The Patriots’ Kicker Scoreboard Trick Debunked | PerThirtySix

January 22, 2012, the AFC Championship Game at Gillette. The Ravens get the ball back, trailing 23-20 with only 1:44 left. Joe Flacco moves them 65 yards in seven plays. With 15 seconds left, Billy Cundiff comes on for a 32-yard chip shot to send it to overtime. It sails wide left, and off the Patriots go to another Super Bowl.

Nerves? Or was it something more?

On Wednesday, Ryan Paganetti, who worked in analytics for the Eagles, Jaguars and Raiders, posted that he had gone through more than a decade of Patriots home broadcasts and found that when an opposing kicker lined up at Gillette Stadium, the video board behind the uprights showed the broadcast's reverse-angle shot of the kick. From the kicker's spot, that put a second set of yellow posts behind the real ones, with the wind streamers blowing the other way. He found dozens of examples on opponents' kicks and none on 52 Patriots attempts from the same games. Cundiff's kick was just one of the ill-fated attempts.

A reply from @barryonhere attached a table: “This is a really awesome find. It also helps explain this absolutely insane stat. From 2001-19, the Patriots made 9.4% more of their field goals than their opponents. No other team in the NFL in that span was even at +4.0%.”

The table posted is accurate (except for the Raiders' mis-signed difference). We reproduced each row from play-by-play from nflverse, and the Patriots really are all alone at the top, with a +9.4% field goal differential. The second-place Colts are a massive gap away, at +3.8%. Is the scoreboard trick really that effective against professional kickers?

What about Patriots away games?

The above +9.4% advantage covers every game the Patriots played, including Patriots away games. 258 of their opponents' 518 attempts were in the opponents' own stadiums. Obviously, the Foxboro board trick will only affect opposing kicks in Foxboro. So, first, let's look at how opposing kickers do in their own stadiums against the Patriots.

Opposing kickers made 75.4% at Foxboro, but only 78.3% in their own stadiums when they played the Patriots, compared to the league average of 82.4%. Which means kickers were worse against New England regardless of whether they were in New England or not. Credit to an exceptional special teams unit, and maybe some increased pressure of high-leverage games against the NFL's greatest dynasty?

Did kickers do worse at Foxboro?

To test kicker performance properly, we built an expected-make model. We trained a generalized additive model (R's mgcv package) on the 25,288 non-blocked field goal attempts from 2001 to 2025 (25,825 before removing blocks). It predicts the chance a kick goes in from its distance, who is kicking (each of 184 kickers gets his own baseline, shrunk toward the league average when the sample is small), the season, home or away, playoff or regular season, and, for outdoor games, the temperature and wind at kickoff. Notably, we exclude a feature for the stadium. Since the stadium is what we're testing, we leave it out of the model and see if it shows up in the residual.

Every kick gets an expected make probability. For any group of kicks, we add up the probabilities to get expected makes and compare that with the actual makes. The difference, as a share of attempts, is the group's shortfall. Adam Vinatieri as a Colt is a small example. He attempted 12 field goals at Foxboro. Given his baseline and the distances, the model expected 9.6 makes; he made 7, a shortfall of 22% on 12 kicks.

Opponents took 498 non-blocked kicks against the Patriots in the 2001–19 regular seasons, the window of the viral table: 250 at Foxboro and 248 in their own stadiums. At Foxboro they fell 4.1% short of expectation. In their own stadiums they fell 2.7% short. (Those are the 75.4% and 78.3% from the first chart after adjusting for distance, kicker, and the home-field edge. The raw gap between them was 2.9%; the adjustment brings it to 1.4%.)

Big as the screens may be, a board in Foxboro won't cause a miss in Buffalo, so the 2.7% that opponents missed by on the road against the Patriots can be a result of the Patriots' special teams rush, the game situations, luck, or, likely, some combination of all of the above. And there's no reason those effects would be absent at Foxboro. Assuming that 2.7% was in effect at Foxboro too, we're left with only the excess, 4.1% minus 2.7% = 1.4%, as potentially creditable to board shenanigans.

