Espn rankings

[LooseCannon]

He's trying to come up with a quick and easy to calculate method that will identify which teams might be better or worse than their win-loss record indicates.

You are what your record says you are.  Has Hollinger's rankings ever done a good job of predicting when seeming bad teams would start winning more or seemingly good teams would fall on hard times?

I think Hollinger has used his stats in the past to suggest that the Mavs are not as good as their regular season wins.

It's actually been demonstrated that point differential is a better predictor of future performance than win-loss record, and that seems to be true across many sports.  Basically, close games end up a crap shoot and there's a lot of variance in teams winning those types of games.  If you root for a team that makes the post-season on the strength of winning more close games than average, you're usually in for a less successful following season, as your team regresses toward the mean and is more likely to have average luck in close games.  People who believe in the mythology think that there is some sort of clutch ability, but a lot of times, it's just luck.  But good teams are involved in more non-close games and win a higher percentage of such games. 

Hollinger's formula identifies underachieving teams which are winning those sorts of non-close games but running into bad luck in close games, as well as overachieving teams which are winning a disproportionate number of close games. 

The value of quick and easy is that it is quick and easy.

[Fan from VT]

This is interesting. We'll do a little experiment just to see.

Here is the current order based on win-loss:


New Orleans 8-0
San Antonio 8-1
Boston 8-2
LA Lakers 8-2
Dallas 6-2
Orlando 7-3
Atlanta 7-4
Chicago 5-3
Miami 6-4
Golden State 6-4
Denver 5-4
Phoenix 5-4
Oklahoma City 5-4
Portland 6-5
Milwaukee 5-5
Indiana 4-4
Cleveland 4-5
Detroit 4-6
Memphis   4-6
New Jersey 3-6
Sacramento 3-6
Houston 3-6
Charlotte 3-7
New York 3-7
Washington 2-6
Minnesota 3-8
Philadelphia 2-8
Toronto 2-8
LA Clippers 1-9

Here is the Hollinger Power Ranking:
New Orleans
Miami
Boston
LA Lakers
Denver
San Antonio
Phoenix
Milwaukee
Chicago
Orlando
Dallas
Houston
Utah
Portland
Memphis
Atlanta
Indiana
Charlotte
Philadelphia
Golden State
Toronto
Oklahoma City
Detroit
LA Clippers
New York
New Jersey
Cleveland
Sacramento
Minnesota
Washington



So here is what I will do. Since real wins matter most in terms of making actually making the playoffs and winning the championship, we need to compare predictions to actual standings. So, in about a month, I will re-look at the Real Standings, and see which current rank order more closely fits the actual win-losses at that time.

Now, Hollinger's rankings should get much more accurate as games are played, so at that approximate 1 month mark, I will also re-rank the teams based on Hollinger and Win-Loss for the next month, and see which matches the win-loss more closely.

Now clearly this is not perfect, because if a team has already won several unearned "fluky" wins, they are going to be further up in the total standings than they "should" be. So, at the 1 month or so mark, I will also rank the teams in order of how they did in the month period from now to then, to really get at the heart of whether win-loss or hollinger ranking better predicts real win-losses going forward.

And, of course, we will do something for the playoffs.

[soap07]

but the idea that stats are a better predictor of greatness than winning has become way too common.

Straw man's argument. Please find me one - just one - analyst that thinks that. I might add - "wins" are a stat in itself.

[MBunge]

I think Hollinger has used his stats in the past to suggest that the Mavs are not as good as their regular season wins.

He also predicted it would be Utah vs. Boston in the 2008 NBA Finals.

Mike

[MBunge]

but the idea that stats are a better predictor of greatness than winning has become way too common.

Straw man's argument. Please find me one - just one - analyst that thinks that. I might add - "wins" are a stat in itself.

Uh, have you seen where the 6-4 Miami Heat are ranked?  If that's not valuing stats over won-loss, what is?

Mike

[Fafnir]

I think Hollinger has used his stats in the past to suggest that the Mavs are not as good as their regular season wins.

He also predicted it would be Utah vs. Boston in the 2008 NBA Finals.

Mike

And this means what? People get post season predictions wrong all the time.

Take a look at his predictions over the past three years. He actually picked the Lakers over the Jazz when they matched up that year:
http://sports.espn.go.com/nba/playoffs2008/news/story?page=Smackdown08#predictions

[Fafnir]

but the idea that stats are a better predictor of greatness than winning has become way too common.

Straw man's argument. Please find me one - just one - analyst that thinks that. I might add - "wins" are a stat in itself.

Uh, have you seen where the 6-4 Miami Heat are ranked?  If that's not valuing stats over won-loss, what is?

Mike

If you just want W-L records you can just use the standings.

[LooseCannon]

So here is what I will do. Since real wins matter most in terms of making actually making the playoffs and winning the championship, we need to compare predictions to actual standings. So, in about a month, I will re-look at the Real Standings, and see which current rank order more closely fits the actual win-losses at that time.

The correct test would be to take the power rankings in the first half of the season and compare that to win-loss record in the second half the season.  Contrast that with first half win-loss record.

[Fan from VT]

So here is what I will do. Since real wins matter most in terms of making actually making the playoffs and winning the championship, we need to compare predictions to actual standings. So, in about a month, I will re-look at the Real Standings, and see which current rank order more closely fits the actual win-losses at that time.

