View Single Post
11-18-2012, 11:15 AM
eklunds source
Registered User
Join Date: Jul 2008
Location: Ed Snider's basement
Posts: 8,325
vCash: 500
Originally Posted by Legionnaire11 View Post
I know all the "stat guys" will spit on it all day and derive complex formulas to measure players defensively. But I guarantee that if +/- didn't exist, there would be a plethora of "stat guys" who would create it. It's a simple measure that fans can quickly understand and apply to what they've seen on the ice.

Is it flawless, and is it safe to use it as a reliable gauge in every instance? of course not, but used responsibly and understood for what it is, +/- is a fantastic stat.
Originally Posted by wickedwitch View Post
+/-, like most other stats, has value if you know how to use it correctly. The difference is that it's really easy to use it incorrectly.
It's not though. Look at Filip Kuba.

Year+/-Shots For @ 5v5Shots Against @ 5v5RatioTeam SH%Team SV%
Both years the Senators outshot (albeit barely) their opponents. Both years he was used in a similar fashion - started roughly the same amount of shifts in the offensive zone as the defensive zone, played against roughly the same level of competition. Yet his +/- went from 886th of 891 players (6th worst in the league) to 9th of 894.

In back to back seasons with a similar performance he went from 6th worst to 9th best in the entire league. Nine hundred players. Solely because of team shooting percentage and team save percentage.

Those numbers are all at 5v5. So, his team shot 6.5% at 5v5 when he was on the ice in 2010-2011... and next season his team scored on 10.7% of their shots. Team shooting percentage is not something players can control. Another thing players can't control is how their goalie plays behind them... In 2010-2011 he basically received sub-AHL goaltending - 88.3% sv% at 5v5 is truly awful... The next season he had almost Vezina-calibre goaltending behind him. 92.6% is pretty darn good - Rask had a 0.929 at 5v5, Lundqvist and Quick were at 0.933, and Rinne was 0.928.

That's a pretty extreme example - but based on nothing more than fickle percentages that Kuba cannot control, he went from -26 to +26.

+/- in the traditional sense has zero useful information in it. None. Find me a player where it provided some amount of meaning and I can explain to you why it didn't... Humans look for patterns that don't exist, it's our nature.

As for "I guarantee that if +/- didn't exist, there would be a plethora of "stat guys" who would create it"... There is a +/- stat that 'stat guys' created. It's called Corsi, and instead of measuring goals, it measures shot attempts. There are only a couple hundred goals for a team each season, but there are thousands of shot attempts. More data = less statistical noise. It's been proven that Corsi and Fenwick (very closely related), do a much better job of predicting future success. They do a better job of predicting team scoring chances (based on people who watched every hockey game, recorded every scoring chance and who was on the ice for/against), a better job of predicting goals for/against, and a better job of predicting which team wins and which team loses.

It's not always presented as +/- because the numbers are so large.. it's usually presented as a ratio. It doesn't matter if you help your team create 1,000 shots per game if they're giving up 950 shots per game, versus creating 30 shots per game and giving up 28.5 - the ratio is what is important.

Using Fenwick, I placed a pretty significant bet on the L.A. Kings winning the cup, despite the fact they were an 8th seed. These numbers work. +/- doesn't.

eklunds source is offline   Reply With Quote