About The Cup Chase
The Cup Chase does two things: it publishes win and top-5 odds for the next NASCAR Cup Series race, and it lets you simulate the season race by race, through the Chase to a champion. Both run on the same model. When you press simulate, you're running the engine that produced the published odds.
How the odds are made
Every driver gets a strength rating for each track, built from their results there since 2022 and blended with their recent form. A driver with only a handful of starts at a track leans on how they run at that kind of track - short track, intermediate, road course, superspeedway.
The model then runs the race 50,000 times. Each run draws a finishing order from those strengths with randomness added, including the chance of a crash or a mechanical problem that wrecks a driver's day. A driver's win percentage is the share of those runs they won; top 5 is the share where they finished fifth or better.
Ratings only ever use races that happened before the one being predicted. No result is ever part of its own prediction.
How well it works
The model was fitted on the 2023, 2024 and 2025 seasons, then tested on 2026, a season it was never tuned on. Across all 36 races of 2026 it beat a uniform guess - every driver in the field given the same chance - by about 18%, measured by log loss on who won. Its win percentages held up too: grouped by how likely the model said they were, drivers won within half a percentage point of the predicted rate in every group.
That's a real edge, not a crystal ball. A 20% favorite still loses four races out of five.
Where it's weak
Superspeedways. At Daytona, Talladega and Atlanta the field runs in a tight draft and one push can reshuffle the top ten; the finish is close to a lottery. Across 2023-2025 the model's edge at these tracks was less than half what it was at short tracks and intermediates, and across the six drafting races of 2026 it came out level with guessing. Those six swung wildly, badly wrong at two of them and clearly right at two others, which is nowhere near enough to say where the truth sits. Six 2027 races are on drafting tracks, the Daytona 500 among them, and their odds carry a caveat on the page.
Top-5 odds are less reliable than win odds. The model tends to be too sure about the drivers it rates highest, so a top-5 number above 55% is the least trustworthy number on the page.
It also doesn't know some things that matter: where a driver qualifies, how teammates' results move together, or anything about a rookie beyond the field average. Championship odds aren't published at all; tested on the 2026 Chase, they were far too confident.
Not betting advice
The Cup Chase is for entertainment. The odds are a model's estimates, top-5 odds especially, and none of it is betting advice. The site is an independent project and isn't affiliated with NASCAR, its teams or its sponsors.
Who runs it
The Cup Chase is built and run by Matt Steven, a longtime NASCAR fan who has been modeling Cup Series data for more than a decade and previously worked as a data scientist before transitioning to the world of sports media. He built the model, the pipeline, and the site himself.
Every prediction is published before the race and scored afterward, win or lose. Odds are updated once a week, after each race.