Compliance notice: A free game: the site's forecasts are generated by a model, and a score is real only where its source is named. Nothing in the game takes money or accepts a wager.

Free high school football predictions

Friday night, called.

Claim your school, call every game in your conference this week and next, and see who got it right on Saturday. Games further out open on the Thursday their week comes up, for everyone at the same time. Start a league with the other parents in the bleachers. Free to play, no email, and no money ever changes hands in the game. For anyone 13 and older.

Look up any school

See this week's slateStart a league

The model has called 77% of winners this season. Think you can beat it?

Call a game

The week's top matchupsTwo strong teams the model can’t separate

MarginWin chance
  • Northwestern(11-0)Division 4
    @Freedom(11-0)Division 4
    Fri, Nov 6 · model has Freedom by 3.5 · both unbeaten
  • Lomira(11-0)Division 6
    @Edgar(10-1)Division 6
    Fri, Nov 6 · model has Edgar by 2.5
  • Racine Case(10-1)Division 1
    @Muskego(10-1)Division 1
    Fri, Nov 6 · model has Muskego by 2.5

See the whole slate

The model called 80% of winners the week of Oct 26 (884 of 1,110) — 77% across all 6,845 graded calls since Sep 21; every call made before kickoff, and toss-ups aren’t counted. Its record

Where your state stands

17 live9 asked, awaiting an answer24 not published

How it works

  1. Claim your school

    Pick a name and your school. That is the whole sign-up: no email, no card. Your school's game leads every page from then on.

  2. Call this week and next

    Every game in your conference this week and next, with the model's call and its win chance beside each side. Tap the side you think wins before kickoff. Games further out open on the Thursday their week comes up, for everyone at the same time.

  3. Start a league

    Send one link to the other parents. Two of you or the whole bleacher. Your own school's games count double, and upset calls break ties.

  4. Report from the stands

    At the game? Put in what the scoreboard says. It shows for everybody once other people there agree, and it's what the table is graded on come Saturday.

Want to read ahead? Season Outlook lays out the rest of your school's season with the model's win chance and margin, week by week. It's for reading: picks on those games open for everyone when their week comes up. Supporter extras are free for everyone this season.

What's real here, and what isn't

Worth saying straight, because the schools are real and a score beside them is real only where its source is named.

  • RealThe schools — 5,762 real programmes across 51 states, with their classifications, and real schedules and results in 17 of them so far
  • RealThe schedules and results — from the source named beside every score: a state association's public results, a credited local publication, a coach, or several people in the stands who agreed. Live in a growing list of states
  • Not realThe predictions — the model's numbers, rebuilt every Saturday from real results. A forecast, never a fact
  • Not realThe units — a score, not money; nothing can be bought, cashed, transferred or redeemed

A score is shown only where its source is named: a state association's public results, a credited publication, a coach, or several people in the stands who agreed on it. Where a real game is not on file yet, a placeholder is generated so there is something to predict, and it never carries a score. The site takes no money, accepts no wager, pays nothing out, and is not connected to any school, district or athletic association.

For athletic directors and associations

  • Every schedule and score names its source and links to it. Only facts are used: the two schools, the date, the site, the kickoff and the final score. No logos, rosters, player names, photos or statistics.
  • Corrections are made the same day. Ask us to stop, and we stop reading your data that day and take it off the site.
  • Free, and it stays a game: no money changes hands and nothing is paid out. Points are only a score.
  • Players must be 13 or older. Players under 18 can’t post in game threads, and their names are hidden on public boards.

Write to partnerships@hometownline.com. See every source we credit and how we keep it a game.

Football margins are lumpy. Most models smooth them.

Margins from 40,000 simulated games between evenly matched teams. Three and seven carry 11.7% of all outcomes between them, and the dips at two, five and nine are real. A normal distribution puts roughly equal weight on three and four, which is why it misjudges every number a predicted margin actually turns on.

Share of simulated games ending on each absolute margin of victory1%2%3%4%5%6%0123456789101112131415161718192021222324
Smoothed trend, for comparison
Absolute margin of victory, 40,000 simulated games at the model's current scoring parameters. Bold ticks mark the most common margins. Standard deviation 22.9 points.
Show values
Simulated margin frequencies
MarginShare of outcomesAgainst trend
00.00%−2.12
13.46%+0.59
22.88%−0.37
35.15%+1.38
44.75%+0.93
52.62%−1.93
63.69%−0.45
76.53%+2.81
83.11%−0.87
92.64%−1.23
103.97%+0.96
113.11%+0.05
122.22%−1.11
133.36%+0.34
144.00%+1.16
152.42%−0.61
162.23%−0.57
173.12%+0.77
182.22%−0.19
191.78%−0.71
202.66%+0.47
212.69%+0.57
221.61%−0.54
231.85%−0.17
241.93%+0.14

18,755 completed games from the 2026 season are on file — generated finals beside the credited real ones — enough to check the model's output against the season it produced. That is mostly a test of internal consistency, not of what happened on a field. One season, not every game ingested — pooling them would compare the model against a mixture.

What makes this prep-specific

A model lifted from the professional game gets high school football wrong in four ways, each of which has a named counterpart in the code.

Blowouts are routine, and most states run a mercy clock
Margin past the threshold is discounted rather than taken at face value
Rosters turn over every year as seniors graduate
Ratings regress toward classification means, and step size rises with churn
Teams rarely play outside their classification
Enrolment anchors a prior that fades as the classes actually meet
Travel ranges from cross-town to four hours
Home field scales with distance and disappears at neutral sites

Start here

What the numbers are not

Every prediction here is the model's own. Nothing here reads anybody else's lines, takes money, or pays anything out, and the ranges shown describe simulation error, not model error. A model can be confidently wrong, and that is exactly the gap your own read of your team fills.