I might be a supporter of statistical analysis in hockey but one stat that I never use is plus-minus. My opinion is that it isn’t very useful and doesn’t provide much value for evaluating players. The reason I say this is because it’s a stat that’s largely derived on luck and how good a player’s teammates are. Charging a player with a positive or negative rating whenever they are on ice for a goal is basically assuming that they did something to contribute to a goal being scored for or against their team. A player could do absolutely nothing to contribute to the play still have his plus/minus rating affected. His plus/minus rating would also be affected even if the goal scored is a complete fluke and the fault of the goaltender.
Luck and randomness is something that occurs in just about every hockey game and goals for/against is usually where it is seen the most. Thus why plus/minus is a bad stat to use when judging a player. All it really tells you is what kind of teammates he played with and how good or bad the team around him was. If you were to look at the players with the worst plus/minus ratings in a given year, you will probably notice that they are mostly skaters who play a lot of minutes on bad teams. It’s also more likely to be a players who are relied on to play the majority of the tougher minutes, unless it’s Jack Johnson then he’s going to look bad no matter what. Someone who plays on a team with bad goaltending is also subject to have a lower plus/minus rating than others and vice versa. The overall point here is that plus/minus really doesn’t tell you much about how a player performs, which is why I tend to favor shot and possession based metrics instead.
There is definitely something that can be drawn from looking at goals against data, though because not every goal a player is on ice for is the fault of the goalie. This is especially true with the Hurricanes, who were a very weak defensive team this season. Charging every single player with a negative rating because they were on ice for a goal is a bit ridiculous but another option would be to penalize the player who was at fault for the goal. Miscues and defensive breakdowns happen all the time and there are usually a couple players who are at fault for it. There may have been a bad turnover, a defenseman may have forgot to tie up his man in front of the net or a forward may have been caught wandering in the defensive zone and allowed an opposing player to sneak in. Those are usually the kind of mistakes that lead to goals against and they are the fault of more than just a goaltender.
Instead of penalizing every player who was on ice for the goal, a more sensible thing to do would be to charge the player(s) who made the mistake that led to a goal with an "error." David Staples of the Edmonton Journal has done this with the Oilers players in recent years and his general rule of thumb with it is that a player who makes a defensive mistake in a scoring sequence is charged with an error. I did this last year with the Hurricanes defense and will do it again this year, only with every Carolina player instead of just the defense. I’ve also made a few changes from Staples’ method. Instead of using a total error percentage from every goal that a player was on ice for, I separated them by game state.
Why? Because players who kill penalties are going to be on ice for more goals (and possibly more errors) simply because they play more defensive minutes than those who are considered offensive specialists. I also did not count errors from 5-on-6 play because less than 2% of the game is played there and the defending team is already at a big disadvantage there. Another thing that is different from last season is that I was able to look at some more detailed footage of the goals thanks to NHL Game Center. This allowed me to get a better look at where certain plays broke down and where errors occurred as opposed to watching only highlights.
After the jump, we will take a look at which Hurricanes players made the most mistakes this year and how many of them ended up turning into goals for the opposing team.
I uploaded a spreadsheet of the plays I logged as errors here if you want to see them and debate some plays that I charged as errors. What I’m going to do here is display a couple charts showing how many errors a player had during each game state and see what percentage of goals he was on ice for were his “fault.”
The numbers here somewhat coincide with my perceptions of most of these players. At the top of the list you’ll see Patrick Dwyer and Brandon Sutter, who were very good at not committing errors at even strength. These two also were not on ice for that many goals against despite playing against the toughest assignments among the forward corps. Conversely, Andreas Nodl is at the bottom of the list but he had only four errors at even strength and wasn’t on ice for many goals to begin with so that is probably why. Small sample sizes tend to make players look really bad or really good and it’s the former with Nodl this year. The PK stats for Dwyer and Sutter in particular are very interesting. Dwyer was on ice for fewer goals against but he seemed to be responsible for more of the shorthanded goals that he was on ice for. This could possibly say something about Sutter and Dwyer’s penalty killing abilities since they were among the best on the team in terms of shots prevented. Dwyer’s error percentage on the penalty kill quite the contrast from what it is at even strength. This is also the case with Jussi Jokinen as he didn’t commit a lot of errors at even strength but was responsible for over a quarter of the goals he was on ice for on the PK.
Jokinen was actually one of the team’s better players at not making mistakes that led to goals. Both him and Chad LaRose were on ice for a lot of goals at even strength but it appears that they weren’t responsible for most of them. Eric Staal also falls into this category and I think this shows how misleading the +/- stat is for judging defensive play. Yes, he was on ice for a ton of goals but not that many were his fault. Then again, if a player is on ice for over 70 goals and is at fault for half of them then they probably shouldn’t be in the NHL. We know that isn’t the case with Staal and this data confirms it.
Jiri Tlusty and Tuomo Ruutu aren’t so lucky in this regard as they were responsible for at least 20% of the goals they were on ice for, which appears to be pretty high number for Carolina’s forwards. I’ve noticed that Ruutu has been protected by the coaching staff the last couple of years in terms of who he is matched up against and I thought the main reason was due to him playing on Jeff Skinner’s line but it appears that Ruutu may have some bigger defensive issues than some may have thought. His error percentage is a lot higher than most of the other regular forwards, and that could definitely be a problem if the team wants to return him to being a tough-minute player.