If we look at this in terms of raw kicks instead of percentages, 1.4% of 250 attempts is about three and a half extra misses across 19 seasons of visits to Foxboro to play against the Patriots. The standard error on that 1.4% is 3.1%, or about eight kicks, so there's nothing to tell us this is anything other than noise.

An unbiased simulation of "worst stadiums": measuring against p-hacking

Foxboro is the worst of the 32 stadiums for visiting kickers from 2002 to 2022, -3.9% below expectation. That -3.9% is the -4.1% from the model section measured on a different window: 2002 through 2022 with playoffs, instead of the 2001–19 regular seasons in the viral table. It is the whole Foxboro shortfall, road rate included; the 1.4% that is specific to Foxboro is the part of it above the road rate. Some stadium has to be worst, so the comparison we do here is against the worst stadium in a league where there are no venue effects. That is, what kind of spread does random noise alone cause, given the volume of kicks at each stadium? We replayed every 2002–22 visiting kick as a flip at its modeled probability, aggregated each stadium's differential, and measured the worst recorded differential amongst the 32 teams. And then repeated that 4,000 times. The median worst stadium in those leagues is -3.9% below expectation, the same as Foxboro. ~50% of them produce a stadium worse than Foxboro.

Because the model has no stadium effects, real venue effects that aren't accounted for in the model show up here. We model for kicker, distance, weather, and more, but not for altitude. So Denver, where the ball carries farther at Mile High Stadium, is +4.8% above expectation, well outside the band. Foxboro is well inside it.

The HD boards arrived in 2010

The last thing we can look at for any cause-and-effect is New England's stadium changes. Gillette opened in 2002 with two 27-by-48-foot screens. For the 2010 season, these were replaced with two HD boards, 41 by 164 feet in the south end zone and 45 by 100 in the north, more than five times the screen area. Paganetti's examples run from the 2011 season through 2020. If the reverse-screen trick affected kickers, opponents' shortfall at Foxboro should get worse relative to their road shortfall starting in 2010.

At Foxboro, opponents' shortfall was -4.4% on the original screens and -3.7% on the bigger screens they introduced in 2010. In Patriots road games it went from -3.5% to -3.8%. The difference in differences, Foxboro minus road in the big-screen era against the same thing in the small-screen era, is +1.0% with a standard error of 5.8%. Which is the opposite of what the theory predicts (with a comparatively massive error bar). The 2023–25 rows are 58 and 47 kicks; over that span the Patriots' own kickers were -6.7% under expectation at Foxboro, worse than the visitors.

The verdict

People have been trying things for as long as there have been uprights to kick through or hoops to shoot through. Dick Ericson, the Metrodome's superintendent from 1982 to 1995, admitted in 2003 that when the Twins were behind late in close games he turned on the fans behind home plate to push the air out toward the fences. Arizona State's students built a "Curtain of Distraction" behind the basket for opponents' free throws; in 2015 the economist Justin Wolfers estimated it was worth a point or two a game, and by 2024, with visiting teams knowing what was coming, analysts couldn't find the effect anymore.

The free throw is probably the closest thing sports has to a controlled test of crowds against professionals: the shooter and the basket are fixed and alone. In Scorecasting, Tobias Moskowitz and L. Jon Wertheim went through more than 23,000 NBA games over two decades. Home teams made 75.9% of their free throws. Visiting teams made… 75.9%.

Football's version, icing the kicker, has been studied repeatedly, and the studies disagree with one another: some find nothing, some find a few percent, and they disagree on whether icing helps or hurts.

Professional kickers rehearse the same swing thousands of times a year and take it in front of 65,000 people who want them to miss. A second set of goalposts on a screen is simply one more thing to ignore, and they probably did.