Model versus market
The model does not beat the market, and this page is not a card to bet. It is published because the comparison is interesting and because a system that only shows you its good weeks is not telling you anything.
Two habits, tracked
Two ways these games have been bet for years: taking the underdog when the model has them winning outright, and betting against teams the market keeps getting wrong. Both are tracked here from the first game of the season. Tracked, not recommended.
A person decides, not the list. Every pick below comes from a fixed rule applied to the model and the market, and nothing more. It is a starting point for human judgement, never a bet on its own: before any money goes down, someone reads the game — injuries, lineups, travel, motivation, whatever the numbers cannot see — and makes the call. The record logs every pick the rule makes, whether or not anyone bet it, so it measures the rule by itself rather than those decisions.
What the backtest says
Break-even at −110 is 52.38%. Graded on the model as it stood the morning of each game, and each team’s cover record from earlier games only.
One corner looks alive: underdog picks getting 10 or more points went 21–6. Almost all are November games where the model barely knows one team, the same failure the disagreement table below describes, and it did not hold on DraftKings. Worth watching, not betting.
This season’s record
| Habit | Graded | Record | ATS | Units | CLV |
|---|---|---|---|---|---|
| Underdog picks | 33 | 20–13 | 60.6% | +5.2 | +0.3 |
| Fade list | 0 | 0–0 | – | +0.0 | – |
Logged the morning of each game at the DraftKings line then available, graded against the result and the closing line, and never edited.
Today’s underdog picks
| Game | Market | Model | Pick |
|---|---|---|---|
| Lindenwood at Omaha | Omaha −4.5 | Lindenwood −0.2 | Lindenwood +4.5 |
| Marist at Lehigh | Lehigh −1.5 | Marist −0.2 | Marist +1.5 |
Today’s fades
None today.
The fade list
Teams covering 35% or less of their games after 10 or more against the spread. The pick is their opponent.
Nobody yet: it takes ten games against the spread to qualify.
The market is the better predictor
| MRI 2.0 | Market | |
|---|---|---|
| Mean margin error, 16,955 games (ESPN Bet) | 9.40 pts | 8.86 pts |
| Mean margin error, 4,498 games (DraftKings) | 9.52 pts | 8.81 pts |
| Picks the winner outright | 71.6% | 73.0% |
The two numbers correlate at 0.90, so most of the time they agree. Where they disagree it is usually the model that is wrong, and half a point of margin error is the whole reason there is nothing to bet here.
Against the closing line, by season
| Season | ATS |
|---|---|
| 2023 | 51.6% |
| 2024 | 50.0% |
| 2025 | 50.0% |
| 2026 | 51.9% |
Break-even at −110 is 52.38%. Four seasons, never close, no trend. On DraftKings over 2025–26 it is 48.3% across 4,498 games, which is worse still.
By size of disagreement
If the model knew something the market did not, the buckets where it disagrees most should be the ones that win. Against ESPN Bet's closing number:
| Gap | Bets | ATS | Units | CLV |
|---|---|---|---|---|
| 0-1 | 3569 | 51.3% | -82.9 | – |
| 1-2 | 3309 | 50.1% | -156.0 | – |
| 2-3 | 2755 | 50.1% | -132.5 | – |
| 3-5 | 3779 | 49.9% | -194.2 | – |
| 5-7 | 1993 | 51.4% | -41.9 | – |
| 7-10 | 1109 | 51.6% | -18.7 | – |
| 10-15 | 361 | 51.2% | -8.6 | – |
| 15+ | 79 | 68.4% | +26.5 | – |
One bucket clears significance: gaps of 15 points or more went 54–25. There are 79 of those in 16,954 games, 42 of them in the first season tested, and reading them is the end of the story — they are November mismatches against opponents the model has barely seen. In one, the model favoured Mississippi Valley State by four where the market had Hawai'i by 25.5. It won the bet while being wrong by 22 points. The bucket is measuring the model's own failures, and it does not reappear on DraftKings.
