Quote:
Originally Posted by Flames Draft Watcher
Stats cannot teach you how to build a winning team. Burke has a lot of experience building teams. I'd take him over a stats guru every day of the week.
|
This is from today. They tend to be right more often than not.
http://www.sbnation.com/nhl/2014/1/2...wAAAAAAAAwmAAA=
It's really not that complicated. You win hockey games by scoring more goals than the other team. You can accomplish this in one of three ways:
- Score lots of goals;
- Don't let the other team score much;
- Both of the above.
Hockey players, like teams, are good at 1, 2 or 3. But all of these correlate to puck possession. Possession leads to scoring chances, and having more scoring chances lead to scoring more goals than the opposition. This is irrefutable:
It's also super obvious. Seriously, does anyone actually think the statement, "a team that generally has the puck more than its opposition tends to win most of its games" is surprising?
Now to this point,
Quote:
|
its being able to know which statistics are relevant statistics and combine that with a knowledge of what you're applying those statistics to.
|
Correct. This isn't to say that possession stats are the be all end all of hockey. You also need guys who can finish, i.e. a decent shooting percentage (see NJD, who have goot possession numbers but don't have this). You need good goaltending (see TOR, who are only afloat right now because of team save %). And you need good special teams (see WSH who are actually getting outscored at evens but have a 25% PP).
New Jersey is the perfect example of this - they're a top 10 possession team quite comfortably, sporting a 52.8% Fenwick close. They're an average PP team and a very good PK team. However, they're last in the NHL in save percentage and 27th in shooting percentage at even strength with the score close.* So, they tend to generate more chances than the opposition's, but their chances turn into goals at a lower than average rate, while the opponents' chances turn into goals at a higher than average rate. Given that, having the puck more isn't enough to result in outscoring their opponents. In contrast, Toronto is 29th in the league in Fenwick close, but their sv% is top 5, and their PP is 6th, so they have a couple of things keeping them semi-afloat still. So, if you have these other things going for you, you can overcome bad possession numbers to some extent, and if you don't have them, then being a good possession team isn't enough.
But generally, goaltending and shooting percentage fall within a range of normalcy, and if you're outside that range, you can usually predict a regression to the norm. For example, New Jersey will probably not finish the year with a team save %age as low as .899 at even strength, while Toronto's .939 will probably drop a bit. Possession doesn't tend to vary as much. Possession also measures even strength play, which is when most goals are scored - while it's possible to have a team be handily beaten at even strength and have their special teams bail them out, over a longer stretch of games, usually the team that's better at evens will end up winning out.
Really, when you think about it, "advanced" stats aren't advanced. They tend to measure things that are pretty commonsense. Identify the purpose of the game: score more goals than the bad guys. Then consider what things lead to that: having the puck more than the other team (especially when the score is close), having good goaltending, and having good special teams. Teams that do this win a lot. None of this should surprise anyone.
*usually the "score close" situation is the most useful one, because teams who are trailing by more than a goal tend to outshoot the opposition while trying to catchup, while the team with the 5-1 lead or whatever tends to sit on it. See http://nhlnumbers.com/2013/12/5/score-effects-and-you