The need for a Player Ranking System

Hockey is, maybe more than any other sport, a team sport. A hitter in baseball can be evaluated based on a series of one on one battles with pitchers which are largely independent of the ability of the hitters teammates. In hockey this is not the case. Jaromir Jagr may be a great goal scorer but if he didn’t have quality teammates around him the number of goals he scores would be significantly impacted. A great defensive play by Marek Malik causing a turnover in the defensive zone followed by a great breakout pass might be just as important in scoring a goal as Jaromir Jagr being in the right spot at the right time to chip in a rebound. Conversely, a good forcheck by Peter Forsberg might be just as important for keeping the puck out of his teams net as a poke check by Antero Niittymaki taking away a shot and a scoring opportunity. As some say, the best defense is a good offense. But there is no stat that “adequately” rewards either Malik’s great defensive play resulting in an offensive opportunity nor Forsberg’s great offensive play limiting the number of offensive opportunities against.

I say “adequately” because +/- attempts to do that and though it has some use as a statistic it is seriously flawed because it doesn’t isolate a players worth independent from his teammates. Conceptually it is a good statistic. Winning games is all about scoring more goals than you give up and +/- measures how many goals are scored by your team while you are on the ice in relation to how many goals are scored by the opposition while you are on the ice. Problem is, if you play on a really good team you are likely to have a decent +/- regardless of how good or bad you actually are. Conversely, if you play on a really bad team you are likely to have a poor +/- regardless of how good or bad you really are. Was Vaclav Varada at +21 really significantly better than Sidney Crosby at -1? No, of course not. There is a reason why Varada is no longer in the NHL.

So the question is, how best can we isolate a players individual ability in such a team oriented sport like hockey. Is it even possible? I believe that this can be done and I believe that I have developed an algorithm that does so adequately. Tomorrow I will describe in more detail what I have done and present some player rankings from last season and Thursday I’ll post this seasons player rankings. I am sure the results with be both surprising and controversial but most of all I hope that are thought provoking and make you think about certain players a little differently.

This article has 8 Comments

  1. You’ve been talking about this for a while now, I’m curious how you’ve did this (I hope you can explain in detail). Anyone who finds out new ways to use hockey data is doing me a favor, because I don’t need to figure it out myself then.

    And I’m curious what would be “controversial”

  2. By controversial I mean some people will get all up in arms and critical of the idea because they find their favourite player ranked much lower than expected or shocked that someone is ranked higher than would be expected.

  3. Umm… just for the record… Varada wasn’t a +21… he had 21 points and was a +2. He’d never been more than a +12 in his entire career. That has to be one of the most inane and poorly researched comparisons in the history of “hockey analysis”. If you can find ONE solid example that looks remotely as bad as the one you just made, I’d be shocked… which frankly blows my mind that you even posted it.

    The only way a player makes it to +21 is by producing points or hardly ever giving them up. It’s a solid indication of either even strength offensive production or excellent defensive play. To argue otherwise is frankly insane.

    Sidney Crosby was a -1 for a VERY good reason. He wasn’t all that good defensively. It’s the same reason good players like Mats Sundin, will come in with a +1 on the season… lots of points but not the most consistent defensively. Alternatively very good defenders won’t finish with high +/- numbers because they don’t produce enough.

    If you compare the top forwards in the NHL in +/- for last season you won’t find a single one that had a +15 or better with less than 38 points (Alexei Ponikarovsky being the only one with less than 40). If you want to know why Ponikarovsky was +15 and Sidney Crosby was -1 all you have to examine is their defensive skills. And as for your inane good teams vs. bad teams argument: COLBY ARMSTRONG PLAYED ON THE SAME TEAM AS CROSBY FOR ONLY 47 GAMES… and he was a +15. I’m sorry but your point is amazingly moot. Chris Clark managed to be +9 on a defensively horrible Washington Capitals team last season, with only 39 points. The more you examine poor defensive teams +/- numbers, the more it’s obvious the players at the low end played poor defensively. Mark Recchi didn’t get to be -28 with the Pens last year by being a great defender. You don’t end up a -28 with 57 points by being good in your own end. Obviously when 30 of your points are on the power play, and 27 are at even strength or short handed, that means you were out there when the other team scored 55 goals, either at even strength or when you were on the power play. Pittsburgh only gave up 300 all year last season, 113 of which were scored while they were short handed. That only leaves 187. If you’re on the ice for 55 of 187 goals against, or 30%, you’re probably part of the problem. Recchi averaged 20:24 of ice time a game last year, 17:36 of which was at even strength or on the power play. So he’s basically out there a third of the time, and on the ice for a third of the goals against on the worst defensive team in the league… Personally I think that likely means he was pretty bad in his own end.


