Omaha 1.9.0
Released 14 September 2026
Near the goal line, real teams hand the ball to different people. The simulation was choosing a carrier as though it were standing at midfield.
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 1.8.0’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 1.8.0 |
|---|---|---|
| Receiving touchdownsrare event | +20.1% | +0.00+0.00 to +0.00 |
| Rushing touchdownsrare event | +18.0% | +0.73+0.43 to +1.20 |
| Receiving yards | +13.7% | +0.00+0.00 to +0.00 |
| Pass attempts | +13.0% | +0.00+0.00 to +0.00 |
| Rushing yards | +12.5% | −0.01−0.04 to −0.00 |
| Passing yards | +11.0% | +0.00+0.00 to +0.00 |
| Completions | +9.7% | +0.00+0.00 to +0.00 |
| Passing touchdownsrare event | +6.9% | +0.00+0.00 to +0.00 |
| Rush attempts | +3.8% | −0.03−0.06 to +0.00 |
| Targets | +1.5% | +0.00+0.00 to +0.00 |
| Receptions | +1.3% | +0.00+0.00 to +0.00 |
| Average across all figures | +10.1% |
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
First and goal from the one is not first and ten. Real teams sneak it with the quarterback about one time in six, and almost never hand it to a wide receiver. The simulation decided who carried the ball from a player's usage on that down and distance, which carries no information about where on the field the play is starting. So it sneaked far too rarely, gave receivers several times the carries they really take, and leaned on the running back to make up the difference.
Quarterbacks take 16.7% of carries inside the five. The simulation gave them 9.9%. Receivers take 1.0%; it gave them 3.5%. In total 7.1% of goal-line carries went to the wrong kind of player, and about one goal-line carry in every two and a half ends in a touchdown.
This is where rushing touchdowns are decided, so getting the carrier wrong there is mostly a rushing touchdown error. Correcting it moved that stat and left every other published figure untouched, because the change alters which player receives a carry and nothing else about the play.
Rushing touchdowns improved from 17.3% better than a season average to 18.0%, and on the measure built for rare events from 9.0% to 10.2%. Ten of the twelve stats we score did not move at all.
What this version still gets wrong
- This moves carries between positions, not between running backs. Real teams also lean harder on their main back near the goal line, and the simulation still spreads those carries the way it spreads carries anywhere else.
- It applies to running plays only. Whether the right receivers are targeted near the goal line is a separate question we have not measured.
- The adjustment is one league-wide pattern. A team whose quarterback sneaks often and a team whose quarterback never does get the same treatment.
- Rushing yards came down very slightly, by about two hundredths of a percentage point. Moving goal-line carries to quarterbacks is the reason: they gain less when they get there.