The Toronto Maple Leafs’ inevitable regression

Back-track to February 8th. In analyzing the then-Toronto Maple Leafs’ chance at a playoff spot, I wrote the following on this blog:

Should the shooting falter, I’m not convinced the team has the defense or goaltending to stay in a playoff spot.

A commenter who went by the subtle handle of “advancedstatsareafraud”, more troll than anything, responded to the facts (the team had very modest possession rates and high percentages that were carrying the team at that point) with the following point:

When the leafs, with excessive PDO and way too low Fenwick, make the playoff – then finally people will come to realize that advanced stats are a complete fraud. The leafs laugh and are better then all this “regression to mean” talk.

This isn’t necessarily to toot my own horn or the horn of the advanced statistical movement that has gained momentum in the hockey blogosphere, but it gives us a perspective at how team records are illusory. 60 minutes is not enough time to determine the best team in a hockey game, and neither is the next 60.

Even after a full 82-game season, which lasts around 5000 minutes, you still don’t quite know exactly who the best team in the league is, and generally coaches and management will make decisions based on a mirage in the off-season to attempt to compete the following year.

That rarely works out.

Case in point, Brian Burke’s former team, the Anaheim Ducks, were one of the worst possession teams in the NHL last season, and yet, after a strong second half with shooting percentages, they began to win some games and were the 4th playoff seed despite being massively out-shot throughout the regular season and having scored just four more goals than their opponent. Unfortunately, however, the Ducks were well out of their element against the Nashville Predators, lost in six games, unable to stave off fate for much longer, and went into the offseason with optimisim rather than regret.

Rather than recognize the issue, general manager Bob Murray traded down at the draft and staved off any attempt at a re-build, going with the same core for the next season, only to have the Ducks crash and burn out of the gate and not improving until they hired former Washington coach Bruce Boudreau after firing newly-minted Leafs coach Randy Carlyle. Boudreau has helped the Ducks’ fortunes considerably, being a possible cause for better underlying performance leading to a better wins and losses record. Boudreau’s replacement in Washington, Dale Hunter, has seen the reverse happen and, fooled by low percentages in the early part of the season, the Capitals are now worse off with one fewer guy around to share the blame.

For a manager, tracking the underlying numbers are important because they give you a clearer picture on your team. Whether Brian Burke calls advanced hockey statistics “horses***” or not, the advanced stats in February showed a team much worse than its record indicated. At some point, percentages will bite back, and the Leafs’ have certainly felt that effect over the last 17 games. This slump could be a blessing in disguise, however; the Leafs weren’t a good enough team to compete in the postseason, and it means that Brian Burke won’t take the team’s first half successes too seriously when it comes time to shed or recruit bodies and salary in July and August.

When I say “percentages” I generally refer to PDO, the most important advanced statistic. It is the simple addition of team shooting percentage and team save percentage, usually calculated at 5-on-5 situations only. Since every shot must result in a save or a goal, the league average PDO is 1, or 1000, or 100%. Part of the reason PDO works (teams slowly trend to 1 throughout the season) is because the more open a game your forwards play and the more scoring chances they generate, the more open they’ll leave themselves on defence. When you “trade chances” you are literally doing this—trading chances, and no team has shown a real ability to have a free-flowing game on one side and tighten up in their own end. The best example are the Tampa Bay Lightning, who are first in the league in shooting percentage, but dead last in the league in save percentage.

Teams with better goalies can expect a PDO of slightly over 1 to account for the fact that their save percentage will be higher in the long run, but for Toronto, who don’t have a goalie who has posted above league average save percentage throughout a good length of time, they should expect a PDO of slightly below 1, in the 98.5% to 99.5% range without getting some serious luck shooting the puck as Phil Kessel and Joffrey Lupul were getting earlier in the season.

This graph here will show how the Leafs’ PDO has balanced out over 7-game segments, each representing one tenth of the season:

[I generated data via timeonice.com scripts – a terrific FAQ can be found here]

As you can see, the Leafs’ PDO never dropped below 1 until the final segment. The Leafs have been stuck with some unfortunate luck in the last 14 games, posting a shooting percentage of 6.6% in Segment 9 and 6.5% in Segment 10 (the season average of shooting percentage before that was 9.2%).

To follow the Leafs’ individual shooting and save percentages by segment…

7-game Segment Segment Save% Segment Shot% Cumulative Sv% Cumulative Sh%
1 91.6% 11.8% 91.6% 11.8%
2 90.7% 11.1% 91.1% 11.4%
3 92.2% 3.9% 91.6% 8.6%
4 91.9% 11.1% 91.6% 9.2%
5 93.5% 8.3% 92% 9%
6 92.8% 11.2% 92.2% 9.3%
7 91.2% 9.1% 92% 9.3%
8 91.4% 8.4% 91.9% 9.2%
9 87.7% 6.6% 91.4% 8.9%
10 90.8% 6.5% 91.3% 8.6%

The shooting percentage started off strong, fell a bit between games 15 and 21, and really picked up, as did the save percentage, between games 36-42, would would have been late December/early January (Toronto went 4-2-1 in that stretch with two shutouts and a 7-3 win over Tampa Bay and a win against Detroit). But it has really come back to earth and its been noticeable in the last few weeks, with Toronto going, as it isn’t either worth repeating, x-y-z over the last 17 games.

If Burke is to make any additions in the offseason, his mindset has to be getting players who can control the shot clock and bring the Leafs up to a team who has the puck more often than the other team. In a game where so much is decided by unreliable, unrepeatable and unsustainable percentages, it’s better off being a team that doesn’t have to bank on them to succeed. Right now, Toronto isn’t one of those teams.