WASHINGTON – So your favorite hockey team is off to a great start. Twenty games into the season, they’ve shocked the world and are first in the division after being predicted to finish last.
You’re thinking playoffs and looking forward to a bright future.
Then it falls apart. Sure, maybe your team still makes the playoffs. Maybe it even wins a series or two. But by the next season, it’s back to the basement, the short success merely a blip in hockey’s radar.
What happened?
Well, evidence of the probable downfall was there all along. PDO, an advanced – and obscure – stat that’s found by adding a team’s shooting percentage and save percentage together, could have warned you.
Here’s how it works: because the goals for and goals against for the entire league are even, league average PDO is roughly 100 percent every year.
More than 100 is considered above average and less than 100 is below average. A PDO between 98 and 102 percent is seen as reasonably sustainable, according to co-editor at hockey-graphs.com Ben Wendorf, a website popular in the hockey analytics community.
Anything outside of that threshold, though, probably means something will change.
“Generally speaking, we don’t see teams sustain levels much higher than two percent points above league average,” Wendorf said. “As long as you’re within the threshold, we generally allow that a team can have that much range of talent. Once you get to say 103, 104, we’re going to say that there’s a percentage in there that’s unsustainable.”
How does a team get a PDO above 102 or below 98? Wendorf said it’s from luck.
“(PDO) measures luck without a doubt,” he said. “Luck is always a part of performance.”
The statistic was started by Brian King, a native of Edmonton, Alberta, and an Oilers fan who frequented the blog site Irreverent Oiler Fans in the mid-2000s.
When Vic Ferrari, the pseudonym used by the site’s founder, posted on-ice goals for (meaning all goals scored when the player was on the ice), on-ice shooting percentage, on-ice goals against and on-ice save percentage for each Oilers player, King’s wheels started turning.
“You could see there was a bunch of players who, some were incredibly lucky and some just couldn’t score to save their lives, or they were getting lit up and it wouldn’t really make sense,” King said. “It occurred to me as someone who had played hockey that there’s times when you’re out there and you can do everything absolutely perfect, and the puck ends up in your net, or you set up your linemate and it hits the post.”
King added the on-ice shooting and on-ice save percentages together and created PDO. The name is not an acronym and stands for nothing. It comes from King’s username on Irreverent Oiler Fans and was originally used as his handle for the video game Counter-Strike.
Although using PDO as a team stat has become popular within the analytics community, King originally made it as an individual player stat and said he prefers it that way.
“There’s so much volatility on the ice. There’s so much you can’t control,” King explained. “We know a team that has Carey Price or (Henrik) Lundqvist on it is going to be a PDO leader.”
“With teams I always thought you lost just a little bit of it,” he added. “For individual players, you’re dealing with a smaller sample size than you are with a team, obviously. I think because of that there’s a lot more room for a whole lot more variance.”
Advanced stats help fans better understand the game they’re watching, but fans aren’t putting together NHL teams. As analytics have gained credibility in hockey, teams have opened their minds to the world of numbers, including PDO.
However, Wendorf said that while PDO could be useful for front offices to understand when to sell a player high or buy low, it can be a sticky situation.
“If things are going really well, how is it going to look for you to turn around to your president or team owner and say, ‘Alright, we need to trade one of our top scorers now because we think those percentages are going to fall,’” Wendorf said.
“Conversely,” he said, “if teams are performing terribly, the puck bounces aren’t going their way, you could sit there and say, ‘Well, the bounces are going to come.’ But a president or team owner, they’re like, ‘Well, we want to win now and we want to win all the time. We don’t want to wait. And it’s not a really good pitch to our fans to say just wait a bit.’ I think there’s a pretty strong love/hate relationship with it because PDO can get you to do things you really don’t want to do as a front office.”
Calgary Flames Director of Hockey Analytics Chris Snow said he uses PDO in his work.
While he admitted “luck” isn’t a well received word in the sport, PDO can be presented in a front office because of its “conversational” nature and high stakes of the league.
“Hockey as a business has never been more competitive,” Snow said. “I think as a consequence, there’s a receptiveness to, ‘We have to be informed, we have to understand when things are going well why, when they’re not why.’ (PDO) is one metric among many that do get discussed that people are receptive to. It’s easy to explain, easy to understand and it’s got some validity to it.”
While PDO has become a poster child for luck, it has its limitations.
Wendorf, King and Snow all said PDO is just one piece of the puzzle that can paint a larger picture. Corsi, a measure of puck possession, Fenwick, an advanced measure of Corsi, and other analytical tools can help outline a team’s underlying talent more accurately.
“If it’s within those thresholds, we probably won’t pay attention to it too much,” Wendorf explained, referring to when a team’s PDO falls between 98 and 102. “That’s why we tend to run our analysis using the Corsi statistic. There’s just more data there.”
For Wendorf, though, PDO helps reign in the outliers.
“I don’t think all the reason people perform above or below average is due to luck,” he said.
“I think that shooting talent is involved,” Wendorf said. “I think that goaltending talent is definitely involved. That being said, I think we believe that there’s a larger range of talent in the NHL than there really is, and luck plays a bigger factor than we’re willing to acknowledge. PDO helps us keep that under control.”