/cdn.vox-cdn.com/uploads/chorus_image/image/24189827/135928244.0.jpg)
The last time we decided to take a look at the Pittsburgh Penguins 2013-14 Hockey Analysis Rating Offense, and the time prior to that we took a look at Point Shares. I had intended to make the next one a follow up looking at Hockey Analysis Rating Defense, but due to the holiday HockeyAnalysis has not been updated since the weekend and the two most recent games are missing from their totals. I then decided perhaps I will look as Goals vs Threshold, but Hockey Prospectus is also missing the most recent game. I then considered perhaps look at Corsi via a bubble chart using data from BehindTheNet, but they were missing the four most recent games. So I will put those ideas on the back burner and we will come back to them in the future.
You may be familiar with Fenwick, which is Shots on Goal plus Missed Shots, and Corsi, which is Shots on Goal plus Missed Shots plus Shots Blocked. They are usually expressed as a differential, Fenwick/Corsi For minus Fenwick/Corsi Against, an on-ice team statistic similar to Plus/Minus except that it is based on shot attempts rather than goals. And occasionally they are then converted to Fenwick/Corsi Relative, which takes on-ice Fenwick/Corsi minus off-ice Fenwick/Corsi, which gives us a very rudimentary example of how the player performs in comparison to his teammates. But that's not the only fancy stats that can be derived through the use of possession metrics, as we saw back when we looked at Hockey Analysis Rating, so let's see what else we can do with it.
Fenwick For Percent
Rather than looking at Fenwick as a differential we can instead look at it as a percent difference, Fenwick For divided by Fenwick For minus Fenwick Against. This isn't that much different than the traditional differential Fenwick, with stats above 50% being the equivalent of a positive possession while stats below 50% are the equivalent of negative. We will be looking at 5-on-5 Close data, situations when both team has 5 skaters on the ice, i.e. not Power Play or empty net situations, and the score is within 1 goal during the 1st or 2nd period or tied during the 3rd period.
On the Penguins the highest FF% players are also those with the smallest sample sizes, Ebbett then Gibbons and D`Agostini. Next is Dupuis, Crosby, Jokinen, Kunitz, and Vitale. Following them is Adams, Neal, Bennett, Malkin, Kobasew, and Megna. Sill broke an even 50%, with the remaining players all falling below that. That leaves us with Glass, and Sutter, followed Jeffrey and Conner, and lastly Zolnierczyk.
On D it is first worth noting that studies suggest that most Defnesemen, and even to a certain extent defensive Forwards, have very little influence on driving possession and in general are deriving their on-ice statistics from whichever forwards they happen to be on the ice with. However, we still look at the numbers because no matter how small a part they play they still have some impact on what happens around them. As it was up front, the highest D is one who is benefiting from small sample size, with Despres leading the pack. Next up is Letang followed by Bortuzzo, then Niskanen, Scuderi, Maatta, and Martin. Orpik is a bit further back, but still well over the halfway mark, while Engelland falls in below 50%.
Corsi For Percent
Just as we did above, except basing the numbers on Corsi rather than Fenwick, and just like above we see some players with a small sample size being uncharacteristically high. Ebbett leads the way followed by D`Agostini. After them is Bennett, Dupuis and Crosby, Gibbons and Kunitz, and then Jokinen. Then we have Neal followed by Malkin and Vitale. The rest fall below 50%. Kobasew, Megna, and Conner followed by Adams and then Sutter. Lastly is Zolnierczyk and Jeffrey followed by Glass and Sill.
On D we have limited sample size Despres leading the way again followed by Letang then Niskanen, Scuderi, and Maatta. After that is Bortuzzo and Martin then Orpik. The lone D falling below the 50% mark is Engelland.
Fenwick Shooting Percentage
This one is an on-ice statistic that measures goals scored per shots attempted. I like it better than standard Sh% because rather than looking at just Shots on Goal it holds the player accountable for chances that completely miss the net as well. Corsi Sh% is another option, but the shooter doesn't really have any control over whether or not a shot gets blocked so the player shouldn't be penalized when a shot that may have otherwise been on net gets blocked by an opposing defender. This is not to be confused with Shot Percent which only looks at the player's individual goals and shot attempts, Sh% is based on attempts by any player on the ice for his team.
We again see some numbers that seem as if they wouldn't be as high it it weren't for small sample sizes. Ebbett leads the way, followed by Conner and Bennett, and then Jeffrey and Kobasew. Following them is Sutter, Kunitz and Dupuis, and then Crosby. Then we have Malkin, Glass, and Gibbons followed by Jokinen, Vitale, Adams, and Megna. Bringing up the rear is D`Agostini and then Zolnierczyk and Sill who have yet to be on the ice for a single goal scored.