How about that? Joni Pitkanen was responsible for the lowest percentage of goals that he was on ice for and this was while playing against some pretty tough competition in the top-4. His underlying numbers were not great this year and he spent most of the season injured, but this does give me some optimism about next year and how he will perform then. If he can handle the toughs then he should be able to make a decent partner for Tim Gleason, who was the team’s best defensive defenseman yet again. Gleason plays the most difficult minutes among Carolina defensemen and was responsible for less than 40% of the goals that he was on ice for. He also did not make that many egregious errors aside from a brutal turnover to Claude Giroux in a game against the Flyers. Most of the errors charged were due to him not being able to prevent a screen in front of the net, which is something that he will hopefully get better at because that was one of the team’s biggest problems for the whole season.
Gleason’s partner, Bryan Allen, was also very good and only slightly worse than Gleason at even strength and on the penalty kill. Allen had similar problems to Gleason but he also seemed to lose his footing in the defensive zone a lot and had a lot of trouble keeping up with top forwards. He’s good enough defensively to play in any teams’ top-four but he certainly has some limitations. I’m interested to see how the Ducks use him for next season. His game is very one-dimensional but I tend to like players like him who can play against tough competition and put up respectable numbers despite that. Allen’s shot-blocking skills also add to his value.
Justin Faulk was pretty solid at even strength in terms of not making that many big mistakes but he struggled big time on the PK. He wasn’t on ice for that many shots against when playing short-handed but he had a lot of errors on the PK that led to goals. I noted earlier that he didn’t allow many shots or scoring chances on the PK and had a low save percentage there, but the high amount of errors he comitted might be the reason for that. Faulk was only 19 for most of season and was playing a big role on the team so things like that are expected. You could say that Faulk was rushed, but he performed very well at even strength in a top-four despite being younger than I am. One might think that most of Faulk’s errors were due to bad pinches since he is more of an offensive talent but that wasn’t the case. He actually struggled the most with positioning and winning puck battles. Both of those mistakes are correctable since he can get stronger and gain an even better sense of the game as he gets older. Same goes for his penalty killing.
The two players who had the tendency to get caught pinching the most were Jamie McBain and Jaroslav Spacek, the team’s third defense pairing for a good part of the year. Both were regularly deployed against weaker competition and judging from this, it looks like McBain fared a lot better than Spacek. McBain was used in a lot of different roles throughout the year and was in over his head in a few instances but he wasn’t responsible for that many of the goals he was on ice for despite that. However, watching some of McBain’s mistakes made me want to break my computer screen. Things like that tend to stick with you when it comes time to evaluating a player and it might be the reason why some people think lower of McBain than I do. I actually think he’s very underrated and can play top-four minutes if given the right partner, and when I say that I mean someone who isn’t Pitkanen.
Jay Harrison was responsible for a little under half of the goals he was on ice for and he was used mostly against second lines with Justin Faulk. He was probably the team’s most improved player this year as he had gone from a third-pairing guy to playing over 20 minutes a night in the span of the year but I still don’t know if he is a legitimate top-four defenseman. His overall numbers from this year show that he performed better when he was used in a more offensive type role but he was able to at least keep his head above water and not be a liability when used against tougher competition. After looking at these numbers, I’m still not sure if he would be a top-four defenseman on a contending team but he’s good enough to at least be in the NHL and play every night. The Canes are getting a ton of value for him for less than $1 mil. per year no matter what.
If you want to talk about players who were liabilities, look no further than Tomas Kaberle and Derek Joslin. Both players were given the softest minutes among the defense corps when they were playing and neither were effective at driving the play forward. Kaberle was barely keeping his head above water and Joslin was lit up like a Christmas tree at even strength. To make things worse, both players were on ice for a lot of goals relative to their ice time and they were at fault for at least half of them. My thought earlier in the year was that Kaberle was getting unlucky (his on-ice save percentage was below .900) but you can see here that he wasn’t helping his cause. I have to say, for a guy who is known for his puck-handling skills, he sure does turn it over a lot and he was next ot useless when it came to winning puck battles and fighting off opposing forwards. Joslin, on the other hand, was just bad in every sense of the word and looked like someone who shouldn’t be playing in the NHL. It’s easy to tell why neither of these two players are on the team anymore.
Overall, I feel like this can be a useful stat but the subjectivity of it is an issue because there are bound to be issues over what should and shouldn’t be counted as an error. If someone else tracked this same data then they would probably come up with different results than me, for instance. It’s also tough to determine what is a good or bad percentage when the data is available for only so many teams. If you look at the Copper and Blue article I linked to earlier, you’ll see that the Oilers had some absurdly high error percentages. That could be due to either scorer bias or the Oilers being an awful team that year. You be the judge. Anyway, I think this stat is a good way to expand hockey analytics but the subjectivity of it might be a problem, along with the fact that not all errors lead to goals so luck plays a role in this.
Either way, doing this project led to some interesting observations about the team’s defensemen, especially McBain. The numbers usually have interesting things to say about him and it seems that the error stat is no different. I’m very interesting to see how he progresses in the next couple of years.