Power Ranking

[EvanZ]

One thing we should do is make sure we are using the same lines. I went back and updated the line to be the closing line, which brought me down to 8-3 and the RMSE to 7.96. I'm listing the games, spreads, and predictions here so you can compare (and use the same lines):

Code: Select all

HOME	AWAY	LINE	PRED	ATS	CORRECT
LAC	DAL	3.5	-1.68	LAC	1.00
ATL	PHX	-2.0	-6.40	ATL	1.00
NYK	PHX	-8.0	-4.08	PHX	1.00
MIN	DET	-8.5	-5.04	DET	1.00
NJN	GSW	-3.0	-0.11	GSW	0.00
SAC	IND	5.5	3.38	SAC	1.00
BOS	TOR	-9.0	-7.41	TOR	0.00
NOH	MEM	4.0	2.85	NOH	0.00
PHI	DEN	-6.0	-4.91	DEN	1.00
ORL	SAS	-5.0	-3.93	SAS	1.00
WAS	OKC	10.5	9.55	WAS	1.00

This "Ensemble" is a simple average of my own ridge regression, LV full season ratings, LV prior week ratings, Jerry's calculated RAPM (column 2), and the ratings from the Gothic Ginobili blog. Interestingly, the LV full season ratings was the best individual predictor with a RMSE of 7.83 and getting 10/11 correct.

[mystic]

One thing we should do is make sure we are using the same lines. I went back and updated the line to be the closing line, which brought me down to 8-3 and the RMSE to 7.96. I'm listing the games, spreads, and predictions here so you can compare (and use the same lines):

http://scores.goldsheet.com/merge/tsnfo ... slist.aspx

I use these lines, I guess these are the closing lines.

Changing the small differences between yours and mine is not changing the 7 wins ATS for me.

This "Ensemble" is a simple average of my own ridge regression, LV full season ratings, LV prior week ratings, Jerry's calculated RAPM (column 2), and the ratings from the Gothic Ginobili blog. Interestingly, the LV full season ratings was the best individual predictor with a RMSE of 7.83 and getting 10/11 correct.

What is LV? Did you test that "simple average" for more than just the last 11 games?

[EvanZ]

What is LV? Did you test that "simple average" for more than just the last 11 games?

Las Vegas (regression on spreads for full season or prior week). I've been testing it here and there, but not systematically. Too much work for me right now to write the proper code to do this. I really should start keeping track of the totals like you are. At any rate, I just wanted to have some point of comparison. I imagine that last night's set of games was just an extreme result, and it will regress to the mean of 50% or so. I really don't think Vegas can beat Vegas systematically. It just wouldn't make much sense.

[bbstats]

So power is at 99, Vegas is at 101...

anybody have Against The Spread handy?

[EvanZ]

3/3 last night.

[mystic]

3 ATS? I was 0 of 3 ATS.

[EvanZ]

Yes, 3 ATS. Well, I'm due for a bad run here.

[J.E.]

No metric has the Clippers above average, they're dead last in DRTG. Opponent teams scored 120 points per 100 possessions when Paul and Griffin were playing. They allow the second best eFG% and the second most opponent OReb%. I have a feeling that their low post players suffer from the Javalee McGee syndrome: Not giving an eff about basketball fundamentals because you can jumper higher than everyone else

Here's a great piece on the subject by NBA.com's John Schuhmann
http://hangtime.blogs.nba.com/2012/01/2 ... ef:nbahpt1
He's saying that if DeAndre Jordan wouldn't go for that many blocks the Clippers might actually benefit by grabbing more defensive rebounds.

(@mods: should we make a thread to discuss internet articles?)

[J.E.]

HCA for the running season is at an average of 4.8 btw ...

[Crow]

(@mods: should we make a thread to discuss internet articles?)

Whatever you want to do is fine IMO.

In most cases a thread will die within a few days and rarely wil it last more than a week. In 2-3 weeks it will usually be off the front page and without that visibility it is pretty unlikely to return, though there are a few exceptions.

I am not that interested in thread organization personally at this point.

The track record of repeat use of threads for discussion on the same topic is not that strong.

If past threads were sorted by a handful or two of broad topic areas, they might be easier to find, read or re-read and update and maybe that would make re-use of existing threads more common. Probably only the administrator could do that. It could be done just going forward if folks wanted to have multiple sub-forums like apbr.org has but I assume some prefer just to have one forum to check instead of many even though the time difference to do so would be pretty minimal.

[EvanZ]

7-3 with one push ATS last night.

[boooeee]

Another site's take on the Vegas Power Rating: http://bettingmarketanalytics.blogspot. ... -2012.html

(Originally created for the NFL at AdvancedNFLStats and then adapted for the NBA).

Thanks for sharing this here. This week's version is up (I plan on updating weekly). http://bettingmarketanalytics.blogspot. ... -2012.html

I've been testing the rankings against the Saturday spreads, and most games look reasonable, but the Knicks-Nuggets spread is puzzling. According to my rankings, the Knicks should be favored by a point at home versus the Nuggets (they're 2.5 points "worse" than the Nuggets according to the rankings, but after you factor in 3.5 points for home court, you get a -1 point spread). Instead, the Nuggets were favored by 2.5 points. Usually, when the rankings miss that much (3.5 points) it's due to a high-profile injury. But as far as I can tell, there were no injuries or roster changes worth noting for either the Nuggets or the Knicks.

At the time the rankings were created, it didn't have the Friday results in which the Knicks lost by 14 points against the Bucks, after being favored by 6.5 points (a 20.5 miss by the market). So, it's possible that that was factored into the spread.

[EvanZ]

I think the final line on that game was +2. A more puzzling line to me was HOU/SAS. I had that game even according to my full season Vegas ratings, but HOU was favored by 4.5. The Spurs are 1-6 on the road now, so maybe Vegas is rapidly adjusting their spreads taking that into account.

[mystic]

I've been testing the rankings against the Saturday spreads, and most games look reasonable, but the Knicks-Nuggets spread is puzzling. According to my rankings, the Knicks should be favored by a point at home versus the Nuggets (they're 2.5 points "worse" than the Nuggets according to the rankings, but after you factor in 3.5 points for home court, you get a -1 point spread). Instead, the Nuggets were favored by 2.5 points. Usually, when the rankings miss that much (3.5 points) it's due to a high-profile injury. But as far as I can tell, there were no injuries or roster changes worth noting for either the Nuggets or the Knicks.

I would say it is an issue with the your ranking. Ridge Regression has the Nuggets being much better than the Knicks, my ranking has them also being much better. In fact I won ATS, because I expected the Nuggets to win between 5 and 9.6.

Not quite sure how your ranking is created, but maybe you are overrating some things here.



Anyway, last night was 8-2 ATS, but the two previous nights it was 3-14 ATS. Overall now 91-68 ATS and two pushes.

Evan, you might have found something, which is a good betting tool. Maybe it is weaker as saying us something about the real strength of the teams, but in essence your algorithm seems to predict pretty well how much Vegas is off.

[J.E.]

Memphis is now ranked very high(7th) in my rankings. Marc Gasol and Conley are 1st and 4th in vanilla RAPM. Boston seems to be in trouble, although RAPM derived and Vegas' ratings still expect them to bounce back somewhat. The Knicks continue to look dysfunctional and Philly has cooled off somewhat. Most other teams are right where people expected them to be

I'm currently working on ridge regressed team ratings with higher weight for more recent games, but I need some more time