The correct test would be to take the power rankings in the first half of the season and compare that to win-loss record in the second half the season.  Contrast that with first half win-loss record.

i think i covered that. I'm going to compare both lists to the total win-loss after a month as well as new win-loss in that timespan, since fluky results now will still count in total win-loss in a month.

[Fafnir]

So here is what I will do. Since real wins matter most in terms of making actually making the playoffs and winning the championship, we need to compare predictions to actual standings. So, in about a month, I will re-look at the Real Standings, and see which current rank order more closely fits the actual win-losses at that time.

The correct test would be to take the power rankings in the first half of the season and compare that to win-loss record in the second half the season.  Contrast that with first half win-loss record.

i think i covered that. I'm going to compare both lists to the total win-loss after a month as well as new win-loss in that timespan, since fluky results now will still count in total win-loss in a month.

Month my month snaps shots could be interesting, but a month is how many games, 10? If you're going to compare overall bigger splits than that are probably necessary. Especially given how tough certain stretches of scheduling can be.

[MBunge]

I think Hollinger has used his stats in the past to suggest that the Mavs are not as good as their regular season wins.

He also predicted it would be Utah vs. Boston in the 2008 NBA Finals.

Mike

And this means what? People get post season predictions wrong all the time.

Take a look at his predictions over the past three years. He actually picked the Lakers over the Jazz when they matched up that year:
http://sports.espn.go.com/nba/playoffs2008/news/story?page=Smackdown08#predictions/

What it means is that if Hollinger's system was WHAT HE AND OTHERS CLAIM IT TO BE, his predictions should be markedly better than others'.

And I would say that him picking the Lakers over the Celtics in the 08 Finals is among many examples of that not being the case.

Mike

[Fafnir]

*shrug* I give up, heaven forbid anyone make a wrong prediction.

Apparently it means they can no longer be taken seriously. Just look at win loss records and be content that no one in the world knows basketball, because everyone makes incorrect predictions.

It is just another way at looking at performance in the NBA. Whenever people get upset about it I think about the Cetlics in 08-09 when they were 27-2, but had played pretty poorly over the last 5 or 6 games of that torrid stretch.

[Fan from VT]

here's an interesting tidbit:

2008 is the only year I could find stat head playoff predictions vs. espn scout expert predictions, scored on accuracy of picking series winners with bonus points for correctly picking # of games per series.

That year the experts scored:

Ford 68
Sheridan 63
Legler 62
Adande 59
Broussard 57
Henry Abbott 55
David Thorpe 53
Marc Stein 51
Jalen Rose 50

The statheads went like this:
Kubatko 73
kurylo 72
Ilardi 71
Ma 65
Hollinger 61
Pelton 54

So by and large, the staties destroyed the scouties. Hollinger, despite being a bad stattie that year, would have been the #4 expert.

Interestingly, Hollinger got better in '09 and '10:
Berri 75
Hollinger 68
Ma 67
Pelton 67
Payne 66
Ilardi 60
Weil 49

'10
Ma 71
Pelton 70
Hollinger 66
Berri 57
Ilardi 54
Paine 52
Voulgaris 40

Now, unfortunately, they stopped scoring the scouties in '09-'10. But this is all on TrueHoop, who's author, Abbot, observes that the statties pretty much kill the scouties like him in overall predictions in the playoffs, at least.

Anyway, this is presented for what it's worth: entertainment.

[MBunge]

*shrug* I give up, heaven forbid anyone make a wrong prediction.

Apparently it means they can no longer be taken seriously. Just look at win loss records and be content that no one in the world knows basketball, because everyone makes incorrect predictions.

It is just another way at looking at performance in the NBA. Whenever people get upset about it I think about the Cetlics in 08-09 when they were 27-2, but had played pretty poorly over the last 5 or 6 games of that torrid stretch.

The whole freakin' point of Hollinger's stats is that they give him a better understanding of the game and what's going to happen when two teams play each other.  THAT'S THE WHOLE FREAKIN' POINT.  If they don't do that (and I don't see any evidence they do), they're not worth anything more than killing time in BS internet discussions.  Which is perfectly fine.

Mike

[Fan from VT]

*shrug* I give up, heaven forbid anyone make a wrong prediction.

Apparently it means they can no longer be taken seriously. Just look at win loss records and be content that no one in the world knows basketball, because everyone makes incorrect predictions.

It is just another way at looking at performance in the NBA. Whenever people get upset about it I think about the Cetlics in 08-09 when they were 27-2, but had played pretty poorly over the last 5 or 6 games of that torrid stretch.

Yeah, it's weird. People act as if  if someone were to lose their entire lifesavings because they bet it all on one hollinger prediction, it means hollinger is the idiot. Ummm...no. There is an idiot, but it's not him.

His stats would be "correct" and "have validity" if, instead of betting 100% of savings on one game (which no stat prediction model would ever suggest doing), you bet .1% of your savings on hundreds of games vs. the spread throughout the season, always following some statistical rule. If that rule could basically guarantee you a positive % return on your investment compared to just picking the team with the better record or going with your gut (two methods that, vs. the spread, end up giving about a zero net gain/loss), then it's statistically valid.

Now, I've never used hollinger to gamble vs. the spread, but I have seen very good evidence of this applied in football betting vs. the spread, so the theory holds.