The one faint signal
Betting the opening number rather than the close, mean closing line value is positive and grows with the size of the disagreement — the market tends to drift toward this model's side after the opener. That is a reason to keep measuring, not a reason to bet, and it is the same shape of hint football produced.
| Gap | Bets | ATS | Units | CLV |
|---|---|---|---|---|
| 0-1 | 884 | 51.4% | -19.0 | +0.08 |
| 1-2 | 846 | 49.9% | -44.4 | -0.02 |
| 2-3 | 763 | 48.2% | -66.5 | +0.11 |
| 3-5 | 1010 | 48.2% | -88.3 | +0.21 |
| 5-7 | 526 | 45.6% | -74.6 | +0.22 |
| 7-10 | 302 | 50.0% | -15.1 | +0.58 |
| 10-15 | 76 | 53.9% | +2.5 | +0.68 |
| 15+ | 10 | 70.0% | +3.7 | +8.20 |
DraftKings openers, 2025–26.
Where they disagree now
| Date | Game | Model | Market | Gap | Model's side |
|---|---|---|---|---|---|
| 11-13 | Tennessee State at Tennessee | +37.9 | +28.5 | +8.4 | Tennessee |
| 11-13 | DePaul at Northwestern | +5.7 | +1.5 | +7.2 | Northwestern |
| 11-13 | Yale at Northeastern | -8.6 | -1.5 | -7.1 | Yale |
| 11-13 | Georgetown at Clemson | +10.5 | +4.5 | +7.0 | Clemson |
| 11-13 | Columbia at Bucknell | -11.4 | -5.5 | -6.9 | Columbia |
| 11-13 | Fairleigh Dickinson at Oklahoma | +26.7 | +31.5 | -6.8 | Fairleigh Dickinson |
| 11-13 | Prairie View A&M at Vanderbilt | +44.2 | +40.5 | +6.7 | Vanderbilt |
| 11-13 | UIC at Saint Mary's | +11.0 | +6.5 | +6.5 | Saint Mary's |
| 11-13 | Stanford at UNLV | -4.5 | -1.5 | -6.0 | Stanford |
| 11-13 | Alcorn State at Tarleton State | +19.4 | +15.5 | +5.9 | Tarleton State |
| 11-13 | Arkansas-Pine Bluff at California Baptist | +11.0 | +15.5 | -5.5 | Arkansas-Pine Bluff |
| 11-13 | Ball State at SMU | +26.8 | +23.5 | +4.3 | SMU |
| 11-13 | SIU Edwardsville at Duke | +39.4 | +44.5 | -4.1 | SIU Edwardsville |
| 11-13 | Milwaukee at TCU | +21.5 | +22.5 | -4.0 | Milwaukee |
| 11-13 | UL Monroe at Arizona State | +34.5 | +31.5 | +4.0 | Arizona State |
| 11-13 | App State at Charlotte | +4.6 | +7.5 | -3.9 | App State |
| 11-13 | UC San Diego at Utah State | +17.4 | +12.5 | +3.9 | Utah State |
| 11-13 | Northern Iowa at Michigan State | +25.7 | +27.5 | -3.8 | Northern Iowa |
| 11-13 | Illinois State at North Dakota State | -7.2 | -3.5 | -3.7 | Illinois State |
| 11-13 | Lindenwood at Omaha | -0.2 | +4.5 | -3.7 | Lindenwood |
| 11-13 | Texas Southern at Murray State | +23.9 | +29.5 | -3.6 | Texas Southern |
| 11-13 | Georgia at North Carolina | +8.0 | +4.5 | +3.5 | North Carolina |
| 11-13 | South Dakota at Drake | +17.0 | +13.5 | +3.5 | Drake |
| 11-13 | Wright State at Samford | -4.6 | -1.5 | -3.1 | Wright State |
| 11-13 | Ohio at Purdue | +23.5 | +18.5 | +3.0 | Purdue |
55 of the next 175 games carry a DraftKings number. Games involving a team the model has barely seen are left out entirely, because the largest gaps in November are the model's ignorance rather than its opinion.