PDO: What Is It and Why Does It Matter?

One factor of the game of hockey that is mentioned sometimes but isn't talked about a lot is the role of luck and variance. Watch any game and you will probably see a player get a perfect shot or scoring chance away only to have it stopped by the goaltender or go off the goalpost. Sometimes a player will miss the net completely despite being in perfect position. Things like this happen fairly often and when they do, you will probably hear a fan or commentator say say "that's just unlucky" or something similar shortly after. Most people talk about how these bad bounces happen from an individual standpoint, but what about over the course of an entire season?

Goals, points and plus/minus are the usual metrics used to judge a player's performance, but if you were to take a look at a certain player's scoring stats and plus/minus over the course of a few seasons, you will probably notice that they tend to bounce around a lot. This is mostly because goals for and against are a result of factors that might be out of a player's control and goal scoring patterns usually come in streaks. I mentioned earlier how the most a player can do is get into a good position to create a scoring chance and the rest is up to the goaltender. The shot might also go off the goal-post or miss the net completely, we saw plenty of that in game 4 of the Predators-Coyotes series last night.

An example within the Hurricanes is the season Jussi Jokinen had in 2009-10 when he scored 30 goals. Expecting this kind of production from him every season isn't fair or reasonable because he shot at 18.8% that year, which is a pretty high shooting percentage, especially compared to his career average of 13.2%. He did spent most of that year playing as a winger on a line with Eric Staal, so that partially explains his high goal total but luck also played a huge role because a shooting percentage that high is going to come down to Earth no matter how good you are. On the flipside, Jokinen shot at only 10.2% this year which is pretty low for him and he had only 12 goals a result, his lowest total in awhile. Keeping in mind that Jokinen was one of Carolina's better possession players at even strength, it is fair to say that he got a bit unlucky this season and is better than his goal total indicates.

A similar concept applies with plus/minus. This stat is usually determined to figure out who is good and bad defensively when it really isn't the best way to go. Remember, every skater that was on ice for a goal is given a plus or minus, even if they weren't involved with the play at all. Let's pretend that a defenseman is on ice for two goals against and both of them were incredibly soft goals that the goaltender usually stops. That defenseman would be charged with a -2 and will have a "bad game" in most people's viewpoints because of that but in reality, there was nothing he could do about either goal that was scored. This is why I usually look at shot metrics before goals for/against stats to judge a player's overall performance. There's just too much luck/variance involved in these stats.

One way to see how big of a role luck had in a player's season is to look at a stat called PDO. This sounds like some big, complex stat with a complicated formula but it is only the sum of a player's on-ice shooting percentage and on-ice save percentage at even strength. The average is considered 1, 100 or 1,000 depending on what numbers you use and a player's PDO is expected to regress back to that point no matter how high or low it is. PDO usually has the biggest impact on a player's plus/minus rating more than anything since it looks at on-ice shooting and save percentages but it can be look at on a team level to help explain certain hot and cold streaks.

For instance, the Los Angeles Kings are surprising a lot of people this post-season with how they eliminated the Canucks in five games and are now one game away from eliminating the St. Louis Blues as an eight seed, but they have actually been a good team all year, just unlucky. Sometimes legitimately bad teams maintain a low PDO all season but the Kings have been controlling shots and scoring chances at even strength all season, they've just fell victim to bad shooting luck for most of the season. The addition of Jeff Carter and simple regression to the mean has played a role in them turning things around at the best time possible.

Now that I've gotten the theory and statistical mumbo jumbo out of the way, let's take a look at how luck has effected some players on the Hurricanes this season. Something we'll do after the jump.

As a team, the Hurricanes shot at 7.8% and had a save percentage of .918, which shows that they got a bit unlucky but they weren't exactly controlling shots or scoring chances, so luck isn't the only thing to blame for their problems this season. That being said, variance played a big role with some players this year and that's easy to see when you look at the table below.

