Retractable Roof Open or Closed: Does It Change NFL Scoring?

Every autumn someone tells me the roof is closed, so expect points. It sounds like a matchup edge: no wind, a fast track, quarterbacks in shirtsleeves. I wanted to test it properly, so I looked at 7,276 completed NFL games with closing spreads and totals from 1999-2025, home venues only, and split the retractable-roof games by whether the roof was open or shut at kickoff.

The headline answer is flat. Roof open averaged 46.7 points, roof closed 46.8, and the over landed at a coin-flip rate either way. The gap that matters is between covered and uncovered football, and that is what I lean on when I build my American football predictions page for indoor games.

Ground 1Open or closed, retractable roofs score the same

The two highlighted bars are nearly identical: 46.7 points with the roof open across 128 games, 46.8 closed across 608. Domes sit at 47.0, while open-air stadiums trail at 43.3 against an all-venue figure of 44.3.

So the lid itself is not the story. Once a stadium is built to be covered, it plays like an indoor venue whatever the roof is doing. The outdoor bar is the one that drags the average down.

The split holds across eras too. Closed-roof games averaged 47.3 against 44.9 outdoors in 2013-2025, and 45.4 against 41.9 in 1999-2012. Scoring has risen everywhere, but the indoor premium has stayed.

Ground 2The market already prices covered stadiums higher

Points matter only relative to the number. The closing total averaged 44.9 with the roof open and 45.6 closed, so bookmakers set nearly the same line in both cases. Across all venues the average line was 43.5.

In every category the dot for points scored sits to the right of the line. Scoring beat the closing total by +1.8 per game with the roof open, +1.2 closed, +1.3 under a dome and only +0.5 in the open air.

That pattern tells me the market has been slightly slow on covered venues, but by margins small enough to sit inside normal game-to-game noise. The open-roof figure looks the most generous, and it rests on the thinnest sample.

Ground 3Over rates hover at a coin flip indoors

With pushes removed, the over hit 50.0% with the roof open and 50.3% closed. Domes reached 52.0%, open-air games 48.8%, and the all-venue rate was 49.5%.

The honest caveat lives here. The open-roof range is ±8.7 points against ±4.0 closed, because there are only 128 open games, and teams choose to open the roof, usually when the weather is mild, so open and closed games were never randomly assigned. Read the open bar loosely.

At flat stakes, backing the over returned −3.2% roof open, −2.5% closed, +0.5% under a dome and −5.6% outdoors. A fair price for the closed-roof over at 50.3% is 1.99 decimal, so the standard juice takes away whatever small lean exists.

Ground 4Individual buildings diverge far more than roof state

Here the averages break apart. Dallas (AT&T) averaged 51.4 points against a 47.4 line and went over 59.4% of the time across 144 games, with the roof open just 6.2% of the time. Arizona (State Farm) went over 55.5% from 165 games.

At the other end, Indianapolis (Lucas Oil) went over 44.7% and Atlanta (Mercedes-Benz) only 39.7%, though Atlanta has 73 games, below the 100-game mark. Houston (NRG), with the largest sample at 203 games, sat at 47.5%.

These stadium rows mix venue with the teams that played there, the defences, the offensive lines and the coaching eras. I treat them as prompts to check the current matchups, not as a stadium effect that carries forward unchanged.

Written byPau Oliveras

I'm Pau Oliveras, based in Barcelona, and I write the live NFL scores and weekly picks at horse-racing.tips — pulling apart spreads, totals and the matchups that quietly decide them.

View profile →