I've been dubbed a "stathead" around the ol' sphere a lot recently. Some use the term descriptively, others as a pejorative. I would like to say for the record, however, that I don't find all stats particularly powerful or meaningful. In fact, I downright hate some of them.
Stats I dislike:
Goalie wins.
My new pet peeve. I've said this before and I'm sure I'll have to say it again: goalies don't win games. Teams win games. True, netminders perform a vital function and a good goalie can improve a club's chances of winning, but, in the end, someone has to score goals in order for a team to win the game. And goalies don't score. They don't draw penalties. They don't do a lot of the things that lead to goals for or against.
It's a good rule of thumb that a goaltender with a lot of wins is probably pretty good. All other things being equal, that's true. Problem is, things aren't always equal. Sometimes a great goalie plays for a lousy team. Sometimes a lousy goalie plays behind a great team. In those cases, wins are an extremely poor barometer of the goalie's efficacy. In fact, in both cases the win stat is probably grossly misrepresenting the guy in question.
In the end, SV% is far more relevant stat when assessing goaltenders, because the rate at which a goalie stops the puck is the only contribution he makes to a team's wins/losses. Anyone claiming that "wins" is a more important stat than "SV%" is inverting cause and effect.
Plus/Minus
I actually don't dislike plus/minus as much as most people these days, but I can still admit that it's a flawed stat - mainly because it's sullied by such things as short handed and empty net goals. Parsing situations and counting events makes plus/minus more powerful, as does considering it in context of a team effects (such as quality of competition, quality of opposition and ES SV%). Naked plus/minus absent of these qualifiers can be fairly misleading, however.
"RTSS" stats.
Take-aways, give-aways and hits are three of the most useless categories in the league, because no one seems to agree on what any of these things is. Different arenas count each thing differently and I have yet to see any rational criteria used to judge each of them. The numbers are almost totally useless therefore.
All that said, there are some stats I'd like to see:
Puck battle differential.
I'd really like to see +/- rate for winning/losing puck battles. I think it's a skill that is almost essential in the NHL and it would be interesting to see how a puck battle rates would correlate with wins/GD etc. In addition, it strikes me as something that is potentially trackable, as long as the defintion of "puck battle" and what constitutes a "win" or "loss" is properly defined and observed.
PP/PK temporal efficiency.
Right now, PP and PK rates are based on "opportunities" and "times shorthanded". Clearly, however, not all instances of special teams are created equal: some last 5 minutes. Other last 10 seconds. For example, if Detroit scores a goal during a 5 minute major, they are "1 for 1". If, on the other hand, the Flames have a 10 second PP due to over-lapping minors and they fail to convert, they're "0 for 1". That's fairly ridiculous. A PP scoring rate (or PK success rate) in terms of goals/time unit would probably be a lot more revealing and powerful, especially if it was parsed by situation (one man advantage, two man advantage).
