In the build-up to Canada’s first game of the Olympics, against Norway, Mike Babcock talked about wanting to reduce the length of the average shift played by Team Canada’s players from 45 seconds to 40 seconds. It’s not as significant a change as it sounds.
Unfortunately, the NHL doesn’t always present the shift-length data that they gather in the most useful way. One of the unfortunate tendencies on the league site is to mash together data from even strength, the power play and penalty killing situations, even though they are substantially different situations, in terms of the demand that they place on skaters and how changes are conducted.
This makes it harder to see how coaches really use their bench. Fortunately, with the league’s play-by-play files and some of the cool tools developed by Red Line Station, we can look at how the length of a specific player’s 5v5 shifts vary. Let’s use Corey Perry as an example.
Perry is a star on the Anaheim Ducks. In the 2012-13 season, he played 15:19 per game at even strength (ES), which ranked 45th in the NHL. In other words, his ice time was pretty typical for a first-line forward. He played 836 5v5 shifts and his average 5v5 shift lasted about 46.9 seconds. Going from an average of 46.9 seconds to the 40 seconds that Babcock wants sure sounds like it’s a big deal—it sounds like he’s being asked to make his average shift 15 percent shorter.
That’s not really the case, however. The vast majority of Perry’s 5v5 shifts—about 76 percent of them—were 60 seconds or less in length. The average length of Perry’s shifts that were less than 60 seconds is right below Babcockian Perfection—38.5 seconds in length. Most of his shifts aren’t going to be affected by a dictate from the coaching staff to shoot for an average of 40 seconds.
That still leaves about 24 percent of Perry’s shifts that were 61 seconds or greater in length. These are the shifts that drive Perry’s average 5v5 shift length up from 38.5 seconds to 46.9 seconds.
If we look at these shifts a little more closely, we see that we’re dealing with 200 shifts that average 74 seconds in length. It’s a bit misleading. The way in which the NHL measures shift length doesn’t take into account the fact the NHL breaks three times a period for TV commercials. Olympic games are three periods in length; NHL games are more like twelve period games, with nine of the intermissions occurring at the first whistle after a fixed point, and three longer intermissions during which the rink is resurfaced.
Looking at Perry’s 2012-13, of his 200 61+ second 5v5 shifts, 31 of them were bisected by TV timeouts. The NHL’s data doesn’t deal with this very intelligently—for all intents and purposes, these are two shifts rather than one. When you take that into account, Perry’s average 5v5 shift length drops from 46.9 seconds to 45.3 seconds.
This isn’t the only factor at work. Star players are more likely to be given leeway at the end of games that their teams are losing to chase a tying goal or play an extended shift with the goaltender out of the net. Four of Perry’s long shifts were shifts with the Ducks goalie pulled as Anaheim tried to get back in the game. Star players are probably also more likely to benefit from a timeout called to give them a chance to catch their breath when chasing a game-tying goal. This all adds up, and it isn’t captured if you think in terms of the average shift.
If you look at the timing of Perry’s long shifts, they disproportionately occur in the second and third period. Perry’s time on ice is split fairly evenly between the first, second and third periods but 74 percent of his time spent playing long shifts occurs in the last two periods of games. The fact that it’s not evenly spread suggests that it’s at least partly a tactical decision on the part of the Ducks’ coaching staff as opposed to something intrinsic to Perry’s game. Perry (and, assuming similar distributions of ice time, Canada’s other skaters) aren’t really being asked to change most of their shifts and make them shorter; they’re just being asked to cut out the really long shifts.
It’s understandable that the Ducks would lengthen the leash on Perry as a game goes on. Coaches shortening their bench isn’t uncommon and, at the NHL level, hockey games are within a goal a lot of the time. In 2012-13, Anaheim spent about 73 percent of their 5v5 time within a goal of their opposition. As the game wears on, it makes sense that you’d want Corey Perry out there if you were Bruce Boudreau.
Of course, Boudreau can’t look down his bench and see what Mike Babcock will see for the next week and half. His options are different. The factors that drive long shifts for stars in the NHL—TV timeouts, regular timeouts, empty net situations and, most importantly, the lack of remotely comparable players on the bench—simply don’t exist at the Olympics, at least for Canada.
Canada’s depth would seem to make this a pretty good strategy. It’s unlikely to be a particularly significant change for the players on Team Canada—they presumably have shift distributions that are pretty similar to Perry’s, with the vast majority of their shifts being consistent with a 40-second average. What this move will change is that, perhaps unlike with their club teams, Canada’s skaters should be able to maintain a higher pace later into games.
Given that depth is Canada’s greatest strength, it’s good that the coaches seem to be focused on how best to take advantage of it. It will be hard, if not impossible, for other teams to match it. They just don’t have the same depth. It will be worth paying attention, as games wear on, to whether or not Canada appears to be having stronger third periods as they’re able to keep rolling four lines and playing short shifts for the entire game.