Thursday, November 10, 2016

Taking a PDO-based look around the league

We’re almost a month into the season, and while the standings are still in flux, teams are starting to settle into tiers. Some are overachieving, others are underperforming, and some are right where we thought they’d be. And if history is any guide, a whole lot of teams will end up somewhere very different by the end of the season.

But which teams? Which of the early-season trends have been real, and which are a mirage? We can’t know for sure. But we can get some important hints by looking at one number for each team: their PDO.

As with many of today’s “advanced” stats, PDO is pretty simple, and most fans are familiar with it by now. (If you’re not, you can find a good explainer here.) It’s just a team’s shooting percentage plus its save percentage, with an average team adding up to 100.

Most teams drift towards that 100 mark over time. That can take a while, but anyone who’s significantly above or below the mark is a good bet to see their fortunes turn around. The easiest way to think of PDO is to treat results outside a realistic range like a warning light on your dashboard. It doesn’t necessarily mean there’s anything wrong, but you’ll want to check into it to make sure.

So what is a realistic range? There are no firm numbers, but we can look at history for some guidelines. We can only calculate PDO dating back to the 2007-08 season, which gives us a total of 270 team seasons to work with. The highest and lowest seasons on that list can give us a sense of what's realistically sustainable over a full season.

The highest PDO over that span belongs to the 2008-09 Bruins, who shot 10.9% as a team and had a .925 save percentage for a PDO of 103.3. Not surprisingly, they had a great year, posting 116 points. It wasn't sustainable, as they dropped to a nearly dead-center 99.9 PDO the following season and fell to 91 points. But the Bruins of that era were an example of a team that generally could sustain an above-average PDO, largely thanks to spectacular goaltending by Tim Thomas and Tuukka Rask; they had two other seasons where they were over 102, including a 102.7 in 2013-14.

The next two best PDO years were by the Maple Leafs and the Penguins, both in 2012-13. Remember, that was a short season due to the lockout, so you'd expect to see more outliers. That Leafs season led to a furious debate over whether it was sustainable, with analytics fans insisting it couldn't be while the Leafs front office vowed that it was. We know how that turned out.

Meanwhile, the worst PDO season ever recorded was by two teams from that same lockout-shortened season, the 2012-13 Panthers (96.3) and Devils (96.9). The worst mark over a full season belongs to the 2014-15 Oilers, who shot a reasonable 8.2 but got awful .888 goaltending to finish at 97.0. They're one of 12 teams to finish a season under 98.0.

So that gives us a range to work with. Anything above 102 or below 98 is extreme, and anything above 103 or below 97 would be almost unheard of over a full season.

We're still very early in the 2016-17 season, so we'd expect to see some teams well outside of those boundaries. And indeed we do. Let's look at the league's five best and five worst PDOs heading into action tonight, as well as a few other teams that are worth mentioning.

(All numbers are across all situations unless otherwise noted. PDO numbers can vary slightly depending on what site you're looking at based on small differences in how the numbers are calculated; we'll be using hockey-reference.com's totals.)

New York Rangers

Well, you knew they were going to show up. And they do, right in the top spot, with a PDO of 105.2 across all situations. That's miles above any kind of reasonable cutoff, so clearly what the Rangers are doing right now can't continue.

What the Rangers are doing, of course, is scoring a ton of goals. They'd scored five or more in five straight games before Tuesday's 5-3 loss to the Canucks, and they lead the league by a mile with an average of 4.14 goals-per-game. Not surprisingly, they're doing all that scoring with a ridiculously high shooting percentage of 14.0%; for comparison, the best team shooting we've seen in the analytics era was 11.6% by the 2009-10 Capitals.

So the Rangers shooters have to cool off. Then again, that's not telling us anything we didn't already know – even the most diehard Rangers fan would have acknowledged that they weren't going to keep scoring like the mid-80s Oilers.

What may be more interesting about the Rangers' sky-high PDO is that it's almost entirely shooting driven. Their save percentage clocks in at 91.2%, good for 14th in the league. If anything, that seems low for a team with Henrik Lundqvist. And it is – the Rangers have finished with a better save percentage four times in the past five seasons.

So overall, we know the Rangers' shooters will cool off significantly, although we probably didn't need PDO to tell us that. But we might see the other half of the equation trend up, meaning the Rangers are a team that could realistically come in close to that 102-or-so range by the end of the season.

Columbus Blue Jackets

This early in the season, it's unusual to see a team with a very high PDO that isn't doing well in the standings. Unusual, but not impossible, as the Blue Jackets demonstrate. Despite tying the Rangers with a PDO of 105.2, they've only won a middling six of 11 games and are sitting just outside of a wildcard spot.

So what's up? The Blue Jackets' numbers are driven by good shooting and excellent goaltending. We'll leave it to you to decide if you think Sergei Bobrovsky can maintain his early season pace. But on the shooting side, there's no great mystery here: That 10-0 win over the Canadiens last week is throwing everything off. Take away that one game, and their 11.2% shooting drops to an 8.9%, or slightly below league average. You don't want to get too clever with cherry-picking games, but unless you think the Blue Jackets will mix in a double-digit blowout every month or so, it's not hard to see where this number will go.

Montreal Canadiens

Speaking of the Canadiens, they slip in right behind the Rangers and Blue Jackets with a PDO of 105.1. Like Columbus, they're shooting well, scoring at an 11.6% clip, but it's their .935% goaltending that's really driving their number.

If you're a Montreal fan, that's good news, since league-leading goaltending from Carey Price seems like something that could be sustainable in the long run. He's putting up a .953 save percentage right now, which has to come down – nobody's ever been better than .940 over a full season – but isn't ridiculously better than the .930 or so he's put up over the last three years. Like Lundqvist or those Bruins goalies from years past, he's the kind of goalie who gives his team a shot at getting well beyond that average PDO of 100.

As for the skaters, guys like Alexander Galchenyuk and (especially) Shea Weber will see some regression, and on paper the Canadiens don't seem like a team that's likely to shoot anywhere near 11% over a full season. So yes, some of Montreal's success this year is fueled by percentages that are unlikely to last. On the other hand, at least they're banking points while they're hot.

>> Read the full post at Sportsnet




1 comment:

  1. Is there an advanced stat that only counts shots for/against that missed the net? If not can we name it after J.T. Miller?

    ReplyDelete