HeatRadar Research

What wind and rain actually do to an NFL offense

Everyone knows bad weather hurts the passing game. But by exactly how much? We measured 4,021 outdoor team-games against the same teams' own calm-weather performance to put a number on the impact, and the honest answer is that wind matters more than rain, both matter less than people assume below 12mph, and above 16mph the passing game falls off a cliff.

How this was measured

We didn't want to just compare windy games to calm games in a vacuum; that mostly measures something else. The teams who play in 18mph wind in December are Buffalo, Chicago, Cleveland and Green Bay; the teams who never do play in domes. Those are different offenses with different quarterbacks, and pooling them tells you about the schedule rather than the weather.

So every team is compared to itself. A team's game in the wind is measured against that same team's average in its own outdoor, dry, under-8mph games the same season.

Dome games are excluded from both sides, and it is worth saying why with numbers rather than asserting it. A dome is not neutral conditions, it is better than neutral. Taking the same team in the same season, indoors against its own outdoor calm games, it gains 2.3% on yards per attempt, 14.8% on touchdown rate and 8.5% on sack rate across 205 team-seasons. Something about indoor football helps the passing game beyond the absence of weather.

Had those games gone into the baseline, every figure on this page would be too large, because wind would have been charged for the indoor bonus on top of its own damage.

One detail that matters more than it sounds: a calm game is left out of the average it is being measured against. Without that, the calm bucket is compared to itself and reads as exactly zero no matter what, which tells a reader nothing about whether the method is centered. With it, the calm row below is a real control, and it lands within half a percent of zero on every measure. That is the number to look at first.

How wind affects passing and rushing

Conditions come from the league's own play-by-play feed: wind speed in miles per hour, and a free-text weather note from which precipitation is read. Games below are dry, so this is wind on its own.

Each figure is the change against the same team's calm dry games that season. The range in brackets is how much the answer moves if you reshuffle which games are in the sample: a narrow range means the number is solid, a wide one means it is not.

The 20mph row deserves a warning

It looks milder than the 16–19 band on passing, which is almost certainly not real. It rests on 84 team-games, the yards-per-attempt figure could sit anywhere from 11.1% worse to barely worse at all depending on which games you count, and games that windy are disproportionately played by teams built for them. Treat 16mph and above as one regime rather than reading the top row as a recovery.

Does wind affect quarterbacks?

Yes, and the damage is to accuracy rather than opportunity. A quarterback throws almost as often in a gale as he does in still air, down 0.8% in the 16–19mph band, so the volume he needs is still there. What he loses is what he does with it. Completion rate falls 4.2%, yards per attempt 9.2% and touchdown rate 10.6%, and he is sacked 10.6% more often.

Below 12mph none of it is worth acting on. At 8 to 11mph the completion figure lands anywhere between slightly worse and slightly better depending on which games you count, which is the statistical way of saying nothing is happening. The honest threshold for a quarterback is 12mph, and the honest panic point is 16.

How rain and snow affect an offense

Rain is held to games under 12mph so the wind effect above is not counted twice.

Rain costs a passer more accuracy than wind does. Completion rate is down 5.8%, yards per attempt 10.0% and passing touchdown rate 17.5%, and all three are big enough that they are not chance. What rain does not do is suppress volume: teams throw slightly more in the rain, not less. They simply throw it worse.

How hard it rains matters far more than whether it rains

The league's own weather note is not a rainfall record. It says "rain" or it does not, it uses the word "heavy" once in fourteen seasons, and checked against a reanalysis record it misses most of them: 423 outdoor games had measurable rain and the note mentions 159 of them. It is also full of forecasts rather than observations, including eleven games whose note says there is no chance of rain at all.

So we stopped using it for rain and took the number from somewhere that has one. Rainfall comes from the Open-Meteo historical archive, hourly, matched to each stadium's coordinates and summed over the three hours from kickoff.

What that archive is, stated plainly, because it changes how you should read the rest. It is weather reanalysis, chiefly ERA5 and ERA5-Land, not a gauge reading taken at the stadium. Reanalysis assimilates station, radar and satellite observations into a physical model and outputs a gridded estimate, here at roughly 9 to 25km. So "the rain at the stadium" is really the rain over a grid cell containing the stadium. That is a genuine limitation: it will smooth a narrow, intense cell sitting over one end of a city, and it is the reason we band the results rather than quoting rainfall to the hundredth of an inch on any single game.

It is still a large improvement on the alternative, which is a free-text note with no number in it at all. And where the two sources overlap they agree: on a spot check of twenty games, every one the league called rain shows measurable precipitation and every one it called clear shows exactly zero.

Rainfall totalled over the three hours from kickoff, at the stadium, from the Open-Meteo reanalysis archive. Wind held under 12mph so the two effects are not counted twice, and snow games excluded.

The genuinely wet games, with a warning attached

A quarter of an inch or more has happened 27 times in fourteen seasons, and after holding wind down and setting snow aside it leaves 25 team-games. That is a tiny number. We are showing it anyway, in the bottom row of the table above, because these are the games people actually want to know about and saying nothing is its own kind of answer.

Read it as a glimpse, not a coefficient. Do not apply it to a projection.

What those 25 games look like: completion rate down 4.7%, yards per attempt down 5.8%, and the running game hit harder than anywhere else on this page at 8.8% fewer yards a carry. Teams threw more, not less, up 6.1%.

The honest reading is that this row does not show a worse passing day than the 0.10 to 0.25 band below it, and it probably should. Twenty-five games is few enough that one strange afternoon moves everything, and the range around that completion figure runs from 10.5% worse to 1.2% better, which is another way of saying we cannot tell. The wettest game on record is Dallas at Pittsburgh in 2024 at 0.88 inches.

If you take one thing from this row, take the running game. Losing almost a yard a carry in the heaviest rain is the only number here that gets worse as the rain gets harder, which is what you would expect and what the passing numbers are too thin to confirm.

Snow is shown without the wind cap, because snow games are windy games and holding wind constant leaves too few to say anything. The figure therefore has wind inside it, which is part of why it is the most dramatic row on the page: passing volume down 8.8%, rushing volume up 12.8%, and yards per carry up 10.4%. On 55 team-games, read it as a direction rather than a coefficient.

What this means for a projection

Passing yards are attempts multiplied by yards per attempt, so the two effects compound instead of being chosen between. Receiving yards ride on the same product (a receiver cannot catch what the offense does not throw), so wideouts and tight ends take the passing adjustment. Rushing yards are carries times yards per carry, and in wind the first rises while the second barely moves.

What this study does not settle

• Wind direction and stadium shape are ignored. A 15mph crosswind in an open bowl and 15mph swirling in a horseshoe are the same number here and are plainly not the same condition.
• Forecast versus actual. These are conditions recorded at the game. Anyone applying this on a Thursday is using a forecast, which is a different and noisier input.
• Teams are not interchangeable. Every figure is an average across offenses built very differently. A team that throws short and quick should lose less to wind than one that pushes the ball down the field, and that split is the obvious next piece of work.
• Cold is not separated out. Wind and low temperature arrive together in December, and this study does not disentangle them.
• Dome teams are thinner in the sample, though they behave the same. A team needs four of its own outdoor calm games to earn a baseline, and a dome team only plays those on the road, so they qualify in a median of four seasons against twelve for outdoor teams. Where they do appear they lose the same amount: in 12mph-plus wind, dome teams drop 4.1% on yards per attempt and outdoor teams 4.9%, which is close enough that there is no real gap to see. That rests on 23 team-seasons against 258, so read it as no large difference rather than as proof of none.