Special Teams Analytics: The Hidden Third of the Game

The Third of the Game Analytics Forgot

Football is conventionally split into three phases - offense, defense, and special teams - but for most of analytics' history the third phase was an afterthought. The early public work poured into expected points and EPA for offense and defense, while kicking, punting, and returns were treated as a rounding error. That was a mistake. Special teams plays decide field position on nearly every drive, end roughly a quarter of all possessions, and swing close games more often than their share of highlight reels suggests. This article covers how analytics finally took the hidden third seriously: judging kickers by makes over expected, framing the whole kicking game in expected points, the long decline of the kick return, and how special teams earns its own line in DVOA.

One-line idea: every special-teams play either gains or surrenders field position, and field position converts directly into points - so the phase can be measured in the same expected-points currency as everything else.

Judging Kickers: Makes Over Expected

The oldest kicker stat is field-goal percentage, and like raw completion percentage it has a fatal flaw: not all kicks are the same task. A 25-yarder and a 52-yarder both count as one attempt, but one is nearly automatic and the other is a genuine challenge. A kicker who attempts mostly chip shots will post a gaudy percentage; a kicker trusted from distance will look worse on the raw number while actually being more valuable. Field-goal percentage rewards easy kicking as much as good kicking.

The fix mirrors every other "over expected" metric in this textbook: estimate the probability that an average kicker makes each attempt given its distance (and conditions), then measure makes relative to that baseline.

FG makes over expected = Σ (Made − Expected make probability)

Distance is the dominant input, and the well-established shape of the make-probability curve is steep. Framed as ranges rather than invented exact rates:

Inside ~30 yd

Nearly automatic for NFL kickers

~40-49 yd

Reliable but no longer a gimme

~50+ yd

Accuracy falls off steeply with distance

Conditions shift the whole curve. Wind, cold, rain, and a poor field surface all lower make probability; altitude raises it - the thin air in Denver lets the ball carry, extending every kicker's effective range. A good expected-make model prices these in, so a 50-yarder drilled into a cold wind counts for more than the same kick in a calm dome.

Example: why percentage misleads

Suppose Kicker A goes 30-of-32 but almost all his attempts were inside 40 yards, where the expected make rate is very high - the model expected roughly 30 makes, so his makes over expected is about zero. Kicker B goes 28-of-34 but a third of his tries were 50-plus, where misses are expected; if the model only expected about 25 makes, his makes over expected is roughly +3. By raw percentage Kicker A (~94%) looks better than Kicker B (~82%); by makes over expected, B was the more valuable kicker. These are round, illustrative figures, not any real kicker's line.

The Kicking Game in Expected Points

Makes over expected handles accuracy, but the real power of special-teams analytics is folding the entire phase into the expected-points framework. The key insight is that field position is worth points. A first-and-ten at your own 20 is worth far less, on average, than a first-and-ten at the opponent's 40. Expected points (EP) puts a number on every spot on the field, and that lets us value a punt, a kickoff, or a field goal as a change in EP - exactly the same way we value an offensive play.

This reframes punting entirely. A punt that pins the opponent deep is not a neutral "give the ball away" - it is a positive EPA play, because it moves the opponent's next possession to a low-value starting spot. Conceptually:

Punt EPA ≈ EP(our situation before) − EP(opponent’s situation after)

A booming punt downed at the opponent's 5 flips field position dramatically and shows up as real value; a touchback or a short shank that gives the opponent good starting position shows up as a loss. The same logic explains why field goals are scored against their alternatives - a made field goal is three points, but the decision to kick is judged against going for it or punting, which is precisely the comparison covered in our fourth-down decision guide. The kicking game is not separate from the expected-points world; it lives inside it.

Why field position is "worth points": because the expected-points model assigns value to every yard line, any play that moves the ball - including a punt or a return - can be converted into the same EPA currency as a touchdown drive. Special teams stops being unmeasurable the moment you accept that framing.

The Long Decline of the Kick Return

The most visible change in special teams over the past two decades is the near-disappearance of the kickoff return. This is not an accident of style - it is a direct, well-documented response to rule changes aimed at safety. As the league moved the kickoff spot and adjusted touchback rules, kicking the ball out of the end zone (or short of it for a fair catch) became the percentage play, and the share of kickoffs actually returned fell steeply.

The analytics story tracks the rules. When a touchback awards reasonable starting field position, the expected value of taking a knee often beats the expected value of a return, where a fumble or a tackle inside the 20 is a real risk. Returners who once flipped games became specialists used sparingly. Recent kickoff formats have tried to coax returns back by re-engineering where players line up, but the underlying lesson holds: the value of a return is set by the rules around it, and analytics simply follows the expected-value math those rules create. The decline was rational, not timid.

Read the era, not just the number: return averages and return rates are only comparable within a given rule set. A returner from one kickoff format cannot be ranked head-to-head against one from a different format without accounting for how the rules changed the task.

Special Teams as Its Own DVOA Component

The clearest sign that analytics finally took the third phase seriously is that the headline efficiency systems give it a dedicated slot. DVOA - the opponent-adjusted efficiency metric - splits a team's rating into offense, defense, and special teams, treating the third phase as a co-equal contributor to overall quality rather than ignoring it.

The special-teams component is itself broken into pieces, each measured against a league-average, situation-adjusted baseline in points:

Sub-phase	What it measures
Field goals / extra points	Makes over expected, adjusted for distance and conditions
Punting	Net field position created versus expectation
Kickoffs	Where the opponent starts versus the league baseline
Punt & kick returns	Field position gained on returns versus expectation

Adding these up in points lets a team's special teams be ranked first to last just like its offense, and lets analysts say a unit was worth, say, a couple of wins of field-position value over a season. The value of the framing is that it puts all three phases in the same currency: a strong special-teams unit and a strong offense both show up as points above average, directly comparable.

The bottom line

Special teams is a full third of the game that analytics long underweighted, and the fix in every sub-phase is the same expected-points logic used elsewhere. Kickers are best judged by makes over expected given distance and conditions, because a 50-yarder and a 25-yarder are not the same task - accuracy is nearly automatic up close and falls off steeply past about 50 yards, with altitude helping and weather hurting. Punting and field position carry real EPA value once you accept that every yard line is worth points, which is also why field-goal decisions belong in the fourth-down framework. The kick return declined as a rational response to touchback rule changes, not timidity. And the clearest proof the phase matters is that DVOA carries a dedicated special-teams component, split into field goals, punts, kickoffs, and returns. The field-goal-by-distance example above uses round, well-established ranges purely to illustrate why percentage misleads - not any real kicker's stat line.