On D the sample size issue once again rears its head, as we again see Despres leading the way. Following him is Scuderi and Niskanen, then Orpik, Engelland, and Bortuzzo, with Maatta and Martin coming in behind them. Letang is way down the bottom in no-man's land by himself, the only players on the team that are lower are the two Forwards that were not on the ice for a single goal scored.
Corsi Save Percentage
As we did above we are taking one of the more commonly seen stats, on-ice Sv%, and turning it on its head by looking at it in a new light. For a player defending the opponents' ability to score goals blocked shots now become an effective defensive strategy, and unlike the Sh% the defenders can actually actively influence the opponent's ability to score goals by getting in the way of the puck. Missed shots there is some impact from defense, getting in the shooting lanes or disrupting the shot can cause it to go wide, but the impact is very minimal. However, since Corsi is a conveniently existing stat that includes blocked shots, so we look at the number of goals against in relations to the number of shot attempts allowed.
The top of the list is once again heavily influenced by sample size as we see a number of players that were not on the ice for a single goal against. Kobasew, Ebbett, Sill, Zolnierczyk, Gibbons, and Megna with a perfect 1.000. Next we have Adams, followed by Glass and Sutter, followed by Vitale and Dupuis, and then Kunitz. A little further down we have Conner and Jeffrey, followed by Crosby, and then Bennett, Jokinen, and D`Agostini. At the far end we have Malkin and then Neal.
On D we again have samm sample size issues evident, as Despres was not on the ice for a single goal against and has a perfect 1.000. Then we have Scuderi and Niskanen, followed by Letang and Maatta, then Bortuzzo, and afterwards Martin and Orpik. Bringing up the rear a good ways behind the rest of the D is Engelland.
PDO
PDO is referred to as the "dumb luck" stat. There is a tendency to refer to "puck luck" or lucky bounces, and this is what PDO attempts to examine. Now it doesn't mean that there is a mystical force at work, some fate which actively influences the way the puck, but just that there are certain aspects of the game that influence play in small ways that we cannot account for, and for lack of a better term we refer to those collective unquantifiable aspects as "luck."
The quick and dirty average is to assume that players should have an average PDO of 1000 and that if a player is above 1000 they are performing above their ability, i.e. "lucky" or on a hot streak, and that we could expect to see them fall back to earth and "regress to the mean" as the season continues. Conversely, a PDO below 1000 suggests a player that is in a slump or on a cold streak and we could expect to see him improve and bounce back as the season progresses. Of course this isn't always true, some players, Crosby being a notable example, consistently have a career average PDO well above 1000, or others may have an average that is consistently below 1000 and should be viewed as such when considered a "regression to the mean." Of course sometimes a player may have a "hot season" or a "cold season" in which their PDO remains above or below their average for the entire year and the "regression to the mean" doesn't actually kick in until the next season.
Conveniently PDO does tend to point out players that may have been better than expected because of sample size effects. The top forward, as we saw time and time again with the other stats, is Ebbett. Next we have Kobasew, then Gibbons, followed by Conner, Megna, and Jeffrey. Those are players that we would have expected the stats to be higher than we would expect because of the small sample size influences. After them we have Sutter, Dupuis, Glass, Kunitz and Bennett, and lastly Adams. Kunitz and Dupuis, and oddly enough Conner, despite being above 1000, have averaged well above 1000 for the past few years so they are actually performing at the level we expect to see.
Sill and Zolnierczyk fall in at an even 1000 having been on the ice for no goals, neither goals for nor goals against. Below 1000 we next have Vitale and Crosby, followed by Malkin, then Jokinen, and lastly D`Agostini and Neal. Those are all players that we can look at as having struggled to regain their usual level of performance this season and we should expect to see them bounce back and improve as the season continues. Although Vitale, despite being below 1000, has consistently finished below 1000 every season so he is actually performing as expected for his average.
On D our sample size issue rears its head one final time as we see Despres far ahead of everybody else, although both Scuderi and Niskanen are well above 1000. They are all performing significantly above their ability so we may expect to see them crash as the season progresses. The rest of the D all fall below 1000. Orpik followed by Bortuzzo and Maatta, then Martin, with Letang and Engelland bringing up the rear. So most of our D have been struggling and we expect to see them to improve throughout the year. While Martin and Engelland are still below average, it is worth noting that both have averaged well below 1000 during the past 3 seasons, so while they should improve they have been doing much better relative to their average performance than the others have been.