Omaha 2.2.0
Released 30 September 2026
Questions with a whole-number answer are answered in whole numbers again. The chance of a receiver catching two or more passes was reading 23.5% where it should have read 31.3%, against an actual 32.7%.
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 2.1.0’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 2.1.0 |
|---|---|---|
| Receiving touchdownsrare event | +20.1% | −0.19+0.00 to +0.00 |
| Rushing touchdownsrare event | +16.7% | −0.13+0.00 to +0.00 |
| Receiving yards | +14.1% | +0.00+0.00 to +0.00 |
| Pass attempts | +13.0% | +0.12+0.07 to +0.18 |
| Passing yards | +11.1% | +0.00+0.00 to +0.00 |
| Rushing yards | +10.9% | +0.00+0.00 to +0.00 |
| Completions | +9.8% | +0.25+0.15 to +0.36 |
| Passing touchdownsrare event | +6.9% | −0.05+0.00 to +0.00 |
| Targets | +2.7% | +0.78+0.62 to +0.90 |
| Receptions | +2.6% | +0.92+0.73 to +1.07 |
| Rush attempts | +0.7% | −0.07−0.08 to +0.14 |
| Average across all figures | +9.9% |
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
This is the figure most worth checking against a price, and it was too low across the board. Every projection is nudged toward a player's own season average before kickoff, because the simulation is weaker there than a season average is. That nudge scales the whole distribution, so a receiver who caught exactly five passes in a simulation came out holding 4.6 of them. Asked whether he cleared five, the answer was no. He had caught five in every simulation.
Measured against what actually happened, on about 93,000 checks: the chance of two or more catches read 23.5% where the real rate is 32.7%, and now reads 31.3%. Three or more read 13.0% against a real 19.2%, and now reads 17.5%. Three or more targets now reads 29.6% against a real 29.5%.
The same scaling left the projection ranges on fractions of a catch. A range whose bottom is 1.4 catches excludes a player who caught one, which is a real outcome, so the ranges were both wrong and quietly harder to clear than they looked. Receiving and rushing yardage are genuinely continuous and were never affected: 61.4 yards is a real answer, 1.6 catches is not.
Receptions improve from +1.7% to +2.6% and targets from +1.9% to +2.7%, the two weakest figures on this page. Completions +9.5% to +9.8%, attempts +12.8% to +13.0%. All four hold up on a game-by-game resampling. The average moves from +9.7% to +9.9%.
Receiving, rushing and passing yards are identical to the last digit, because nothing about the simulation changed. This release does not teach the engine anything new about football: the simulated games are the same games, read correctly. We would rather say that than find a number to lead with.
Receiving yards +14.1%, rushing yards +10.9%, passing yards +11.1%, all unchanged to four decimal places.
What this version still gets wrong
- Four figures on this page read slightly worse: receiving touchdowns +20.1% from +20.3%, rushing touchdowns +16.7% from +16.8%, passing touchdowns +6.9% from +7.0%, and rush attempts +0.7% from +0.8%. Touchdowns are scored two ways because they are rare, and on the second measure they moved by a hundredth of a point in opposite directions. We do not believe any of the four is meaningful, and we would rather show you the numbers than decide for you.
- Receptions are still the second-weakest figure here. The remaining gap is not the shape of the projection but its level: the engine projects about 8% fewer catches than actually happen, and scaling the projection up makes matters worse, which means the error is in which receivers get the ball rather than in a global total. Nothing in this release addresses that.
- An 80% range on a whole-number figure is wider than 80%, and that cannot be fixed. With outcomes of zero, one, two or three catches there is no way to draw a range covering exactly four fifths of them, so these ranges are a little conservative by construction. We checked this against the correct distribution rather than assuming it: on receptions the honest best is 89%, and ours is 90%.
- Rush attempts are the exception to the point above. There the counts are large enough that a true 80% range is reachable, and ours now runs wider than it should at 85%.