There is a significant amount of disagreement regarding the value of certain statistical metrics in the game of hockey today. Some fans and analysts hold these metrics in very high esteem, using them as the basis for their arguments about teams and individual players. Others suggest stats like Corsi and Fenwick are less telling, are manipulated to help support flawed arguments and are generally overrated.


I’ve found many of the statistics to be helpful in evaluating individual players – especially in my fantasy leagues. When it comes to assigning value to skaters, I believe one statistic stands above the rest. That statistic is PDO.


This stat is simply the sum of shooting percentage plus save percentage. Advanced statistical research suggests shooting percentage and save percentage primarily driven by luck over the long term. Add them together and you have a stat that accurately reflects a player’s luck while he’s on the ice. Generally speaking, the stat is designed to regress toward 1.000. For a more in-depth explanation of PDO, check out this article from Arctic Ice Hockey.


Let’s use this statistic to evaluate one of the game’s most talented skaters and (mostly) reliable fantasy assets: New Jersey Devils winger Ilya Kovalchuk.


It has been a little more than two years since the Devils pulled the trigger on a blockbuster trade to acquire Kovalchuk.  In September of 2010, the Devils made a significant long-term commitment to the then 27-year-old sniper, signing him to a 15-year, $100 million deal. The price to retain Kovalchuk was steep, but he was seen as one of the game’s great offensive talents and a player whose track record suggested he could provide point-per-game production for at least a few more years.


The price of acquiring Kovalchuk’s services in one-year fantasy leagues was also extremely high. Entering the 2010-11 campaign, Kovalchuk was coming off draft boards late in the first round or early in the second round. That came as a surprise to no one, considering Kovalchuk had been a point-per-game performer in four out of the previous five years and had scored 27 points in 27 games with the Devils during the previous season.


However, no one could have predicted what would happen next.


Kovalchuk’s production dropped significantly in his first full season in New Jersey.  He scored just 31 goals and totaled 60 points, his fewest of each since his rookie campaign of 2001-12. To make matters worse, his plus-minus of -26 was the worst of his career. Simply stated, he was a massive disappointment for the Devils, his fans and his fantasy owners.


It is impossible identify just one reason as to why Kovalchuk’s numbers nosedived in his first full season with the Devils.  In fact, there were a number of key factors:


  • A player that averaged 286 shots per season and 3.69 shots per game in his first eight seasons in the NHL, Kovalchuk took just 245 shots in 2010-11, or approximately 3.02 per game.
  • The Devils stumbled out of the gate as a team, leading to poor statistical production from a number of previously-reliable fantasy assets.
  • Teammate and fellow star forward Zach Parise appeared in just 13 games all season due to a serious knee injury.
  • The weight of great statistical expectations following Kovalchuk’s decision to sign a $100 million deal with a perennial playoff contender.


Luck, or a lack thereof, also played a significant role in Kovalchuk’s underwhelming season.


In doing some preseason research for my one-year fantasy drafts, I decided to look up Kovalchuk’s PDO on the hockey site Behind The Net.ca. I discovered he posted a PDO of 1.035 in 2009-10. However, it promptly dropped to .971 in 2010-11, good for second worst on the team among forwards that played at least 50 games.


Kovalchuk’s PDO has rebounded a bit to .983 this season, as has his production (26 goals, 38 assists and 64 points in 61 games) thanks to a little bit more luck and a lot more shots. After averaging just 3.02 shots per game a season ago, the Devils sniper is averaging 4.04 per game in 2011-12.


To get a better grasp on the how luck has affected Kovalchuk’s production over the course of the past few years, let’s take a look at Kovalchuk’s year-by-year PDO in his past four full seasons:























It can safely be stated that Kovalchuk’s fortunes took a significant turn for the worse in 2010-11. Furthermore, his PDO of 1.035 in 2009-10 may help explain his career-best plus-minus of 10 (the first season in which he finished as a plus player in his entire career).


As with Kovalchuk, we can quickly and easily determine who has the highest and lowest PDO among players who have played in a significant number of games this season by looking it up on Behind The Net.


Some fantasy-relevant names appear among the league leaders in PDO (minimum 50 games played):


  • New York’ Derek Stepan (1.055)
  • New York’s Michael Del Zotto (1.044)
  • Calgary’s Olli Jokinen(1.036)
  • Phoenix’s Ray Whitney(1.035)


Some interesting names also appear on the list of NHL players with the lowest PDO this season (minimum 50 games played):


  • New York’s Mark Streit (.956)
  • Montreal’s Tomas Plekanec (.967)
  • Columbus’s Fedor Tyutin (.968)
  • Anaheim’s Cam Fowler (.973)


It is no secret fantasy owners need to rely on a wide variety of information when making roster decisions about their respective squads. Understanding PDO and its ability to suggest if luck has had an effect on a player’s performance can significantly help with individual player evaluation. It’s just one of many factors fantasy hockey enthusiasts need to consider, but few statistics are more reliable and easy to understand.



Write comment
Comments (7)add comment

Mike Schmidt said:

... @ David Goodburn

In addition to such things as health, PP/PK time, linemates, number of shots and quality of competition faced are more or less equal from the previous year, I tend to look at if a player's shooting percentage varies significantly from his career mark. I also look to see if his on-ice save percentage at even strength varies greatly from years past. If a player is placed in the same situation he did just one year ago and isn't getting the results, something is amiss.

A guy like Staal is a great example. If you are confounded by the lack of production and you see his role hasn't changed much if at all, I'd check and see what his PDO looks like.

March 08, 2012
Votes: +0

David Goodburn said:

How do I utilize this stat? Hi Mike,

Great article. I agree with you that luck does play a role in on ice production in terms of incredibly high and unsustainable SH% or SV%(just ask Steve Mason what a little luck can do for a career).

