What Does TOI Mean in Hockey? Time on Ice and ATOI

Ice time is a young official stat. The NHL's stats glossary dates time on ice, and its even-strength, power-play and shorthanded splits, to 1997-98. By then the league had been counting individual skaters' shots for nearly four decades without recording anyone's minutes.

People track it. Rule 36 of last season's NHL rulebook lists a time-on-ice scorer among the Real Time Scorers assigned to every game, next to three stats-entry scorers and an event analyst, and requires their reports during the first and second intermissions and again after the game. The minutes in a box score come from that log, with a league-assigned scoring system manager overseeing the data.

The NHL calls the total TOI and the average TOI/GP. Hockey-Reference calls the average ATOI and prints it in minutes and seconds per game, so the TOI meaning in hockey tables is identical at both sites and only the label moves. The NHL's glossary pins down the arithmetic. A player's TOI is his even-strength, power-play and shorthanded time added together, defined as playing time in all situations, so overtime minutes count, and the league breaks those out in their own OT TOI column.

Ice time is game-clock time summed over a player's shifts, and the NHL sorts every second of it by manpower. Stoppages and TV timeouts add nothing, since the clock isn't running, and time in the penalty box isn't time on the ice. These are the inputs:

• Even-strength time (EV TOI), when both teams have the same number of skaters
• Power-play time (PP TOI), when his team has the man advantage
• Shorthanded time (SH TOI), when his team is killing a penalty
• Shifts, the separate trips over the boards that the time is summed from
• Games played, the denominator for the per-game average

Here's a worked example, with invented round numbers. A defenseman takes 24 shifts in a game and averages 50 seconds a shift. That's 20 minutes of ice time. If three of those minutes come on the power play and two on the penalty kill, his even-strength time is 15 minutes. Keep that up for an 80-game season and he's logged 1,600 minutes.

Analysts use the total as a denominator more often than as a stat. Divide goals, shots or expected goals by ice time and multiply by 60, and you've got a per-60 rate that puts a fourth-liner and a first-liner on the same scale, even when one of them plays twice as long as the other.

TOI measures trust, and position sets the baseline. Among the 327 forwards who played at least 60 games last season, the median averaged just under 16 minutes a night. The 161 defensemen who did the same had a median a bit over 20, on more shifts that ran a couple of seconds longer. Compare a forward to forwards and a defenseman to defensemen, or you'll be comparing two different jobs.

Quinn Hughes led the league at 27:44 a game, and Moritz Seider was third, behind Zach Werenski, at 25:40. Seider played every game at that rate, which made him the NHL's leader in total minutes with 2,104.7, about 50 more than anyone else. Among forwards, Connor McDavid led at just under 23 minutes, and Dylan Larkin's average of just over 20 ranked 29th.

The splits tell you what kind of trust it is. Taking the median of each piece, the typical forward played about 13 and a half minutes a night at even strength, a minute and 40 seconds on the power play and 40 seconds shorthanded. Alex DeBrincat was a power-play specialist, with more than three minutes a game on the man advantage and almost nothing on the penalty kill. Larkin did both jobs (more than three minutes a night on the power play and a minute and a half shorthanded), and J.T. Compher was a penalty killer first.

Why does a coach's opinion belong in a stats glossary? Because it's one of the steadiest numbers in hockey. Among the 379 skaters who played 60 or more games in each of the last two seasons, ice time per game correlated at 0.90 season over season. It barely moves. A player's minutes this year are a strong hint about his minutes next year, and about where his coach slots him on the depth chart today.

Reading the depth chart

Sort a roster by TOI/GP and you've got the coach's depth chart without the press release. It's that blunt. Detroit's two heaviest-used defensemen last season were Seider and Simon Edvinsson, who averaged better than 22 minutes, with Ben Chiarot next at nearly 21. Up front, Larkin, Lucas Raymond and DeBrincat all averaged more than 18 minutes and carried the heaviest loads. At the other end, Axel Sandin-Pellikka's 16:13 ranked 149th of the 161 regular defensemen.

Turning totals into rates

James van Riemsdyk's 31 points in 72 games look modest at 0.43 a game, a step behind Andrew Copp's 0.54. Per 60 minutes the order flips, 2.20 for van Riemsdyk to 1.97 for Copp, because van Riemsdyk did it in less than 12 minutes a night to Copp's 16 and a half. Some of that is power-play time (almost two of van Riemsdyk's nightly minutes came with the man advantage), which is the first thing to check behind any rate.

Tracking workload

Seider's 2,104.7 minutes came on 28.6 shifts a game, and he didn't miss one of Detroit's 82 games. That total is a defenseman's version of a goalie's starts, the plain workload number, and nobody in the league carried more of it. The NHL's share versions (PP TOI% and SH TOI%) show the same thing for special teams, the slice of a team's power-play or penalty-kill time that one player soaked up.

It records the coach's opinion

Minutes say how much a coach played someone. Whether the player earned them is a separate question, and on-ice goal and expected-goal shares are the stats that answer it. TOI can't tell the difference.

Power-play time inflates rates

NHL teams scored almost three times as many goals per minute at five-on-four last season as at five-on-five (7.32 per 60 to 2.48, according to MoneyPuck's season file), so a big power-play role lifts any all-situations per-60 number. Split the rates by situation before you compare two players.

It doesn't say where shifts started

Two players with the same 18 minutes can face different work, one starting in the offensive zone against third lines and the other taking defensive-zone draws against the other team's best, and TOI records none of it, which is why zone starts and quality of competition exist as stats of their own.

One bad night drags the average

A player who leaves a game hurt after one shift still has that game in his denominator, because a single shift earns a game played, so his TOI/GP drops through no choice of his coach's. It's a small effect over a full season and a big one over five games.

Do goalies have time on ice?

Yes, and it's the denominator of goals-against average, which the NHL calculates as goals against times 60 divided by time on ice. Karel Vejmelka led all goalies with almost 3,700 minutes last season. John Gibson's 3,181 minutes and 144 goals against work out to his 2.72 GAA.