Kevin Bieksa, appearance vs. reality

There are numbers for everything in sports.

Luck? Sign me up.

I got into a bit of a discussion on Twitter with another competitor in the Province’s #ReplaceTheKB competition last night about the play of Kevin Bieksa. Bieksa, I’m sure you’re all aware, is a team-low minus-9, has yet to score a goal, and has been on the ice for 21 goals against so far through just 14 games.

May I suggest to you that Bieksa is going through an unlucky streak. This is something that we can quantify with something called PDO. What is PDO? Follow me after the jump.

I often get asked what “PDO” stands for, but the truth is that it is simply the Internet-handle of the guy who invented it in the comments section of a long-time hockey stats blog run by Vic Ferrari. PDO, simply, is the shooting percentage of a player’s team when he is on the ice at even strength plus his save percentage.

Over the course of many games, this number will approach 100%. Anything above 100 we call “unsustainably high” and anything below it we call “unsustainably low”. The idea is that the best players are the ones with the highest shot differentials, and, over the span of a season, the goal totals sync up with the shots taken.

Kevin Bieksa, among defensemen with at least ten games played, has the sixth lowest PDO among NHL defensemen who have played at least 10 games. Last season, Canucks goalies saved 94.2% of the shots against when Bieksa was on the ice. In the playoffs, it was 93.3%. This season tells a different story, as that number has dipped to 87.5%, and Bieksa’s overall PDO has dropped to a gaudy 93.1%. He can’t take all the blame for the things that have happened to him and Keith Ballard this season. Some bounces have done the trick as well.

The way that I can best describe this effect is that the skill required to score a goal is overstated. The true talent in hockey is creating shots at one end, and preventing shots at the other. It is the core tenet of advanced hockey statistical analysis, and the geeks like to talk about “shot differential” when discussing players rather than pure goal differential.

This effect is easily seen on a player such as, say, Maxim Lapierre. Lapierre had a rough go last season, with him and his teammates only putting 5.8% of their shots in the net last season. His PDO was the lowest on the Canucks at 97.9%. This season, his PDO has started off in the opposite direction, with his on-ice shooting percentage at 11.1%. His PDO has gone on an up-swing as a result, at 104.3% Thus far, there are very positive reviews about Lapierre’s play, and while it has been good for a fourth line player, what you see out of Lapierre is the reverse of what’s happened with Bieksa.

PDO understands our emotions. If it seems like a guy is skating to the bench head bowed in solemnity, we recognize that. But PDO is slightly more sophisicated, understanding that there is a certain fluke about being on the ice for a goal against. A player has about a 3% chance of getting scored on every shift he takes.

When we think that a player isn’t playing up to par, if you check his PDO, you can expect to see it a little lower than 100. The reverse effect happens as well. If a player looks to be at the top of his game, it’s probably because he’s riding the percentages, and that will expect to come down. It’s a measure of luck, but in the more personal sense, it’s a measure of our personal emotion.

-Ville Leino was never really good, but since the playoffs of 2010 until this season he’s been lucky. This is what the stat community calls “riding the percentages”.


It is an easy statistic to find, and the most important of all the advanced statistics, one that is simple enough to understand and gives us so much more perspective when crafting opinion about hockey players. Go to, and under the “Statistics” tab go to “Player Breakdown”. 

From there, there is an almost un-navigable sea of numbers and ratios, but if you click on ‘On-Ice Shooting’ on the options menu on the right side of the screen, along with your criteria for which players you want to show up (team, position, minimum games played, year) you are left with his on-ice shooting percentage and on-ice save percentage, added up on the right-most column to form PDO.

The Canucks numbers from this season can be found here.

If you scroll through that, you can note some of the highest PDOs in the NHL last season. Nathan Horton had the highest among playeg last season at 103.8%, while David Booth was the lowest at 95.8% (and that shows you how the Canucks “bought low” when it comes to Booth). Those numbers are closer to 100 than the numbers we’ve seen at the start of this season but players have been known to have bad luck over a long period of games (or good luck: see Leino, Ville, who had a 109.5% PDO in the 2010 playoffs, a 102.3% PDO last season with Philadelphia, and has started his Sabres career with a much lower 98%, leaving people wondering whether his contract really was that shrewd of a signing).

So rather than make judgments of hockey players based on a couple of games where a player appears to have been sub-par, just keep in mind the perspective and the larger context. Over a small number of games, goal totals don’t quite match quite up to the number of shots for and against, but in the long scheme of things, it generally does.

Kevin Bieksa will turn it around. Well, he won’t, necessarily, but his percentages are unsustainably low, so he could play the exact same way towards the end of the season, and his +/- number will slowly begin to improve.

  • it seems like a deceiving stat. Take for example Max Lappierre last year. You say his line scored on 5.8% of their shots and yet his PDO was 97%. Shouldn’t it have been in the 20-30% range? Sorry this stat is too complicated I think +/- is a more telling stat because it tells how many times a player was on the ice when his team score and when they were scored on. Also how do you define a scoring chance? A player can shoot the puck from center ice and score as has happened on occasion. To me, any shot has a chance of going in and thus should be considered a scoring chance.

  • Calling anything above 100 unsustainably high and anything below unsustainably low is a major exaggeration. Depending on your team’s goalie, something like 101 can be very sustainable for a ranger or canuck or bruin. Additionally skill of line mates factors in. It is luck driven, but there is more to the stat then just luck

    • Oh, it can be. That’s why teams with better goalies have players with higher PDOs. But you take that player off of that team and stick him on a team with a worse goalie, and he won’t keep up that nice +/- number that he’s put up.

      Similarily, take David Booth out of Florida (Vokoun did a pretty poor job when Booth was on the ice) and his +/- should improve with better goaltending.

      Looking at shot statistics, you can estimate each player’s +/- in front of a normalized goalie.

  • @Tina A scoring chance is a shot taken within “home-plate.” While any shot has a “chance” at going in, research has shown that these are what we should (and do) consider “higher percentage opportunities.”

    As for PDO’s complexity: it is the simplest advanced stat around. Add one number to another, and produce a sum. Really quite simple.

    +/- is flawed because it is so luck based, you can get a – for jumping off the bench for a line change at the wrong time, or because your goalie duffed an easy save attempt. It’s certainly of less value when analysing the quality of an individual skaters contributions than that player’s underlying numbers and PDO…

  • @tina

    The other reason that +/- is flawed is that it doesn’t take into consideration the context of play. Players that take more face-offs in the defensive zone against the other team’s Crosbys and Ovechkins are more likely to see scoring chances against. The Sedins and Ehrhoff had high +/- last year in part because AV was using them as much as possible in the offensive zone, and against weaker competition, whenever possible (fourth line icing? here come the Sedins…).

    And, there are many other events in a game besides goals, so that’s why stats geeks turned to using Corsi, scoring chances and the like to evaluate players.