It is no real surprise to me to see 2 NYR at the top of this year's list given Lundqvist's SV% dominance.

Do I have to do a team baseline then compare players on that team to it to determine a one year league luck that maybe I can capitalize on? I am thinking here that Eric Stall would have had a low PDO at the start of this year and since then the "market correction" has normalized his stats a bit. But I would guess certain players Ialways score quite low on this despite elite production. I am thinking here of a guy like Zetterberg who has atrociuos SH% numbers every year so likely comes out under 1.00 on the scale.

I guess my question is what is teh best way to use this in a one year league to take advantage of this stat?

March 08, 2012
Votes: +0

Mike Schmidt said:

... @ Pengwin7 First and foremost, I appreciate the response. You are one of the smartest commenters on the hockey and baseball site.

A few points:

1.I should have been more clear in the article. All of the PDOs are for even-strength. That levels the playing field. You can't compare the PDOs of guys who see a ton of PP time with those who don't. That would skew the numbers considerably.

2. Some players are better shooters than others. This is undeniable. But look at Kovy's shooting percentages in Atlanta. One year it's .120, the next it is .161, the next it is .125, the next it is .184. Two of those years he was a 50-goal scorer. The other two years he was a 40-goal scorer. I'd argue luck plays a significant role in shooting.

3. I do agree with even-strength production/minute as a valuable stat. That's a good call on your part.

4. Hockey is not baseball, for sure. But I think there's some value in this stat. I take it into consideration when evaluating players on my rosters. For the majority of players, the stat is less relevant because their PDOs aren't overwhelmingly high or low. But a 60-point swing for Kovalchuk from one year to the next caught my eye. Again, bad luck wasn't the only factor. Shot quantity played a significant role, among other factors. But I'd argue bad luck did play a role.

Thanks for your feedback!

March 08, 2012
Votes: +0

Pengwin7 said:

Disagree I love, love, love metrics.
And I love the number crunching here. Very solid.

Overall though, I wouldn't point to PDO as an important statistic.
You're overvaluing it, IMO.

1. NHL elite players like Kovalchuk are pure snipers. Watch these players in warm-ups, or the NHL shooting competition. If these guys have >1 second of control time, they'll put a puck on a 8" diameter target. Luck plays a very small part of shooting.

2. An analysis of SH% coupled with a player's linemates/team is sufficient. With Parise & Henrique, Kovalchuk is getting more passes onto his stick = more SOG = more scoring.

3. People always blah-blah-Atlanta-bad-team, but those were some fantastic PP units while Kovy was there. Marc Savard, Kozlov, Hossa/Heatley, Enstrom. Very good players. Kovalchuk typically put up 30PPP as a minimum in Atlanta... which means that a good percentage of his scoring was PP-based, which is higher scoring rate altogether. I'm not sure if PDO is simply even-strength based... but a player's SH% is going (typically) be better on the PP and goalies SV% will be lower. Whether this PDO will also work out to 1.000 for Kovalchuk is important - as he bombs a lot of his shots on the PP. (I didn't see a separation of production for Kovy, ES vs. PP, which would/could have been telling, IMO)

You've got several statements in here that I find to be, well, misleading.
1. Furthermore, his PDO of 1.035 in 2009-10 may help explain his career-best plus-minus of 10 (the first season in which he finished as a plus player in his entire career).
Or... contract year? +/- is all about hockey smarts, focus & effort (Landeskog?). Any player is capable of this, even Kovalchuk, if he knows his next contract depends on it. We've all seen Kovalchuk put effort, and PK, from time-to-time. I'm a big Kovalchuk-hater, mostly because he regresses to lazy hockey without effort to backcheck... which I hate.

2. Understanding PDO and its ability to suggest if luck has had an effect on a player’s performance can significantly help with individual player evaluation. It’s just one of many factors fantasy hockey enthusiasts need to consider, but few statistics are more reliable and easy to understand.
Disagree. If you want remove the words "significantly" and "more reliable" and "easy to understand"... then I'll agree.

3. When it comes to assigning value to skaters, I believe one statistic stands above the rest. That statistic is PDO.
Most valuable fantasy statistic: Even-strength production/minute. If a player can score at even-strength, he'll get more ice-time and more PP-time. Finding these guys is the key to fantasy hockey (keeper) leagues.

ps. All this rant made, I still LOVE, LOVE the work... but hockey ain't baseball.
There ain't no BABIP here - so don't try to force one. smilies/wink.gif

(Overall, though... A+ for the work... just do not agree with your interpretation of its value)
March 08, 2012
Votes: +1

Ari Steiner said:

... This probably varies a lot on a player-by-player basis, and might be more an indicator of style than luck; some players just shoot a lot (Nash/Carter), whereas others (Tanguay, say) wait for quality shots. I'd like to see the stats for the three guys I mentioned, I think they would show that this depends more on a players style than on luck.
March 08, 2012
Votes: +0

Mike Schmidt said:

... You can see the effects in the player's plus-minus, most certainly. A lot of that comes down to luck. If you look at some of the numbers closely, you'll find a big swing between on-ice save percentages for many players from year to year -- even for those players who face the same level of quality of competition.

Phoenix's Ray Whitney is one example. Whitney's plus-minus was 0 last year. His PDO was 1.002 and his on-ice save percentage was .911. This year he's a +22 with a PDO of 1.035 and an on-ice save percentage of .933.
March 08, 2012
Votes: +0

ultrawhiteness said:

advanced stats PDO is pretty great for evaluating how a guy's season is going, but i wonder if the on-ice SV% portion serves any usefulness for fantasy purposes.
March 08, 2012
Votes: +0
You must be logged in to post a comment. Please register if you do not have an account yet.