Game Simulator

How the NFL Game Simulator Works – Ratings and Drives

Where the numbers come from

Every team in the league carries two ratings on a 0–100 scale, and the whole model is built on the gap between them.

  • Attack is how consistently an offense moves the ball and finishes drives. A 90 attack means drives end in points often; a 70 means a lot of punts.
  • Stop is how often a defense ends a drive without points allowed. It is not a points-allowed average, it is a per-drive shutdown rate.

Both numbers are the site's own editorial ratings. They are kept in one table that covers all 32 franchises, they are reviewed through the season, and they are deliberately rounded rather than quoted to a decimal. They are not official NFL statistics, they are not a betting line, and nothing on this page is a forecast of a real fixture.

Why every team gets eleven drives

A game is modelled as 22 possessions, eleven a side, alternating, with the road team starting. The possessions are grouped by quarter: the first three for each side land in the first quarter, the next three in the second, and so on. That grouping is the only reason the line score in the scoreboard adds up to the final — there is no separate quarter simulation underneath.

How one drive is decided

Each possession is rolled on its own, against the defense it is facing:

  1. The offense's attack rating and the defense's stop rating are compared.
  2. The gap between them sets the odds of a touchdown, a field goal and a empty drive.
  3. One roll picks the outcome, and the points go on the board.

Because the roll is independent every time, the same pairing does not produce the same game twice. A 31–27 one minute and a 23–35 the next is the model working as intended, not a bug.

Six points or seven

A touchdown is always worth six. After it, the model rolls a conversion: 92% of the time the extra point is good and the drive is worth seven, the rest of the time it misses and the drive stays at six. There is no two-point conversion in the model, which is why you will see 6 and 7 on the same scoreboard and never an 8.

Field goals are three points. Safeties are not modelled.

Home field

The home side carries a small edge, worth roughly two and a half rating points of extra attack. It is enough to tilt a coin-flip matchup without deciding it, which is why swapping home and away sometimes changes the winner and sometimes only changes the margin.

Overtime

If the fourth quarter ends level, both sides get extra possessions until one scores and the other does not. The model gives overtime six rounds. In the rare case that six rounds pass with neither side scoring, the result stands as a tie rather than inventing a winner, so a tie is possible but you will need a lot of runs to see one.

How to read a result

What you see What it means
A wide final score A big attack-versus-stop gap between the two teams
A one-score game Two ratings that sit close together
A home win by a field goal The home edge doing just enough
Six points on a drive The extra point missed

A single run is a sample of one. Three runs of the same pairing tell you more about the matchup than any single scoreline, and running it both ways tells you whether the venue mattered.

What the model does not do

It does not use injuries, weather, travel, rest days or in-season form. It does not read a betting market. It does not simulate a season, a playoff bracket or a Super Bowl — those modes are not built, so no page here offers them.