Chase Analytics: Kevin Harvick enters the playoffs as favorite

Interesting, I still think these things are less useful in racing than other sports. So much of the past driver performance that goes into the measure is based on the quality of the cars at the time the races were conducted. I do find it hilarious that some of their simulations (.016% of them) showed Buescher coming out on top.
 
Does the analytics algorithm include a figure for debris or tires bouncing down pit road?
 
Being unbeatable at Phoenix definitely skews it towards his favor.
 
Interesting, I still think these things are less useful in racing than other sports. So much of the past driver performance that goes into the measure is based on the quality of the cars at the time the races were . I do find it hilarious that some of their simulations (.016% of them) showed Buescher coming out on top.
conducted.

I thought the comments about Shrub were interesting and accurate as he could be a 2 time lottery winner or crash out in the first 3 races.
 
Does the analytics algorithm include a figure for debris or tires bouncing down pit road?

Dubious debris cautions are in but no decision has been made yet the bouncing tire thing. Nascar got roasted for calling 1 unseeded tire caution at the Glen and then got roasted again for looking the other way when one was needed.
 
I'll fearlessly predict that Kurt Busch will top 5 himself (aided by a win) to the final round and a second championship.

<steps back to don asbestos jumpsuit>
 
Harvick seems like a logical choice, but this format is far to unpredictable to forecast a winner with any degree of certainty. Hopefully he won't have to kamikaze half the field at Talladega again this year to stay alive.
It depends on whether or not he has to. If he has to wreck half the field to stay alive he will. That's Harvick 101
 
I'm glad this weekly analysis has returned. Of course, it is only as good as the model used to simulate each race. But it does present an interesting analysis of the probabilities, and how they stack up over all the remaining races. Good stuff, IMO.

The weekly Pit Rho analysis of win probabilities has consistently shown Harvick as a heavy favorite, every week. For example, right now Harvick is shown with a 20.2% probability to win Chicagoland, nearly double the second highest driver (Logano 11.9%). I don't think Harvick is as strong as Pit Rho thinks he is. But I have not done any analytical work on it, so my argument is just based on gut feel.

PitRho2016_Before_Chicagoland.png


Las Vegas sportsbooks list Harvick as the favorite at 5/1 odds, so if you believe the Pit Rho probability of 34%, that would make a good bet. Keselowski at 8/1 and Kurt Busch at 20/1 are in line with the Pit Rho analysis. All of the other drivers are too expensive in Vegas for bettors who believe the Pit Rho probabilities. (I'm not a bettor, but I do follow the odds from time to time.)
 

Attachments

  • PitRho2016_Before_Chicagoland.png
    PitRho2016_Before_Chicagoland.png
    31 KB · Views: 172
Harvick seems like a logical choice, but this format is far to unpredictable to forecast a winner with any degree of certainty. Hopefully he won't have to kamikaze half the field at Talladega again this year to stay alive.

He shouldn't have had to last year. NASCAR should've stuck to the rule they wrote earlier that week instead of changing the rule at the end of the race. That race, and Phoenix (20 minute caution for a single car wreck and calling a race at around 6pm local time) will never sit well with me. Plus, Kyle being in the Chase last year and Tony and Chris Buescher in it this year.
 
I dont bother with these things, as Aunty pointed out, one "entertainment" caution can change everything.
 
Such old school crap . Using numbers from the best car all year , the best driver all year , with the best crew chief all year and ranking him first , while you take the worst driver , with the worst car all year and rank him 16th . Did this guy go to school to learn that ? Neither one will make the final four or whatever basketball term they are using today.
 
It only takes 1 Kenseth to change everything or is his mood in the calculations?:sarcasm:
 
"we ran the numbers" - what numbers? I wonder what bias they used, consistency, finishing positions, bad luck, what parameters? The Gibbs guys have been running up front most of the season and I'd expect one of them to have a top chance. I do think that Harvick is one of the favorites.
 
The past few years there seem to more "perfectly timed cautions" for a few of the "superstars" than ever before. Reality TV? The way a few of the drivers "fight" it would seem so.
It's football season, I can watch the highlights of the race on Monday in a 1/2 hour if I so desire.
I dont bother with these things, as Aunty pointed out, one "entertainment" caution can change everything.
 
