In the late 90ís, the CBC shot a documentary entitled The New Ice Age. The documentary had a great scene where Don Meehan was negotiating a new contract for Tie Domi. He asks Bill Watters how he’d replace the PIM that he might lose if Domi were to leave, slaps him around for a while and then gets Domi an absurd contract. Domi giggles, asks whether the contract is payable in US dollars and accepts the terms.
If I was sitting in Watters’ chair, I would have asked Meehan how useful those PIM were to the Leafs. I’d love to hear Meehan’s answer to that question. It’s a tough question to answer but someone suggesting that PIM would need to be replaced should the player in question be lost ought to be able to answer it.
Part of the problem with the statement Meehan made is that PIM are not created equally. Some make a team shorthanded while others do not. There’s argument about whether or not certain penalties (such as fights) can give a team a boost or send a message that their stars are not to be touched which can be beneficial down the road. Some are justified as being useful in the service of the greater good (beating the holy hell out of Sean Avery).
Objectively, penalties matter because they impact the number of points a team can expect to get out of the game. At the start of each game, both teams will have an expected number of points. When Edmonton plays Pittsburgh later today, my best guess is that the Oilers expected number of points is something like 1.38 points while the Penguins can expect something 0.76 points.
Of course, hockey has only three possible point outcomes for teams-2, 1 and 0. As the game progresses and events occur, a team’s expected number of points from the game changes as well. Every event in a hockey game, no matter how minute, changes the points that a team can expect from a game. The most obvious of these events is goals, but this idea touches on everything that occurs including the passage of time, icing, offsides, players getting injured and penalties.
Penalties change the expected point outcome because they alter the probabilities of goals occurring. The penalized team sees their expected goals for drop and goals against rise. The unpenalized team sees their expected goals for rise and goals against drop.
The impact that a particular penalty has on the expected points that a team can expect to earn from the game depends on a number of factors including but not limited to the time of the game, the score and the manner in which the penalty was taken. The last one is essentially impossible to quantify in my opinion-a penalty could actually net out to a positive change in expected points in certain circumstances.
Consider the following hypothetical: Jarome Iginla and Daymond Langkow are on a two on one in the final minute of a game that the Oilers are leading 2-1. Igor Ulanov is playing to force Langkow to shoot as opposed to passing it to Iginla while Cory Cross is in the hottest pursuit that he can put together. At that point in the game, the Oilers expected points from the game would be very near 2 while that of the Flames would be near 0. It’s rare that a defensive team like Calgary would manage to get the goal that they need to extend the game. The 2 on 1 would probably raise it higher than usual (as would the identity of the participants) but it still wouldn’t be very high.
At the last moment, Ulanov sprawls, taking away the pass and forcing Langkow to shoot. Characteristically, MacT has chosen to start Ty Conklin, who equally characteristically misses the puck. It hits the post and goes across the net to Iginla. Cross, who has caught up with the play (it’s a hypothetical) is faced with a choice-allow Iginla to shoot it or tackle him. Obviously, tackling him is the smart play to make and the play that has the best outcome in terms of expected points. Although the Oilers will be penalized, the Flames crappy PP and Oilers solid PK reduces the chance that the Flames will score. A goal will cause the Oilers expected points to drop dramatically. While he can’t know if the Flames will score on the PP, all of the known information suggests that he should take the penalty.
The best way to evaluate the cost players impose on their teams for the penalties that they take would be through an analysis of how it changed the expected points that a team could expect to earn from that game. While this is beyond me at the moment, I’ve assembled a list that splits a players penalties into coincidental and non-coincidental penalties. For some reason, the NHL does not make this available. It is, I think, a fascinating collection of information.
Contrast a player like Alexei Zhitnik with someone like Barrett Jackman. They both have 72 PIM this season but while Jackman has piled up 30 of those PIM in coincidental penalties all of Zhitnik’s 72 PIM were non-coincidental. Without getting into the specific changes in expected points, it seems obvious that Zhitnik’s penalties hurt more than did those taken by Jackman, just by the nature of the fact that so many more of them put the Islanders a man down.
What was the cost of the penalties taken by Zhitnik? At the very least, we can take a stab at estimating it in terms of goals. The Islanders kill penalties at the rate of 80.4%. Zhitnik has been responsible for 36 opportunities against. That comes out to a goal differential of about -7.5. The number that can be directly attributed to Zhitnik’s penalties would be reduced by a number of factors though-the Islanders would be expected to give up a certain amount of goals in the time that they spent on the PK if they were at ES and they’re expected to score a certain amount of goals while shorthanded. The flipside of that is that he’d also be pulling down the amount of goals that the Islanders would be expected to have scored if they were at ES. There’s also the possibility that the other team will take a penalty while on the PP (and other factors which I’m sure I’m forgetting). I’ll make an examination such as this the subject of a future post to get a better idea of the actual cost imposed by players such as this. If we take -7.5 goals for someone like Zhitnik as a starting point though, it’s a significant cost. As a rough figure, I look to a swing of about 6 goal differential as being the price of an extra win. If Zhitnik stays at his current pace, his penalties could cost the Isles as many as two wins.
The impetus for setting up the spreadsheet to track this was a comment I made over at BOA about PIM. I mentioned that I remembered looking at the subject of who takes penalties that put the team a man down and who takes penalties that lead to another guy going to the box. My recollection was that Horcoff and Moreau were in the former category and Smith in the latter. This has proven to be correct-Smith has only 18 PIM this year that have put the Oilers a man down while Horcoff and Moreau have 42 and 38 respectively. As it stands, Smith actually looks to be a pretty unique player. While he’s got a pretty high PIM total, the fact that so many of them are coincidental puts him in a group with very few guys who play regular minutes. It’s frequently been said that he was undervalued-while I don’t know about this year, it seems likely to me that he’s probably had less respect than he should have in the past for being a pretty solid defenceman while staying out of the penalty box.
My next post on this will involve taking a look at the different types of penalties that players take. I’ll be interested to see where players like Hemsky and Moreau rate in terms of things like high sticks and roughing, which I’d have to imagine rarely fall into the class of penalties with positive net outcomes in terms of winning percentage.
Oh, and as for Domi…he doesn’t take a lot of penalties that put the team down, at least when looked at as a counting stat as opposed to a rate stat. Given that he plays a reasonably regular shift, I’d guess that he probably doesn’t put up a high number of penalties that hurt the team while still putting up a number that are more likely to have the subjectively highly valued benefits like inspiring the team by having his face beaten in by Chris McGrattan.
The stats are available here. NC is non-coincidental while C is coincidetal and the numbers are penalties in minutes.