Forwards

Sh% = Shooting percentage, Career Sh% = Career shooting percentage, On-ice Sh% = On-ice shooting percentage at even strength, On-ice Sv% = On-ice save percentage at even strength, PDO = Sum of on-ice shooting and save percentages

Eric Staal had some very poor shooting luck for most of the season but he still finished the year with a respectable goal and point total. 9% is around the league average for shooting percentage but he's obviously a better finisher than that and could have had more than 24 goals if the bounces went his way. However, on-ice save percentage is where Staal got really unlucky this year. Carolina's goalies were stopping only 89.7% of opposing team's shots when he was on the ice, which is the lowest among forwards. This helps explains his awful plus/minus rating. Chad LaRose and Jeff Skinner's plus/minus ratings also took a hit due to poor goaltending whenever they were on the ice.

Speaking of of LaRose, he had his highest shooting percentage in years this year, but the rest of the team wasn't scoring much at even strength when he was on the ice. Players like LaRose are interesting because their shooting percentage is never very high and they always have a pretty low goal total as a result. This usually indicates that they aren't as good of finishers as other players. Patrick Dwyer and Andreas Nodl also fit in this category. Although, there isn't much LaRose could have done about that awful on-ice save percentage or his -15 rating.

One player who some fans have said has gotten unlucky this year is Andreas Nodl and that is very true when looking at his shooting percentage, but that isn't too far off from what he has shot at over his career. Nodl actually got somewhat fortunate at the other end with the Carolina goalies stopping most of the shots they faced at even strength when he was on the ice. You will probably notice that the shutdown line of him, Dwyer & Sutter have very high on-ice save percentages which could indicate that the goalies could have bailed them out more than a few times. I am open to the idea that players can influence team save percentage but numbers that high are going to come down eventually.

As for players who did get lucky, look no further than Anthony Stewart and Tim Brent. Yes, Brent's career shooting percentage is pretty high but he scored on almost 17% of the 71 shots he took this season. That means he scored once almost every six shot attempts. I have a tough time believing that will sustain in the long-run. Stewart also had a career high in shooting percentage and the Hurricanes scored on almost 10% of their even strength shots when Stewart was playing with them. Tuomo Ruutu also had a very high shooting percentage in both areas this year, but he's actually been a great finisher throughout his career and shot only a percent less than his career average.

Defensemen

Jaro Spacek was the fortunate one on the blue-line. He scored on almost 12% of the shots he took, the goalies were stopping pucks at a rate that was way above average and the team was shooting at a pretty normal rate when he was on the ice. He played well as a third pairing defenseman but luck helped him out a little bit.

Similar to the shutdown line, Tim Gleason and Bryan Allen have pretty high on-ice save percentages. Like I said earlier, I am open to the idea that they could have played a role in it but luck was obviously a factor, too. Gleason actually had a pretty high PDO for a tough-minute defenseman and I'm not too certain that it will carry over into next year. I expect him to play well but his plus/minus rating might take a tumble if the goalies aren't stopping 93% of the shots they face.

The most unlucky defensemen were Joni Pitkanen and Derek Joslin. Both had abysmal on-ice save percentages and the Canes couldn't score much at all when they were on the ice. Their personal shooting percentages were higher than their career averages but all of their apparent problems in their own zone might not be all their fault. I actually took a closer look at Joslin's season and his play as a defenseman was not terrible. It was when he was moved to forward that the problems started. Shows that he could be worth keeping around as a #7/spare defenseman for depth purposes.

Jay Harrison and Justin Faulk got unlucky in the shooting department as the team was shooting at less than 7% with these two on the ice, so I think that has played a role in their bad plus/minus rating. The fact that both had a normal on-ice save percentage concerns me a little, though.

To sum things up, PDO is one way to predict future performance but it does need context. Knowing what kind of situations a player is being used in and what kind of player he is helps determine if his good or bad year is due to what kind of luck he had. You never know which way the bounces will go in hockey, so looking at stats like this are a helpful way to explain a hot or cold stretch that a player might be going through.