forecasting method, we predict the chance each driver has of making the top 10.

For those of you who are too eager to wait, here is the final answer:

How did we arrive at these percentages?

One way to forecast how these drivers will perform this weekend is to consider their performance in the first 25 races this year.  This is a large enough sample size to suggest the full range of each driver’s results, based on their circumstances, equipment, and team quality this season. The table below shows the number of points scored by each driver in each race this season. For example, Dale Earnhardt, Jr. scored 42 points in the first race this season, 40 points in the second race, and 37 points in the third race, a sizzling start to the season.

Let’s assume this weekend’s Richmond race will have a similar profile to one of the previous 25 races this season.  In that case, we have 25 possible outcomes for what the points standings will be following the race, by adding the current points of each driver to the 25 theoretical results. The "T" in front implies the theoretical outcomes that could occur, so "T9" is how many points each driver would have if the results from the ninth race this season are repeated this weekend.

With these 25 possible outcomes, we can now rank the points standings for these 10 drivers.  Remember that only the best 4 drivers from this group will make the top 10. The "R" implies what the rankings would be based on our theoretical results, so "R9" is the ranking that would happen if the results from the ninth race this year are duplicated.

Based on these rankings, we can count how many times each driver finishes in the top 4 of this group, and divide by 25 to get his percentage chance of making the Chase:

So there you have it: based on the previous races this season, Dale Earnhardt Jr. would qualify for the Chase every time.  On the other hand, Brad Keselowski and Jamie McMurray would not qualify, given how they've performed in their previous 25 races.

Notice one interesting observation that emerges from the data: even the narrowest of points leads can translate into a significant edge in making the top 10. For example, Kurt Busch has a small 6 point lead over Jeff Gordon, but that’s worth an additional 16% chance of making the top 10.

The method we've seen here is just one approach to calculating the odds of making the top 10.  This approach is intuitive because it's based on real performance data from the season, and gives you a quick sense for what we should expect from each driver.

All this is good news for drivers like Earnhardt and Joey Logano, who should feel very confident about making the cut.  In contrast, drivers like Keselowski and McMurray will have to pull off their best performances of the year to have any chance of qualifying.  And for the viewing audience, the most interesting battle for the top 10 should be among the three drivers with chances closest to 50%: Kurt Busch, Jeff Gordon, and Kasey Kahne.

And yes, there is one fact we all know is true: anything is possible this weekend.  That's why they run the races while we sit and watch them.

By the way, in case you were wondering, the Richmond race from this spring is race #9. If history repeats itself this weekend, Jeff Gordon would knock Greg Biffle out.