Drance Numbers is the silly research wing of PITB. While Messrs. Wagner and Mooney blog nationally and solve mysteries, Drance Numbers will look into the minutiae of quantifiable NHL data and bore you with it every Friday.

Before last night’s first period explosion, were you worried about the Canucks slow start, and their early season inconsistency? On Tuesday, the Canucks out-shot and out-chanced the Rangers 2 to 1, but couldn’t beat Lundqvist. In the third, the Rangers got lucky and capitalized on four quality chances. Much hand-wringing ensued as Canucks fans spent the time between Tuesday night and Thursday night’s games exhaustively debating Roberto Luongo, his level of play, his sponsorship deals and most preposterously, his future in Vancouver.

Then the perception surrounding the team pulled a clean 180 in a first period that saw the Canucks score 4 goals on 16 shots against all-universe netminder Pekka Rinne. The goals were scored in some of the silliest ways imaginable: a bank shot by Daniel Sedin, a slap-shot from Henrik that was tipped by former Fish Called Wanda cast-member Kevin Klein, a preposterous Dale Weise tip of a rolling snap-shot from the point, and a Chris Higgins goal off of what should’ve been a failed two-on-one. Maxim Lapierre also hit a post, and Marco Sturm had a goal disallowed that probably should’ve counted.

The hockey gods taketh, and apparently also giveth. The Canucks fortunes could not have flipped more dramatically from one game to the next than they did this week.

Going into last night’s tilt against the Predators, several of the team’s games had been of the “hard to watch” variety, with the game against the Rangers taking the cake. Luongo had been – to put it tactfully – shaky, and the defensive coverage had been even worse. The guns hadn’t shown up in a big way, the power-play was frustrating and the team had already been shutout twice. Sure, all of that sure sounds grim, but the team wasn’t struggling as badly as the results indicated – it was mostly just bad-luck! The devil, as it so often is, is in the percentages.

Despite all the irrational negative energy surrounding the team this week, the Canucks had quietly out-shot their opponents and controlled play at even-strength in most of their games. The bounces were bound to come, and last night, the percentages swung in the Canucks favour in a massive way. Sometimes that’s just how it works.

When judging the performance of a team or a player, we all rely primarily on what we see on the ice. We observe Marco Sturm struggling, we notice Mikael Samuelsson’s indifference, and we wonder, “when is last years version of Jannik Hansen going to show up!” This is all well and good; being an observant hockey fan is an essential step on the way to being a knowledgeable one. But as we’ve talked about previously, your eyes can deceive you, especially when it comes to the ratios and percentages at play over the course of a handful of games of hockey.

Luck may seem “intangible” but stat-heads smarter than myself have long since figured out a rough way of measuring it in hockey. The stat is called PDO, Brian King created it, and I’ll send you over to mc79hockey if you want a more thorough primer. It’s one of the most important advanced metrics in hockey.

Here’s the long and the short of it: PDO is the sum of a specific team, or players, on-ice shooting% and sv%. It can be calculated with this simple equation: Goals divided by shots-for plus saves divided by shots-against. What’s useful about this number is that it’s “unstable”, and over a large-enough sample will regress to 1000. Because of this, we can use PDO to determine whether a team’s struggles over the course of a small sample (say 7 games to start the season) are a legitimate concern, and likely to continue, or not.

In the Canucks case, it was not a legitimate concern; it’s just that the team’s percentages going into last night’s shellacking of the Predators had cratered. Through 6 games their PDO was a 978, but that figure will stabilize and begin crawling back towards the mean, accelerated by random goals-for bursts like we saw in last night’s first period.

For a good example of how “bad luck” can be confused for “lack of effort” or “not playing well enough,” lets check in with Tony Gallagher who yesterday wrote that “Mikael Samuelsson, Marco Sturm, Jannik Hansen and Manny Malhotra,” are “all seemingly struggling to get anywhere near their traditional form.” While we can all agree that Sturm and Samuelsson especially have been underwhelming in the early going, Tony picked the four Canucks forwards with the lowest PDO to single out for their performance. A bounce here or there, and in as little time as it takes to play a single period, the perception surrounding a particular player or team can change dramatically. That’s what we saw last night.

Checking out an individual’s or a team’s PDO number before jumping to a conclusion about their performance is essential in evaluating what is actually going on in the games, and what is likely to occur going forward. Most stat-heads don’t believe in things like clutch play, and only super elite players drive on-ice shooting percentage. We would all do well to pay attention to more than the results, especially in October, and instead pay attention to who is generating shots, and chances. They are of more predictive value than the surface stats (points, goals, assists) that we grew up with.

Tags: , , ,

5 comments

  1. madwag
    October 21, 2011

    Great post! More of the same by Danielson on Backhand Shelf.

    VA:F [1.9.16_1159]
    Rating: 0 (from 0 votes)
  2. Jes Golbez
    October 21, 2011

    It’s a long season, and things tend to smooth out over the course of 82 games. The Canucks skill and talent will lead them to a good record, the Leafs will falter, and the cream will rise to the top. One-off games like the one vs the Rangers happen here and there, but people always treat early-season games as if the whole season will turn out that way.

    VA:F [1.9.16_1159]
    Rating: 0 (from 0 votes)
  3. tom selleck's moustache
    October 21, 2011

    Interesting post. I have a question though. I get the idea that high and low PDO’s would eventually normalize to the mean of 1.00; but I’m curious, and I couldn’t really see this in Dellow’s article, what were the final PDO’s for the top teams vs the bottom teams at the end of a given season? Because, intuitively, you would expect that a higher seed team, like Vancouver, would have a year end PDO that should be higher than, say, a much lower seed team such as Edmonton; and that this would be a general trend you would see with the other top vs bottom teams. If so, how great is this difference?

    Because, if that’s the case, then it makes one wonder what would be the correlation between the sum of individual player’s PDO’s and a team’s overall PDO. Could this then be another metric to look at when deciding to include a particular player on a team?

    VA:F [1.9.16_1159]
    Rating: 0 (from 0 votes)
  4. ArtemChubarov
    October 21, 2011

    Indeed teams do finish about 1000 – the Canucks in fact have finished well above 1000 in each of the past three seasons. Here’s some work the guys over at pensionplanpuppets did on team PDO a few years back with full team PDO numbers for 2008-2010 https://spreadsheets.google.com/spreadsheet/pub?hl=en&key=0AlBZxpXinYkCdFB5WnFJenp1YVBlSF8wRXZjanpXV1E&hl=en&gid=1

    VA:F [1.9.16_1159]
    Rating: 0 (from 0 votes)
  5. Anonymous
    October 21, 2011

    Very interesting post. I completely agree with

    “We would all do well to pay attention to more than the results, especially in October, and instead pay attention to who is generating shots, and chances. They are of more predictive value than the surface stats (points, goals, assists) that we grew up with.”

    VA:F [1.9.16_1159]
    Rating: 0 (from 0 votes)