Does icing the
kicker work?

Every color commentator has the same move: defense trailing by two in the final minute, the kicking team lines up for a field goal, and just before the snap the defensive coordinator calls timeout — “icing the kicker,” the theory that making him stand there and think about it for an extra ninety seconds will rattle him into a miss. It is one of the most confidently stated claims in sports, and one of the most tested: this run rebuilds it from 26,633 NFL field goal attempts across 26 seasons (1999–2024), flagging a kick as iced when the play immediately before the snap, in nflverse's own chronological play-by-play, is a timeout called by the team on defense.

The naive number looks like a real effect. Across all 26,633 attempts, iced kicks go in 77.4% of the time against 83.0% for everyone else — a 5.5-point gap, 95% CI [-7.65, -3.45], p=2.2e-07 — excludes zero, and by a wide margin. This is the number a broadcast graphic would show, and on its own it would look like proof the folk claim holds.

It's also the wrong test. “Icing the kicker” is a claim about late-game pressure, not about a defensive timeout called for any reason at any point in a blowout. Restricted to the 2,175 attempts actually in the window commentators mean — the final three minutes of the fourth quarter or overtime — 24.2% of clutch field goals (526 of 2,175) were iced by this definition. The make-rate gap on that subset falls to 3.2 points (iced 76.8% vs. not-iced 80.0%), 95% CI [-7.28, +0.92], p=0.13 — already contains zero. A season-clustered standard error on the identical spec agrees almost to the decimal (CI [-7.29, +0.92]), and a distribution-free two-proportion z-test cross-checks it independently of the OLS assumption (z=-1.57, p=0.12).

The confound that's left standing: distance. A coach doesn't call timeout before a 22-yard chip shot, and doesn't bother before a desperation 58-yarder either — icing is reserved for kicks genuinely in doubt. In the clutch window, iced attempts average 40.5 yards against 38.8 yards for the rest, a 1.7-yard gap by itself enough to explain a chunk of the naive miss-rate difference, since make probability falls steeply with distance regardless of any timeout. Holding kick distance fixed, the icing effect drops to 0.94 points, 95% CI [-4.78, +2.90], p=0.63 — contains zero, and sits closer to zero than to the naive estimate. Adding a quadratic distance term, since make probability isn't linear at the extremes, barely moves it (CI [-4.86, +2.77], p=0.59).

Read plainly: the desk cannot confirm icing the kicker works. The headline-looking 5.5-point gap is an artifact of comparing all defensive timeouts, in every game situation, to a pool of easier, less contested kicks. Once the sample is narrowed to the situations the claim is actually about, and distance is held fixed, every interval this run fit contains zero. That does not prove icing has zero effect on any individual kicker on any individual night — it means 26,633 attempts across a quarter-century are not enough to distinguish a real effect from none.