Wednesday, November 02, 2005

 

Yipeeee!

The Oiler are back to .500 after another very solid performance last night.

I've got an impending deadline, so posting will be light from me for a while.

Fortunately, Fenwick will presumably be around to cry that he still can't watch his Flames on tee-vee, and the boys over at Covered in Oil do a terrific job.

Speaking of which, I completely agree with them about Hemsky. Last night he was doing everything right that he had previously done wrong. Most impressive, was that every single time he got the puck he attacked the net. On the powerplay it was amazing - if nobody came to cover him he would shoot and get a good rebound or score, and if someone came to cover him he dumped it off perfectly for someone else. It was textbook.

If he (and the rest of the team) can keep it up, this tough extended road trip might prove to be better than expected.

Comments:

".500" doesn't mean what it used to when a team can have 7 losses and only 6 wins and be ".500"
 


What, you mean like back in 19-whatever when there was no such thing as a tie game? Or pre-1992 before the OTL point was introduced?

I think the only logical way to calculate winning % is out of potential points available.

Oil have 13 points out of a possible 26. That's .500. It still isn't a playoff spot, but it is a nice benchmark to clear nonetheless

Some other ramblings on the topics of ties, points, and winning percentages are here (and associated comments.
 


A point for losing in overtime really cheapens the concept of ".500".

It ensures that more than half of the league will be ".500".

As I see it, a .500 tema has the same number of wins as they do losses.
 


Lets say the NHL had a hypothetical point scheme where you got 10 points for a loss, 11 points for an overtime loss and 12 points for a win.

Effectively its the same as the current scheme except you get 10 points for each game played.

Now if a team loses all their games they still get 10 out of a possible 12 points in each game making them an .833 team. Everyone would be guaranteed to be well above ".500".

Wouldn't it be far more meaningful to look at teams that have the same number of wins as they do losses then to look at getting half of the potential points available? The point scheme can be manipulated so that the majority (or all) of the teams in the league are ".500". It becomes meaningless in these scenarios.

I argue that the current NHL point scheme is one such scenario - just less manipulated then the hypotehtical one I gave.
 


What are the odds that hemsky is going to be one of those guys that everyone thinks is going to break out one day and be consistent, but never really does?
 


Listen, I'm on your side (see the link above). The only useful measure is the final column: total points.

But I think talking in terms of winning percentage in terms of points available (under the currently slightly skewed system) is still useful as a relative benchmark - if only to gives us Oilers fans something to cheer about.
 


No no no. Look. Obviously everybody concedes that the teams are actually going to be slotted into the playoffs according to their total-points rank. "Percentage of possible points earned" is just a proxy for that in the end, since everybody's going to end up playing 82 games. But since we're unfamiliar with the point total OR the p.p.p.e. that will be required to make the playoffs, we have no useful norms. P.p.p.e. is the opposite of a "useful benchmark."

Whereas, for thumbnail in-season judgments, the won-loss percentage calculated treating OTLs as outright losses IS useful. It's useful because (a) by this measure the whole league always remains at .500 and (b) about half the teams make the playoffs, meaning that (c) .500 means you're probably right on the bubble. That gives us an extremely useful, slightly rough way to discuss where the team's at. There's no sense incorporating OTLs as a special category in the formula, because we don't have any full-season norms for those either, and because their effect is marginal (but, by that token, significant for marginal teams).

I'm making all this sound a lot more complicated than it is. What I'm saying is, just go with the measure that adds up to .500. We all know what .500 means, and what it implies informally. (Treating OTLs as outright losses, by the way, also enables truly useful binomial calculations like the ones I had a stab at in a previous thread.)
 


To put it another way: it's irresponsible and dumb to put something in the form of a winning percentage, and call it a "winning percentage", when the league mean isn't going to be .500.
 


Whoa, slow your engine big-rig. You're as bad as Fenwick about this thing.

As I've mentioned in the past, I could give a rat's ass about winning percentage as a tool for looking at relative standings.

What I do care about is the metaphysical importance of getting to .500. Virtually every player/coach/gm/water boy for the Oilers mentioned it before last game. Every newspaper, every fan. Was last night's game just another game? Hell no! It was the game that would put us back to .500.

The fact is, precious few of these people care if .500 ain't what it used to be in terms of how well it predicts the final standings. They are only interested in .500 for the simple reason that it is better to take more points than you give up - which is all p.p.p.e is.

Whether it is good enough to make the playoffs is, at the moment, entirely beside the point.
 


And since when did "winning" become defined as: "beating exactly half the other teams"?
 


So you don't care about whether the team is playing at a playoff-qualifying pace. Fair enough--that strikes me as a solitary and eccentric position for a fan, but I suppose it's justifiable.

But if you don't care about projecting final standings, why is it so important to count an OTL as half a win? You may have noticed that overtime defeats feel a lot like losses, that players regard them and react to them as losses, and that they count as losses for the purpose of talking about streaks. Only for the projective purposes you're disavowing could they possibly be regarded as anything but a loss.

As to your question, it should be phrased "When did 'playing .500' become defined as beating exactly half of the other teams?" So did you need some help with that one?
 


Bah. Winning percentage, no matter how you calculate it, is a suckers game for predicting anything with any degree of accuracy.

If you really want to know how your team stacks up and you don't want to just look at the number of points they have, the magic number is the only way to go.
 


[twirls finger, rolls eyes, makes cuckoo-clock noise]
 

Post a Comment

<< Home

This page is powered by Blogger. Isn't yours?