Over the past few weeks I’ve taken a look at some of the advanced statistics (Corsi, Fenwick) that have been getting a lot of attention lately and an example of what they can be used for. Another statistic that has been gaining momentum recently is PDO, and like Corsi and Fenwick, this ‘advanced’ statistic is relatively simple to calculate.
Originally conceived in 2008 by (and named after) Brian King (username: “PDO”) in the comments of an Edmonton Oilers blog, the PDO of a team is the sum of their save % (shots saved/shots allowed) and their shooting % (goals for/shots for).
Since the sum of all saves and goals scored league-wide equals the league’s shot totals, the PDO for the entire league is 100%. As the season wears on, a team’s PDO will normally regress towards the league average of 100%(1).
The idea behind PDO is then to look for the teams that are the most above or below 100%, as they are probably playing at an unsustainable pace and are likely to see changes going forward as their PDO trends toward 100%. Now in late January half of the league falls in the 99%-101% range.
Through the month of October, Winnipeg (4-5-1; 45 Pt%) had a PDO of 96.8%. Now over halfway through the season, their PDO has rebounded to 101.1%, their record has improved to 26-14-8 (62.5 Pt%), and they are sitting quite comfortably in the first Western conference Wild Card position. At the other end of the spectrum, Ottawa began the season strong (5-2-2, 66.7 Pt%) with a PDO of 103.2, but currently they sit 10 points out of a playoff spot (19-18-9, 51.1 Pt%) with their PDO having regressed to 100.8%.
To try and predict effects of PDO with playoff implications, we look to teams either just inside the playoffs with a high PDO (that may see their performance dip in the near future) or teams on the outside looking in with a low PDO (that may see bounces start going their way).
In the Eastern conference, Florida currently has a PDO of 98.9% while Boston’s is 100.2%, and while I would expect to see some improvement from Florida moving forward, their 10 point deficit behind Boston may be too much to overcome, even with their 4 games in hand, and I doubt they will be able to knock the Bruins from the final Wild Card position. The Western conference is a little more interesting when we try to predict using PDO, as Calgary, holding on to the final West Wild Card position, is currently slightly high at 100.8%. Challenging them for that spot are the defending Stanley Cup champion LA Kings with a PDO of 99.7% and Dallas with 99.9%; with Colorado (100.7%) and Minnesota (97.7%) being long-shots to compete. Should Calgary falter at all in the next couple of months, at least one of these teams will be able to take advantage and move into the final playoff spot. I suspect this Wild Card position will change hands several times before the season finishes, with LA being my pick to finish ahead of the others.
In keeping with the trend in my articles on Corsi and Fenwick ratings, I wanted to see if I could use PDO to predict playoff results. I compared how far teams have made it in the playoffs with their PDO rank (1-30) like I did with Corsi and Fenwick. I also reseeded teams once in the playoffs by their PDO rank.
In the first graph, it is obvious that Corsi and Fenwick are a much better predictor at post-season success, as PDO doesn’t seem to be a good predictor at all. Looking at the reseeded results confirms that a team’s PDO at the conclusion of the regular season has no impact on how well they will do in the playoffs, which makes sense as we know all PDO ratings trend toward 100% throughout the season. However, PDO’s strength shows in the regular season by predicting if a team is over-performing or under-performing their true potential.
1. PDO Explained, blogs.thescore.com
2. 2005-2014 PDO statistics from sportingcharts.com
3. 2014-2015 PDO statistics from war-on-ice.com