APBRmetrics

EvanZ wrote:I got a scraper from someone for that site. I can ask him if it's ok for me to share it with you.

That would be great.

For those of you that are interested, I've just posted my first stab at vegas rankings for ncaa basketball (see link in my signature). I've got Ohio State at number 1 (consistent with kenpom.com).

Regarding Franchise-by-Franchise HCA. I wonder if opponent's travelling times/distances could be accounted for, along the lines of http://weaksideawareness.wordpress.com/ ... -a-season/

Philly, Boston and New York are geographically very close so it doesn't surprise me that (assuming a large number of intra-division games) they're bunched together down the bottom. That logic doesn't seem to work for the Californian teams though...

Pretty happy with these at the moment...award 3 points for home court...and anywhere from 1-2 for a back-to-back advantage (rested vs. tired). Key players may be worth 1-2 points for a subtraction if they're out with an injury.

+7: Miami, Chicago
+5: Oklahoma City
+4: Denver
+3: Portland, Philadelphia, LA Clippers, Atlanta
(note here that we may be seeing something where Philly and Atlanta are treated as +3 vs. weak teams, but only +2 vs. quality...Philly was a two-point home dog to both Miami and Chicago, suggesting the market saw those powers as five points better on a neutral court)
+2: San Antonio, LA Lakers, Dallas
+1: Boston, Indiana, Memphis
0: Orlando, Houston, Minnesota (Orlando may go back up with a rest break)
-1: Utah, New York, Milwaukee
-4: Golden State, Phoenix
-5: Cleveland, Toronto
-6: New Jersey, Washington, New Orleans
-8: Sacramento, Detroit
-10: Charlotte

Some teams seem to get penalized more for b2b's than others because of lack of depth...or a poor track record in earlier b2b's.

Again, these are estimations of the market ratings rather than personal ratings...

SRS has 12 teams over +3, you have 7. Evan has 8. I didn't check others but would be interested in knowing if someone else wants to.

"...award 3 points for home court..."

Do bettors deduct 3 pts from an average power ranking for the other team being on the road as well? I assume they should / do but it isn't said.

The market pointspread would be:

*The point difference between teams based on the above estimations
*Adjusted three points to account for home court advantage
*Adjusted for any b2b influence (usually +1 or +2 in favor of the more rested team)
*Adjusted for injuries to stars (usually -1 or -2 for a significant player)

Sometimes the b2b adjustment is even bigger if the tired team is also playing a 4th game in 5 nights, or even something crazy like Orlando was doing recently with 6 games in 8 nights. Both Orlando and Indiana are in a b2b tonight, but the market seems to be saying that Orlando has more sustained fatigue because of their killer recent stretch (Indiana -5.5 tonight rather than -3 or -4).

Again no direct mention of the road team, so is the stated HCA advantage = the amount of performance improvement for the home team when they play at home over their average for all games and the amount of road disadvantage to the road team compared to their overall average and thus you only need one adjustment (rather than two)? I thought that was the case before but then began to wonder and I want to be sure.

I guess the best way to phrase it is that the market considers the sum of the issue to be worth 3 points when adjusting from Power Ratings...whether you're talking about home court advantage, road court disadvantage, or the combination of the two.

If you allotted 3 points from both directions, that's a six point total adjustment...and that wouldn't line up at all with what we're seeing in the market prices. It's differential on the scale...then a 3 point boost for the home team for whatever reasons one deems necessary to create that 3 point value.

Miami shows up as 3 points better than Denver on the scale. Were they to play on a neutral court, the market would make the game Miami -3. If the game is in Miami, the market posts a line of Miami -6. If the game is in Denver, it's pick-em.

Generally speaking:
NFL: 3 points
College Football: 3 points
NBA: 3 points
NCAA: 4 points

The NBA used to be as high as 4.5 or 5 points back in the 1980's when I started tracking it. I'm guessing that's because tempo's were much higher. Might have been other influences too (an emphasis on shooters rather than defenders in a way that caused the scattergram to spread out more or something).

Hope I'm answering your question. The market doesn't really deal with the "nuances" of road court disadvantage and home court advantage in a clear way. Oddsmakers generally use a blanket 3 points from what they tell me...and then sharps fire away at the number. Many sharps are certainly nuanced enough to shade their action to anything that pops up (San Antonio is getting respect at home tonight vs. OKC, though part of that is OKC in a b2b).

The scale I threw together represents neutral court estimates. If a person is trying to estimate a line for a given game, they would add three points in favor of the host to the differential between teams...then make any additional adjustments they deemed appropriate based on their knowledge of the teams and any injuries or schedule dynamics...

Ok, thanks for the replies. I guess I making the question harder than normal.

I tend to trust that the market is rational.

But isn't average margin of victory for NBA games something like 6-7 points? Is that because the average absolute difference in team quality in games is 3-4 + the 3 pt HCA factor equals 6-7 or whatever? This plays into my question, my lack of 100% certainty of understanding.

Computer Sports World used to have a feature called 32A back in the old days that would show you everyone's point differentials at home and on the road...and then sum up the league at the bottom of the page. Not sure if any place has that now...where you can just look it up and say "the average NBA game is an xxx point win by the home team."

Sagarin is showing 3.48 on his regular ratings, 3.10 in ELO Chess, and 3.59 in Predictor, if that helps.

Maybe a poster has that information handy within their own stuff. I would be surprised if the average NBA game this year was a 6-7 point win by the home team given all the bad home teams. More than a third of the league right now has a losing home record.

The logic is getting sticky here.
Perhaps an average road win is around 5 points, and an avg home win is around 8. No win is ever by less than 1, and some are by a lot more than the avg.

Look at the 2005 Finals, Spurs and Pistons going to the final minute of Game 7. Earlier, SA had won an overtime game.
But the first 4 games were won by the home team, by 15, 21, 17, and 31.

Overall, Det outscored SA by an avg of 2.1 ppg; but the avg winning margin was 14.4 pts.

Crow wrote: But isn't average margin of victory for NBA games something like 6-7 points? Is that because the average absolute difference in team quality in games is 3-4 + the 3 pt HCA factor equals 6-7 or whatever? This plays into my question, my lack of 100% certainty of understanding.

The average MOV this season is 11.9, for home wins it is 13.1 and for road wins it is 10.0. Last season it was 10.6, 11.4 and 9.4 respectively.

At the moment I have the HCA at 3.8 instead of usual 3.

Thanks Mystic.

Are you able to see the median performance by any chance? Sharps will tell you that a settled line represents the median of possibilities for that game. Maybe I should have explained it as the market seeing home court advantage being worth a median of 3 moving forward, rather thinking about averages. I just started out with that 32A thing about 30 years ago so I still think of home court as an average as a default. When one is adding 3 points for home court advantage to Power Ratings...it's because 3 is seen as the likely median for home court advantage moving foward. Analysts can then tweak as they see fit if they believe they have meaningful information about certain locales or team tendences.

The mean Vegas line this season is -3.5. The median line is -4.5. The mean absolute spread is 6.3. The median absolute spread is 6.0.

Went through and did it by hand tonight. Median result is home team by 5 through Friday Night's action.