Breaking Down The Odds For NASCAR's Playoff-Bubble Drivers
It's going to be a cruel numbers game for someone.
A few weeks ago, I ran some relatively back-of-the-envelope numbers on what it would take for Chase Elliott to make the 2023 NASCAR Cup Series playoffs. While there was a remote points-based path to a postseason berth, the far more straightforward route under NASCAR’s wins-obsessed playoff system involved grabbing a checkered flag before the Aug. 26 cutoff to decide which 16 drivers compete for the championship.
Obviously, Elliott is still working on that first win of 2023. But while he does that, I figured it might be fun to do the same exercise for every driver hunting for wins down the stretch of the regular season. And that meant I needed to level up the analysis from a one-off to a full series of simulations for the entire field. I’ll explain how it all works at the bottom of the post — but first, let’s look at the big takeaways from the endeavor.
For one thing, the win-and-you’re-in path is getting a little dicey even for the top names who aren’t in the W column yet. According to my method, Elliott still has a 52% chance of getting at least 1 win over the rest of the regular season — with nine races now remaining — but he’s the only zero-win driver with better than a 50-50 shot at nabbing that first win. Kevin Harvick (47%) and Brad Keselowski (42%) aren’t far off, but then the odds fall to 21% for Alex Bowman and 14% for Noah Gragson, meaning most of the winless field is pretty likely to stay that way by the time we reach judgment day for the playoffs.
Things could get extra-desperate as we approach the cutoff race at Daytona, the final chance for a driver to win his way into the playoffs. For some of the top-ranked drivers in that zero-win group — guys like Chris Buescher, Bubba Wallace and Daniel Suarez — there’s roughly a 90% chance that they still will be trying to get rid of a goose-egg in the wins heading into that race.
And then we get into the standings crunch created by NASCAR’s playoff format itself. This is a great explainer from last season, about how convoluted things can get if 16 or more drivers win a race in a year, thereby sort of breaking the playoff system. We in the racing media always talk as though winning a single race “locks” a driver into the playoffs, but that is not necessarily true — and there’s some chance we see this tension play out over the next two months.
After Ross Chastain won in Nashville last weekend, there are now 11 unique race winners in the 2023 season so far. That currently leaves 5 playoff slots for non-race winners to be sorted out by the points standings, as is usually the case at the end of the regular season. But my simulation says there is a surprisingly high chance that we keep adding new winners to the ranks over the next nine races. On average, 4.4 winners were added from the ranks who hadn’t already won one of the first 17 races, versus 4.6 from the list of existing winners.
For each remaining race, here are the odds that the winner would be a driver who had not already won through the previous race of the season:
The mix of races left is contributing to this potential chaos. We all know that Daytona can produce wild results, while the Chicago Street Circuit is a brand-new track — and track type! — for NASCAR, in which we truly have very little sense of who might win. (As you’ll see below, I tried my best to balance this uncertainty against the track records on road courses, where track records are far more predictive… but ultimately I’m just guessing like anyone else.) The only previous event there was staged on iRacing in 2021 and was won running away by James Davison, who has an average finish of 32.4 in 35 “real” Cup Series starts. It’s a perfect recipe for a new race winner — and perhaps a very random one, at that — to enter the fray.
Additionally, 2023’s winless ranks include some of the all-time winningest active drivers in NASCAR — Harvick and Keselowski — plus one of the best by winning percentage in Elliott. So the particulars of this group of bubble drivers is also begging for the doomsday scenario of Too Many Winners to play out.
Add it up, and according to my simulations there is a 31% chance we see at least 16 distinct race-winners this season, and a 9% chance that there will be more winners than available playoff slots for the first time since NASCAR introduced this postseason format in 2014.
The average simulation saw 14.9 unique winners, leaving only 1.3 slots freed up to be determined by the points on average.1 More likely than not, the doomsday scenario will not come to pass — but one of these seasons it will, particularly if NASCAR keeps trying to level the playing field between teams and introducing the chaos of specialty events.
So where does that leave the 2023 playoff bubble brigade? Elliott remains in decent enough shape, given his better-than-even odds to win a race from here on out. But even at his per-race points pace from when he’s been available, he’s tracking to finish the regular season ranked 17th overall, and sixth among current non-winners. That wouldn’t even be good enough to make the postseason now, much less after more drivers gobble up potential playoff slots by converting from non-winners to winners. Simply put, Chase needs a WIN.
Harvick is arguably in a much more solid spot than Elliott. Not only does he have that 47% chance of winning at least one of the next nine races, but he is also comfortably No. 1 in points among current non-winners. Even if he doesn’t win, there’s a 69% chance that at least 1 playoff spot will be available on points, and that top spot will very likely belong to Harvick.
As for Keselowski, he has that relatively favorable 42% chance of winning at least one remaining race, plus he sits second in the standings among current non-winners — and there’s a 40% chance that at least 2 playoff slots will be filled with non-winners. So he sort of has the poor man’s version of Harvick’s situation, which could be a lot worse.
But I’m less certain about the fates of Buescher, Wallace, Bowman and Suarez. None have greater than a 21% chance to win over the rest of the regular season, and with the exception of Buescher (who is dueling neck-and-neck with Keselowski in the points), none are particularly close to the top 3 among non-winners by points. Since there’s only a 3% chance that four or more playoff slots will go to non-winners, it’s very likely that one or more drivers from that group will get caught up in the numbers game and miss the playoffs without a win.
Methodology
To determine each driver’s chance of winning each race, I simplified each remaining field to the top 36 qualified drivers on points (swapping in J.J. Yeley for Cody Ware, who is suspended indefinitely for domestic assault) and then calculated probabilities based on two factors:
His previous career winning percentage at the pertinent common track type (i.e., Superspeedways, Intermediate Ovals, Road Courses, Short Tracks, etc.).
A naïve probability where all 36 drivers have an equal win probability.
The amount of weight given to each factor is determined by a combination of the driver’s previous career sample size within that track type, plus an empirically derived “padding” factor for the naïve probability at each track type, designed to minimize the Brier score for predictions of races since 2007.
What follows are the track types and padding values by track type:
(The 70-race padding value for street courses was admittedly somewhat arbitrary. On the one hand, the value for road courses — the closest analogue for a street track — was far lower, at about 9 races. On the other hand, this is the first street race in Cup Series history and NBC’s broadcasters stressed on Sunday’s post-race coverage that road racing was an imperfect comparison at best for the street race. Combine that with the relatively random-seeming order of finish for the iRacing event, and I set a value that was just slightly lower than that of Superspeedways.)
After blending the career observed winning percentages by track type with the naïve probabilities via the padding method, each race total was normalized so the sum of all drivers added up to 100% win probability. The rest of the regular season was then simulated 10,000 times to attain the results found in this story.
Filed under: NASCAR
Those two numbers don’t add up to 16 because the number of unique winners can exceed 16, while the number of slots determined by points can never be below zero.