Everyone is writing off Buescher and the 34 bunch, saying they have no shot. Well maybe the joke's on us and they have been sandbagging, saving their best for the Chase. Watch them come out and win half the races like Stewart in 2011. Surprise! Nobody will see it coming :D
 
I gotta say I love the storylines going into this chase. All of the contenders have their unique strengths and weaknesses that will influence the strategy of the whole thing. Some of them:

  • Harvick - Most consistently fast driver and a lock at Phoenix, but his pit crew is near the bottom of the chase grid. If it comes down to a late stop in a must-win race, he's screwed.
  • Brad K - The Deuce seems to trade raw speed for fuel mileage. He can win that way only if the races play out in his favor, otherwise it will be hard to keep up. His ace in the hole is Talladega though
  • Truex Jr - Cole Pearn can cook up up a monster of a car. The one he built for Charlotte was the fastest, most dominant car of the year. Too bad Truex wrecked it at Pocono. The big question is, if and when the 78 team decides to wheel out the annihilator again, is Truex a good enough driver to handle it and can it be reliable enough to get him to Homestead?
  • JJ - What the heck happened? Reliability issues, lack of speed, everything is working against him. Many have picked an early exit for him.
 
"we ran the numbers" - what numbers? I wonder what bias they used, consistency, finishing positions, bad luck, what parameters? The Gibbs guys have been running up front most of the season and I'd expect one of them to have a top chance. I do think that Harvick is one of the favorites.
Yeah, the JGR guys are winning more but Harvick has been the most consistent. I wouldn't count Keselowski out either. I think it'll be Harvick, Kez, and two of the JGR/FRR drivers in the championship four.
 
Yeah because plate races take no skill whatsoever and big wrecks absolutely never happen at other tracks.
 
Harvick had that race blown wide open before the weekly late race caution.
I think he's certainly the favorite there but as we saw in that race and in last fall's all it takes is an untimely caution or weather to throw a wrench in the works.
 
I think he's certainly the favorite there but as we saw in that race and in last fall's all it takes is an untimely caution or weather to throw a wrench in the works.

Or NASCAR deciding to spend 20 minutes cleaning up a single car wreck and calling the race at the first drop of rain because Junyer is leading.
 
I just wish they would lock the field under caution! When they pit they come back out in the same position! I could care less about moe the jackman or bubba the front tire changer!


Sent from my iPhone using Tapatalk
 
The Pit Rho probability model for Chase week 2 has been released, and it continues to show Harvick as their favorite for the Cup title. Link to article. I'll point out that the Pit Rho simulation model loves the 4 team and rates the 4 as a heavy favorite just about every week. Personally, I'm pulling for karma to bite Harvick hard on his whiny little ass, but that's just me..:mad:

From the article:
“Harvick and Johnson are fast enough to make it to the next round,” said Josh Browne, a former Sprint Cup crew chief and now co-founder at Pit Rho, “but they’ll need to limit mental lapses to make it to Homestead with a shot at the title. They both had fast cars Sunday, but mistakes took them out of contention for the win. The current Chase format really penalizes mistakes.”

The interesting area right now is the cutoff for 12th between Jamie McMurray (right now at 78 percent) and Kyle Larson (at 59 percent). The two Ganassi teammates are straddling the border for 12th place probability. Larson has a lower chance of making the next round, but a higher chance of winning the title. That’s because Larson has had more volatile results this season, including a win and running up front late in many races. McMurray has been more of a steady burn.


PitRho-2016-chase-2-After-Chicago.png
 
The Pit Rho probability model for Chase week 2 has been released, and it continues to show Harvick as their favorite for the Cup title. Link to article. I'll point out that the Pit Rho simulation model loves the 4 team and rates the 4 as a heavy favorite just about every week. Personally, I'm pulling for karma to bite Harvick hard on his whiny little ass, but that's just me..:mad:

From the article:
“Harvick and Johnson are fast enough to make it to the next round,” said Josh Browne, a former Sprint Cup crew chief and now co-founder at Pit Rho, “but they’ll need to limit mental lapses to make it to Homestead with a shot at the title. They both had fast cars Sunday, but mistakes took them out of contention for the win. The current Chase format really penalizes mistakes.”

The interesting area right now is the cutoff for 12th between Jamie McMurray (right now at 78 percent) and Kyle Larson (at 59 percent). The two Ganassi teammates are straddling the border for 12th place probability. Larson has a lower chance of making the next round, but a higher chance of winning the title. That’s because Larson has had more volatile results this season, including a win and running up front late in many races. McMurray has been more of a steady burn.


View attachment 21040

I really wish they would release how they do their calculations. My background is in statistics, and I suspect there are some big underlying problems here. It seems they take each driver's average performance at a track and add in some randomness. They then run 100,000 simulations and see where things shake out. The results are heavily dependent on how they account for that randomness. This is not how most sabermetric baseball statistics are created.
 
