On The Teardown this week, Jeff Gluck put a question to Jordan Bianchi that this site cannot answer.

"Does it at all hurt JGR that they have three cars that are all trying to chase Larson and they keep sort of taking points away from each other that could be wins?" Gluck asked. "Where, like, Larson is not really battling with his teammates."

Bianchi's answer split the garage in two. Penske has two drivers in the fight and one who is not, which makes the third a helper: "At that point, those guys are effectively wingmen. Their mission is to help Logano and Blaney." Hendrick is simpler, in his words "just Larson or bust." Joe Gibbs is the awkward one, with four drivers who all need the same thing. "At some point you kind of got to draw a line in the sand and say like, 'I'm not a teammate anymore.'"

He is describing something real. Joe Gibbs holds four of the sixteen Chase seats, three of them in the top four: Hamlin second, Bell third, Briscoe fourth, Ty Gibbs ninth. The race they were discussing ended with Briscoe winning and Bell second, a JGR one-two.

The model has nothing to say about any of it.

Strength here is built one driver at a time, from his own rating at the track, his record on tracks of that shape, and his recent form. There is nothing in it about the car next to him. That is not a discovery, either. It is listed on the about page among the things the model does not know, next to qualifying position and rookies: how teammates' results move together. Every row of results it reads says which team the car belongs to. Nothing uses it.

Here is how often that matters. In the thirty-two races of 2026 so far, one organization has put two or more cars in the top five in twenty-nine of them. Two or more in the top three, in thirteen. Joe Gibbs alone has doubled up in the top five twenty times, and swept the podium outright at Nashville and Chicagoland.

Be careful about what that does and does not prove. The model is not blind to JGR being fast: all four drivers carry high ratings earned in that equipment, so their numbers are high. For Phoenix on Sunday it gives Joe Gibbs a 28.3% chance that one of its four wins, the largest figure in the garage, against 23.8% for Hendrick and the same for Penske. What it cannot produce is the togetherness. It has no way to express a Sunday where all four JGR cars are fast at once, or all four are off, which is the kind of day that settles a season. How wide that gap runs has not been measured here, and saying otherwise would be inventing a number.

Which brings up Talladega on October 25, the eighth of the ten Chase races, and the place where this stops being academic. Bianchi's wingman is literal on a superspeedway, where a teammate's push is the difference between winning and finishing fourteenth. It is also the one kind of track the model already admits it cannot fit, because a drafting race has a predictable fast car and a random pack. So the race that turns most on who helps whom is the race the odds are least equipped for, twice over: no teammate term, and no demonstrated skill in the draft.

One detail cuts against the podcast's framing. "Larson or bust" is a statement about the championship, not about speed. At Phoenix the model's best Hendrick car is not Larson at 9.2% but William Byron at 10.3%, and Byron is 208 points back without a win this season. The fastest Hendrick entry on Sunday is the one the team's plan is not built around.

A shared team effect is a known fix rather than a mysterious one, and it sits next to the other thing these odds cannot see: that a driver who changes teams brings his old car's record with him. Both wait for the refit after Homestead. Until then, when the numbers say Joe Gibbs has the best chance on Sunday, they are counting four drivers rather than one team.