View Single Post
04-03-2012, 01:26 PM
#84
Lafleurs Guy
Moderator

Join Date: Jul 2007
Posts: 28,425
vCash: 500
Quote:
 Originally Posted by MathMan On "point totals matter": This very much reminds me of the baseball arguments that "RBIs matter". We also see "wins matter" both for goalies and pitchers (I don't really follow baseball except in the most cursory manner, but I do remember the ruckus about Hernandez's Cy Young and his 13-12 record, even in this day and age!) This is a pet peeve of mine actually: people dimissing "microstats" by bringing up "macrostats" such as point totals. Often, this goes so far as arguing statistics don't matter... by bringing up different statistics! We're no longer discussing "some things aren't measurable" here. We're making the implicit assumption that points total is a superior measure than others, purely because it is traditional, because it has been introduced earlier. That's not a good enough reason to favor it. It's not that point totals don't matter, it's that the amount of information they contain is suprisingly low, despite often being the only metric players are judged by, consciously or subconsciously. Less informative metrics do not matter even if they have been commonly used in the past (and still are) and if there is some sentimental attachment to them.
The starting point for the SABR people is that hits don't win games... runs do. It matters how many times you score.

In hockey it's the same principle. Scoring is all that matters. Points in hockey aren't the equivalent to RBIs in baseball. Points would be the equavalent to RUNs which is what you're trying to look for when you build a baseball team... how good is the player at generating runs?

I don't think anyone is dismissing SABR in hockey here but you seem to be dismissing totals and I think that's a problem. We've seen Crosby's CORSI lower than Gomez and Tanguay's and yet he's ten times the players they are. How do we know this? Because he scores points (the objective of hockey is to score more than your opponents) at a much higher rate.

Those microstats (at least the CORSI ones you've shown me and where players rank) do not do a good job of predicting how effective a scorer a player will be. In fact, I've seen the opposite from Gomez. And that's problematic because if it's not effective in projecting at least the offensive effectiveness of a player, then their value is greatly diminished. And I know you point to luck, but no matter how you slice it Gomez is not Crosby and yet his CORSI numbers were actually better.
Quote:
 Originally Posted by MathMan On counting discrepancies: there are ways to compensate for counting biases. However, virtually nobody seriously uses takeaways, giveaways, and hits however, which have issues that go way beyond counting biases.
Statiticians can do what they can but I think it's an uphill battle on some of this stuff. And if they're not tracking giveaways/takeaways aren't they missing a big part of the picture? I mean it's kind of a damned if you do, damned if you don't proposition there. Either they're doing it but subjectively or they're not doing it at all and missing out on vital information.
Quote:
 Originally Posted by MathMan On color-blindness: a color-blind person could easily use a digital camera and computer software to detect and measure color. We use instruments to get around the limitations of our senses all the time. The inability to directly perceive something does not mean it is not otherwise perceptible or measurable, even if it is only by its effects.
A colour blind person wouldn't know to look for red to begin with. He'd have to be told by somebody else that 'red' actually existed in the first place.

Moreover, you STILL won't know what it looks like digital camera or not. And couldn't it be argued here that the 'digital camera' here would be looking outside of analytics at the intangibles? Isn't that the only way to see red here?
Quote:
 Originally Posted by MathMan Of course, if you ask someone who hates red "is there too much red in this picture?", he's liable to say "yes" to any amount. So it is with stuff like the work ethic of disliked players, or more formally, on the appearance of work ethic. I personally don't care one whit if a million guys tell me the Designated Scapegoat doesn't work hard enough. He's the Designated Scapegoat, he'll never work hard enough -- this has been true for Gomez, Cammalleri, Ryder, Kostitsyn, Kovalev, etc, etc. IMO, such opinions are invariably mostly observer biases and as a result hold no value whatsoever.
Maybe he's the designated scapegoat for good reason. Maybe he doesn't hustle to get back into plays or just flat out doesn't care if he helps the team win. Maybe he can't score a goal to save his life. Maybe he's out there smoking weed before games and partying in Vegas like Giambi did. THAT matters doesn't it?
Quote:
 Originally Posted by MathMan The aesthetics of a player, team, or style cannot be measured and are a matter of opinion, true. Analytics are only concerned with processes and results. The appearance of hard work is an aesthetic. I recognize that this is controversial, possibly because traditional hockey values are very blue collar and place a high value on dogged hard work while skill is viewed with some suspiscion.
Getting hit by Chara is worse than getting hit by Desharnais. We may not be able to measure it but we know it because God has given us the gift of common sense.

As for the 'hard work' argument being controversial... pure numbers were put into practice by Billy Beane and it was a howling failure. You have to take this stuff into consideration when building your team or you're going to flame out big time.
Quote:
 Originally Posted by MathMan Aesthetics are personal, and I'm not going to argue those. If anyone wants to argue on that, their opinion is their opinion and I wouldn't expect analytics to convince them otherwise. But me, I don't care how the player looks. What I want to know is if he is effective at helping my team win hockey games.
It's not about 'looks' though. Nobody has brought up 'looks'... But there are things that can't be measured and they seem to show up in the win column. If you want to win games, maybe you should open your mind to what's out there outside the spreadsheet.