  4. Oh and in case you were curious about how Recchi did AFTER he went to Carolina – who were obviously a much “better” team: He went from a -28 with Pittsburgh, to -36 at the end of the year. That’s right, after the trade deadline he was a -8, with another 7 points tacked on to his 57 for a finishing 64.

    Doug Weight, another late season addition to the same Stanley Cup winners went from being a -11 with 44 points n 47 games with St. Louis to an exemplary -6 with 13 points in 23 games with Carolina. Obviously those “good teams” don’t seem to make much difference for those players being dragged down by “bad teams” do they.

    Seriously I could go on all day with this one… Varada a +21??? scary.

  5. Sorry, my mistake on Varada. Lets take Bryan Smolinski at +8 then.

    Is Smolinski a better even strength player than Sidney Crosby? No, I doubt it. But Smolinski had the benefit of some pretty good linemates which boosted both his point totals and his +/-. The point I am trying to make is that with +/- you can’t compare players on different teams because the influence their teammates have on +/-. Colby Armstrong’s +15 looks pretty impressive on the Penguins but on the Senators +15 is only good for an unspectacular 14th best.

    All the stuff you discuss in your post is all true and that is the reason why I want to create a single ranking system so we don’t have to try to put Chris Clarks good +9 into context.

    As for Recchi and Weight, one has to consider that they both went from playing on the top line on their original teams to 2nd or 3rd line status on Carolina. Are you able to tell me whether they were playing with better or worse players with Carolina than with St. Louis and Pittsburgh.

  6. Ok, lets look at Smolinski vs. Crosby then.

    Last season Smolinski was a +8 with 48 points. 16 of those came on the power play. So he had 32 points at 5 on 5 or short handed. At a +8 that means he was only on the ice for 24 goals against even strength or on the PP.

    Crosby on the other hand was a -1 with 102 points. 47 of those came on the power play. That means 55 of his points came 5 on 5 or short handed. At -1 that means he was on the ice for 56 goals against at even strength or on the PP.

    Considering both of them were hardly on the ice for any penalty killing time, at around 35 seconds a game, I would say the difference in their defensive play would necessitate the preference for Smolinski over Crosby. Obviously most teams would prefer to have Crosby, but that likely has a lot more to do with the fact that he puts butts in seats, has HUGE potential, and will be winning Stanley Cups long after Smolinski has retired.

    Oh and just to tack on a bit here, Smolinski was a +22 the year before last in Ottawa with 11 of his 46 points coming on the power play. That means 35 goals with him out there, but only 13 against when he was on the ice… considering your points about Sean Donovan and Martin Gelinas that would make him pretty valuable in your new system wouldn’t it??

    I still don’t know how valuable this new rating is.

  7. Maybe an easier way to discern whether or not it’s the team affecting the player or the player affecting the team would be to take the team’s overall +/- and compare the player’s number to that number, ranking them based on the difference between their contribution and those of the rest of the league.

    I’m guessing Colby Armstrong would do pretty well.

  8. Firstly, if your trying to tell me Smoke was only on the ice for 32 goals the entire season even strength or short handed your clearly misguided. Yes he got 32 even or shorthanded points which are clearly pluses, but to assume he wasn’t on the ice when the sens scored and he didn’t get a point is crazy.
    There is no way you can figure out how many goals for and against he was on the ice for by those stats. You pretty much have to look at every scoresheet for every game and add it up.

Comments are closed.