Simplest statistic to rate performance
Tally of shooting and save percentages gives a good idea of the effectiveness of players, teams
In listing the most important advanced hockey statistic for his website earlier this year, statistical guru Gabe Desjardins, a Canadian-born turned California-based engineer, wrote: “If we had a reliable way to measure how many wins a player contributed to his team, that would be our choice.”
Instead, Desjardins, who runs the Winnipeg Jets blog Arctic Ice Hockey, formerly known as the advanced hockey statistic Behind the Net, wrote: The simplest, yet most-useful statistic is PDO.”
I often get asked what PDO stands for. The truth is that it stands for nothing and was simply the Internet handle of Brian King, a hockey fan who came up with the idea behind PDO in the comments section of “Irreverent Oilers Fans,” an Edmonton Oilers hockey blog.
Quite simply, it was the addition of a team’s 5-on-5 shooting percentage, and save percentage, when a player was on the ice. The basic formula was then adapted and applied to teams as well.
“The idea, and I know this will make some people uptight, is that there’s unsustainability in that number — if it’s high, it’s going to come down; if it’s low, it’s going to go back up,” another Oilers blogger, known online as mc79hockey, wrote back in November 2008.
The same post showed that the 2006 Ottawa Senators, who had one of the fastest starts in NHL history, also had the highest first-quarter PDO between the 2003-04 and 2007-08 seasons.
After the first quarter of that 2006 season, where the Senators had piled up an astonishing 17-3 record, the Globe and Mail published a headline asking, “Are the Sens the best team ever?”
The answer turned out to be, of course, no, and their success largely hinged on percentages, not play.
The Senators had a team save percentage of .954 and a shot percentage of 12.6 for a PDO of 107.9 for those first 20 games. Over the larger scale of games for the rest of the season — the final 62 games — the Sens’ PDO would drop to 100.2, along with their record dropping to 35-18-9.
Since each shot must equal a save or a goal, the PDO of the entire NHL is 100. The longer the stretch of games, the more likely it is that a team or player’s overall PDO will inch toward that number. There are outliers each season, but they tend to come back to earth the following year.
PDO is the reason why modern hockey analysts prefer to use shot indicators such as the Corsi number, not goals, as a measure for the quality of overall team play. The PDO effect can work in reverse as well.
In his post, mc79hockey found that the 2007 Philadelphia Flyers, after a 92.7 PDO at the 20-game mark, finished up with a 98.1 PDO over the next 62 games, making their record much better (although not enough to ensure that the team didn’t finish last overall — but they were a miserable possession team, as well).
It’s an idea that goes against common intuition as a hockey fan.
Are some shots not more likely to go in than others? Clearly, but the thing is that PDO works.
After a small number of games, players who are “hot” will have high PDOS, as will certain teams. By paying proper attention to PDO, fans, newspaper columnists and radio hosts may avoid making terrible mistakes about players or teams based on results over a small sample.
For instance, Kevin Bieksa drew a tonne of criticism in the Vancouver market for his plus/minus rating after the Canucks’ first 20 games; he was a minus-five at even strength, with the Canucks scoring 11 goals when he was on the ice and 16 against, despite a very favourable shot rate.
His PDO — adding up shooting and save percentages — was just 95.6. Over the next 27 games, a slightly larger sample, Bieksa is a plus-10 in the same situation, but with a 103.9 PDO.
The wild PDO swing is not uncommon, but the high number is probably bound to regress.
Probably the best example from this season comes from the Minnesota Wild, who, as late as Dec. 12, had won 20 games out of 30 and were first overall in the NHL.
Since then, the Wild have won just two of 17 and, before Saturday’s game, were out of a Western Conference playoff berth.
Excuses range from injuries, to effort, even coaching, but the savvy commentator will notice that, after those first 30 games, the Wild, according to Irreverent Oiler Fans’ sister site Time on Ice, had a PDO of 102.1. What happened next was elementary. (PDO numbers can be found at behind the net.ca, under the “player breakdown” link from the “Statistics” tab.)
Cam Charron is a freelance writer and member of Province Sports’ Legion of Blog. You can find his work at theprovince.com/legionofblog, at canucksarmy.com and he writes monthly for the White Towel. Follow him on Twitter @camcharron.