I really wish they would release how they do their calculations. My background is in statistics, and I suspect there are some big underlying problems here. It seems they take each driver's average performance at a track and add in some randomness. They then run 100,000 simulations and see where things shake out. The results are heavily dependent on how they account for that randomness. This is not how most sabermetric baseball statistics are created.
I really don't know anything about Pit Rho or how sophisticated or rigorous their model may be. They came out with these multi-colored probability charts last year for the chase, and NBCSN.com did a short article each week. It is an interesting way to summarize how the probabilities cascade downwards for each elimination round.

Pit Rho has a web site for fantasy players ( https://fantasy.pitrho.com ). I have no interest in fantasy sports, but I signed up for the free service level, which shows their probabilities to win each week. I get the impression the weekly model is influenced by: type of track or maybe the specific track for this week or maybe both; how the car has performed in general over the last one or two years; and weekly qualifying results have a substantial impact.

Obviously, there must be a random element as well, and I know that Pit Rho has analyzed randomness on a per track basis.

PitRhoRandomness.jpg


The bottom line is they rate Harvick the favorite almost every week, Kyle Busch second, and a big drop in probabilities after that. My gut says these two are over-valued by the model. But maybe that is my bias showing through, as those two are the drivers I dislike the most..:mad: So I don't vouch for Pit Rho, and for Nascar quant, I prefer David Smith of Motorsports Analytics (but he also does not reveal his proprietary model parameters).
 
I just don't get why the media gush over Harvick week after week, regular season or chase. All he ever does is quietly sit somewhere inside the top 10 the whole race, rarely going to the front or dominating except at Phoenix. He's the epitome of a points racer and for some reason that irks me. He's got the worst or 2nd worst pit crew in the chase to boot too. Sure, he's going to homestead simply because he's good enough to point his way into the round of 8 then win at Phoenix again, but I have next to no confidence with him winning it given the competition he's likely to face.
 
I really don't know anything about Pit Rho or how sophisticated or rigorous their model may be. They came out with these multi-colored probability charts last year for the chase, and NBCSN.com did a short article each week. It is an interesting way to summarize how the probabilities cascade downwards for each elimination round.

Pit Rho has a web site for fantasy players ( https://fantasy.pitrho.com ). I have no interest in fantasy sports, but I signed up for the free service level, which shows their probabilities to win each week. I get the impression the weekly model is influenced by: type of track or maybe the specific track for this week or maybe both; how the car has performed in general over the last one or two years; and weekly qualifying results have a substantial impact.

Obviously, there must be a random element as well, and I know that Pit Rho has analyzed randomness on a per track basis.

View attachment 21042

The bottom line is they rate Harvick the favorite almost every week, Kyle Busch second, and a big drop in probabilities after that. My gut says these two are over-valued by the model. But maybe that is my bias showing through, as those two are the drivers I dislike the most..:mad: So I don't vouch for Pit Rho, and for Nascar quant, I prefer David Smith of Motorsports Analytics (but he also does not reveal his proprietary model parameters).
I've seen Andrew Maness tweet out that list before but not with the figures. I'm not surprised that the difference between the most random non-plate oval (Bristol) and the plate tracks is bigger than the gap from the most random non-plate oval to the least (Kentucky). I've been told on here that plate tracks have no more of a random element than any other type but that certainly isn't the case.
 
I just don't get why the media gush over Harvick week after week, regular season or chase. All he ever does is quietly sit somewhere inside the top 10 the whole race, rarely going to the front or dominating except at Phoenix. He's the epitome of a points racer and for some reason that irks me. He's got the worst or 2nd worst pit crew in the chase to boot too. Sure, he's going to homestead simply because he's good enough to point his way into the round of 8 then win at Phoenix again, but I have next to no confidence with him winning it given the competition he's likely to face.
If he were the epitome of a points racer he wouldn't have 1200+ laps led and the best driver rating by a significant margin.
 
All he ever does is quietly sit somewhere inside the top 10 the whole race, rarely going to the front or dominating except at Phoenix.
That is the definition of a highly experienced, very successful racecar driver.

Any reasonably fit, co-ordinated person with good vision and depth perception can learn to go fast in a relatively short time.

It takes forever to learn how to race.
 
I really wish they would release how they do their calculations. My background is in statistics, and I suspect there are some big underlying problems here. It seems they take each driver's average performance at a track and add in some randomness. They then run 100,000 simulations and see where things shake out. The results are heavily dependent on how they account for that randomness. This is not how most sabermetric baseball statistics are created.
I defer to your credentials and experience here.

I have found that for me, the best way to interpret these "statistics" is to ignore them because .... racing will reward you or screw you over as it sees fit. Any attempt to factor that into an algorithm is a waste of time.

jmo, of course.
 
this format is far to unpredictable to forecast a winner with any degree of certainty.

Nobody could, but you could probably pick the five best performing cars/teams, and have the winner within that group....which is amazing given this format. Shows how good the best teams are.
 
Back
Top Bottom