Two days ago I posted the girls forecast for the 2026 CIF state championship, including who the model projects to qualify for Nike Cross Nationals.
Today the boys model goes up. Here is what it says. Check out FRESNO here.
Division 2 is the race
Cooper Stream of Vacaville projects 14:46.2. Aidan Antonio of Woodbridge projects 14:47.1. They are the two fastest returning boys in California, separated by nine tenths of a second, and they run in the same division.
Their teams are just as close. Woodbridge projects 86 points, Vacaville 88. Woodbridge wins the division in 37.6 percent of the ten thousand simulations and Vacaville in 32.8.
The interesting part is what happens when you change how deep the scoring goes. Vacaville and Woodbridge are the top two teams whether you score three runners, four, or five, but they trade places:
scoring depth 1st 2nd three Vacaville, 46.3% Woodbridge, 24.7% four Vacaville, 44.2% Woodbridge, 31.3% five Woodbridge, 37.6% Vacaville, 32.8%
Vacaville has the stronger front three. Woodbridge has the better fifth man. Score it the way the state meet actually scores it and Woodbridge wins by two points, which is not a lead.
And then there is Jesuit. Score the division through three runners and they project third at 46 points, winning it outright in 12.8 percent of simulations. Score it through five and they vanish from the standings entirely, because they return three athletes and scoring takes five.
Two of the eighteen fastest returning boys in the state are on that team: Lucas Alberts eleventh at 15:07.5 and Matthew Ogilvie eighteenth at 15:10.4. Jesuit is a historically strong program that graduated a lot, and programs like that reload. The model has no way to know that. What it can tell you is that the front of their roster is already good enough to win the division, and the only question is what fills in behind it.
So Division 2 is a three-team conversation that the standings only show as a two-team race.
Division 1
San Clemente’s to lose at 74.2 percent. They return seven athletes, two of them in the projected top sixteen, and depth like that is what produces a number that high.
Mountain View is the only other team above ten percent at 16.0. Trabuco Hills at 6.0 and Crescenta Valley at 3.4 are live but need help.
Mario Montoya of Redondo Union is the individual to watch here, third in the state at 14:50.6 and a junior, the fastest returner in California who is not a senior.
Division 3
Campolindo is an overwhelming favorite at 72.9 percent and 46 points, with Clark Gregory fifteenth in the state and Caden Ehrhorn twenty second.
The reason to watch this race anyway is that Hart moved up from Division 2. They project 77 points and 11.5 percent, tied on win probability with St. John Bosco at 80 points. Bosco returns seven, which is the deepest roster in the division after Campolindo’s.
Hart arriving from a division they used to contend in makes Division 3 the strongest field in the state below the favorite.
Division 4
Menlo moved up from Division 5, and they land as the favorite at 67.6 percent and 55 points, with Henry Hauser fourteenth in the state.
Chaminade at 22.0 percent is the only team that projects to give them real trouble. La Canada at 6.8 is the only other one above two percent.
Oaks Christian is the team the standings undersell. They return Sterling White, who projects 15:02.5 and is eighth in the state as a sophomore, the only underclassman anywhere near the front, plus Arizechkwu Nwosu at twenty fourth. Two projected top twenty five runners in Division 4.
Division 5
The most lopsided race in the state. San Francisco University wins it in 91.9 percent of simulations at 46 points.
St. Margaret’s projects second at 110.5 and no team in the division breaks four percent. Everything after first is racing for the podium.
The NXN picture
California does not run a regional meet for Nike Cross Nationals. The state meet is the qualifier. Score every division together and the top two teams go to Portland, with a third or fourth sometimes taken at large when the race is close enough.
So the combined standings are the NXN projection.
Campolindo, 161 points. San Clemente, 168.
Seven points between them for two spots that both currently belong to them. Campolindo finishes first in 40.1 percent of simulations and San Clemente in 35.6, which is close to even for the top line and mostly beside the point, since both go.
Then the field falls off. Third projects at 283, a hundred and fifteen points back with nobody in between. The two best teams in California run in different divisions and never face each other until the scores are merged.
Neither Woodbridge nor Vacaville projects into NXN position, at seventh and eighth. Winning Division 2 and qualifying for Portland are two different achievements this year.
The at-large race
Third and fourth are separated by two points. Mountain View projects 283, St. John Bosco 285, Hart 298.
Their win probabilities say the same thing another way: 4.9 percent, 4.7, and 2.9. Three teams inside fifteen points and two points of probability. If an at-large bid is in play, it comes from this group, and the model cannot tell you which one.
Worth saying that the top two are not locked either. Campolindo finishes top three in 78.3 percent of simulations and San Clemente in 76.2, so each has roughly a one in four chance of finishing outside it on the day.
The individual NXN spots
Athletes on non-qualifying teams can still go. Ranked by projection, the boys most likely to be in that conversation are:
Cooper Stream, Vacaville, 14:46.2
Aidan Antonio, Woodbridge, 14:47.1
Mario Montoya, Redondo Union, 14:50.6
Michael Scarince, South Pasadena, 14:55.1
Jaxon Northcutt, Mt Carmel, 14:55.5
Adrian Cantu of Saugus is next at 14:59.0.
The useful thing here is that this list does not move. Whether two teams qualify, three, or four, the same five names sit at the top of it, because Mountain View, St. John Bosco and Hart have nobody in the projected top twenty. The at-large team race and the individual race are independent this year.
What the model cannot see
FRESNO stands for Forecast of Returning Entrants, Simulated Not Observed, and both halves are caveats.
It only knows athletes who ran the 2025 state meet. It has never heard of the incoming freshman who is about to run 15:30, or the runner who missed his qualifying meet hurt and is now the best on his team. Transfers are in there when I know about them. If you know of a boy who has changed schools and is showing up on the wrong team, put his name and the move in the comments and I will add him.
Jesuit is the clearest case, but Beckman returns three and La Serna returns two, and neither appears in the standings either. All of them will reload, and when those athletes race this fall the model will see them.
One note on the times. The fastest projection on the page is 14:46, and somebody will run faster than that in November. Somebody does every year. A projection is the middle of a range and the middle is never the fastest thing that happens.
Does it work
Tested the same way as the girls model. I hid the 2025 season completely, built and tuned everything on the years before it, then let the model predict 2025 once.
Average error came out just under 20 seconds per athlete.
That number is not comparable to the girls number and I want to say so before anybody puts them side by side. Boys run faster and in a tighter pack, so everything measured in seconds comes out smaller on this side for reasons that have nothing to do with whether the model is any good.
The comparable test is the floor. If you assumed every returning athlete simply ran three percent faster than last year, how close would that get you? Nobody forecasts that way, but it is the minimum a model has to beat to be worth building. The girls model beat that floor by about 16 percent. The boys model beats it by about 9.
That is the honest number and it is the smaller of the two. One shot, on a season the model had never seen, published regardless of what it looked like.
One thing worth its own post
When I put up the girls forecast I said the boys model would be built from scratch, because boys and girls develop differently enough that nothing carries over. That was a rule about how to build it, not a prediction about what I would find.
Building it twice turned the rule into a result. Several quantities came back materially different on the boys side, and one of them is different enough that a sentence on the girls methodology page was wrong. Not wrong about girls. Wrong for being stated as a fact about cross country when it was only ever a fact about girls.
That is getting its own post rather than a paragraph here.
What is next
Every race an athlete runs at Woodward Park feeds his projection, so after Clovis in October this will know more than it does today. I will post updates as that happens.
The full methodology, including the parts I got wrong, is linked on the dashboard.