Omaha 0.7.0
Released 3 September 2026
Play calling now responds to the scoreboard. A team two scores behind throws far more often than one two scores ahead, and until this release the simulation called the same game either way.
How to read these numbers
Every figure below compares the simulation against a player’s own season average so far, scaled to how much of the game is left, roughly what a reader could work out without us. Zero means we add nothing over that. Positive means we beat it; negative means we are worse than it.
Measured across the 2024 season, weeks 1–18, rebuilding the model each week using only what was known at the time. 400 simulations per situation.
The change column is simply this version’s accuracy minus 0.6.0’s, each measured across everything that version covers, so the two columns you can see always account for it. We also measure the change on shared situations alone, which is a more sensitive test; where the two disagree we publish this one.
Accuracy by stat
| Stat | vs. season average | Change from 0.6.0 |
|---|---|---|
| Receiving touchdownsrare event | +15.0% | +0.01−0.36 to +0.36 |
| Rushing touchdownsrare event | +12.8% | −0.29−1.60 to −0.39 |
| Passing yards | +9.8% | +2.43+1.01 to +3.80 |
| Pass attempts | +7.7% | +4.70+2.76 to +6.56 |
| Passing touchdownsrare event | +7.6% | +0.02−0.72 to +1.82 |
| Completions | +6.5% | +3.25+1.59 to +4.90 |
| Receiving yards | −4.0% | +0.18+0.13 to +0.44 |
| Rushing yards | −10.8% | +0.26−0.08 to +0.76 |
| Receptions | −21.0% | +0.40+0.34 to +0.78 |
| Targets | −26.5% | +0.65+0.44 to +0.96 |
| Rush attempts | −38.1% | +0.73+0.30 to +1.29 |
| Average across all figures | −3.7% |
The average is an unweighted mean across the figures above. It is a summary of how we are doing, not a single score for the model: yards and counts are not measured in the same units, so combining them any more cleverly than this would just let passing yards decide the answer.
Each change carries the range the measurement can actually support. A change whose range crosses zero is shown in neutral rather than as a gain: we cannot tell it apart from no change, and we are not claiming it as one.
What changed
Real teams pass on about 78% of snaps when 15 or more points behind and about 30% when 15 or more ahead. The simulation used to call about 58% passes regardless of the score, so a trailing team ran nearly twice as often as it should. That fell hardest on the situations that produce the most carries and the most attempts.
Pass attempt accuracy improved 4.7 points (95% interval 2.8 to 6.6), completions 3.3 (1.6 to 4.9), passing yards 2.4 (1.0 to 3.8). Across the four margin bands the simulation's pass rate now runs 0.680 / 0.600 / 0.596 / 0.475 against an actual 0.673 / 0.610 / 0.612 / 0.515.
Calling the right number of runs feeds through to who carries the ball and how often receivers are thrown to. The gains are smaller here because a full game averages across leading and trailing states, so the old errors partly canceled.
Rush attempts improved 0.8 points (0.3 to 1.3), targets 0.7 (0.4 to 1.0), receptions 0.6 (0.3 to 0.8), receiving yards 0.3 (0.1 to 0.4).
Deliberately barely affected. Third and short is a running down whatever the scoreboard says, measured, it moves about 7 points across the full range of game states against 25 points on first down. A model that made it swing as hard as first down would be fitting noise.
What this version still gets wrong
- Rushing touchdowns got slightly worse, by 1.0 points. They were already under-predicted, we expect about 0.22 per opportunity where the real rate is 0.26, and this release nudged that existing bias further. The likely cause is the goal line, where the swing in play calling is steepest: a trailing team close to scoring now passes more than it used to. Measured against 0.5.0, rushing touchdowns are still marginally ahead overall.
- Rush attempts remain the least accurate figure we publish, at −38.2%. Three releases have improved it by 3.6 points in total; it is still the weakest thing we do.
- Team quality barely affects running plays. A better rushing offense does not reliably produce better runs in the simulation. The effect is about a third as strong as it is for passing, and it is not even consistently ordered. This is the next thing being worked on.
- The simulation runs about 68 plays per team per game against a real 63. The extra plays are spread across everyone and inflate nothing in particular, but the figure should be right.