Learning about / from the betting sphere

Home for all your discussion of basketball statistical analysis.
Crow
Posts: 6250
Joined: Thu Apr 14, 2011 11:10 pm

Learning about / from the betting sphere

Post by Crow » Thu Nov 06, 2014 1:00 am

I've never followed the betting sphere closely but have the impression that the books and sharps were aware of the importance of referees, travel, rest, altitude and coach behavior in blowouts before most media and fans and even teams on. What other angles do they study that fan analysts should pay more attention to? Do they look for and coach player conflicts, from the tape, sources, stats or all of the above? Do they do a good job at tracking coaching strategy changes and coaching influence changes? Do some teams play significantly worse on weekends than weekdays compared to average? Are the affects biggest on Fri, Sat, Sun or Mon.? How much does time of season affect lines (early, pre and post all-star, the weird mix that is found in April)? What else? How good have people gotten at predicting match-up pace or four factor matchups and factor interactions? Does referee analysis reach down to player level? Does the analysis of rest, travel, altitude and weekends reach player level? Do the best know how Popovich us going to divide minutes and when he is going to not try hard or give up early?

BigLeagueInsights
Posts: 63
Joined: Thu Oct 23, 2014 9:59 pm
Location: Las Vegas
Contact:

Re: Lessons from the betting sphere

Post by BigLeagueInsights » Thu Nov 06, 2014 7:33 pm

Obviously, the oddsmakers don't want people to know how the lines are made, so we have to guess what's inside the black box algorithms that make the lines.

From my experience, most lines are created by complex algorithms, models developed by statisticians and computer scientists.

My guess is that the models are regression-based and all of the inputs are relatively simple box score stats.

But the most important adjustment the bookies make is based on public betting patterns. The model spits out a predicted total or spread, and the bookies use past public betting numbers to move the line a few points in order to better balance the books.

One of the most interesting aspects of line watching over the past few years is what happens after an injury. Injuries to superstars obviously break models, and it's not clear to me how the bookies handle this. OKC this year is a great example. How do you make a line when your model has very few data points to train it with?

Crow
Posts: 6250
Joined: Thu Apr 14, 2011 11:10 pm

Re: Lessons from the betting sphere

Post by Crow » Tue Dec 02, 2014 7:50 pm

Post on early betting trends. http://bigleagueinsights.com/early-seas ... ng-trends/

Will you be updating the the league stat trends soon?

Crow
Posts: 6250
Joined: Thu Apr 14, 2011 11:10 pm

Re: Lessons from the betting sphere

Post by Crow » Tue Dec 02, 2014 8:03 pm

Anybody on this free "fake" betting site? http://www.fakepuppy.com/GetStarted.aspx I might try it. If others do, we could self report occasionally. I saw vbookie on spurstalk long ago and have wondered if folks here might be interested in that, with its ability to allow custom propositions. RLee and DSMok1 does this version bulletin board allow it and if there were signs of interest from the group, would you consider it, facilitate it? http://www.vbulletin.org/forum/showthread.php?t=94128

BigLeagueInsights
Posts: 63
Joined: Thu Oct 23, 2014 9:59 pm
Location: Las Vegas
Contact:

Re: Lessons from the betting sphere

Post by BigLeagueInsights » Tue Dec 02, 2014 8:16 pm

I should have an updated version this week.
Crow wrote: by Crow » Tue Dec 02, 2014 11:50 am

Post on early betting trends. http://bigleagueinsights.com/early-seas ... ng-trends/

Will you be updating the the league stat trends soon?

sndesai1
Posts: 133
Joined: Fri Mar 08, 2013 10:00 pm

Re: Lessons from the betting sphere

Post by sndesai1 » Tue Dec 02, 2014 8:23 pm

Crow wrote:Anybody on this free "fake" betting site? http://www.fakepuppy.com/GetStarted.aspx I might try it. If others do, we could self report occasionally. I saw vbookie on spurstalk long ago and have wondered if folks here might be interested in that, with its ability to allow custom propositions. RLee and DSMok1 does this version bulletin board allow it and if there were signs of interest from the group, would you consider it, facilitate it? http://www.vbulletin.org/forum/showthread.php?t=94128
thanks for the link - i've been looking for something like that.

that vbookie mod looks pretty cool

talkingpractice
Posts: 194
Joined: Tue Oct 30, 2012 6:58 pm
Location: The Alpha Quadrant
Contact:

Re: Lessons from the betting sphere

Post by talkingpractice » Tue Dec 02, 2014 11:15 pm

BigLeagueInsights wrote: My guess is that the models are regression-based and all of the inputs are relatively simple box score stats.

Injuries to superstars obviously break models, and it's not clear to me how the bookies handle this.
No, and no.

Crow
Posts: 6250
Joined: Thu Apr 14, 2011 11:10 pm

Re: Lessons from the betting sphere

Post by Crow » Wed Dec 03, 2014 1:01 am

I understand that sharps are unlikely to reveal much of their unique knowledge... but if anyone knew of an already public article on the topic that was "worth reading", I am open to such casual suggestions.

BigLeagueInsights
Posts: 63
Joined: Thu Oct 23, 2014 9:59 pm
Location: Las Vegas
Contact:

Re: Learning about / from the betting sphere

Post by BigLeagueInsights » Wed Dec 03, 2014 11:24 pm

Updated against-the-spread trends here: http://bigleagueinsights.com/2014-again ... ng-trends/

jmethven
Posts: 6
Joined: Wed Jun 19, 2013 6:34 pm

Re: Learning about / from the betting sphere

Post by jmethven » Thu Dec 04, 2014 2:28 pm

It's a misconception that bookmakers have any particularly sophisticated insight into betting lines. The general pattern of most books is to set a line with low betting limits and then gradually raise the limits as the start time of the game draws nearer. This way, they have the opportunity to determine if the line is off BEFORE they get a ton of one-sided betting action from people who probably DO have sophisticated betting models. Along the same lines, most Vegas books in fact copy the lines of off-shore books with higher limits like Pinnacle Sports and CRIS, which are more accurate because there is more liquidity in the market. The reason that spreads are the most accurate predictors of sporting events is because of the efficiency of the betting market and the ability to effectively draw on the knowledge of anyone who places a bet, not because of some special insight or models that bookmakers possess.

BigLeagueInsights
Posts: 63
Joined: Thu Oct 23, 2014 9:59 pm
Location: Las Vegas
Contact:

Re: Learning about / from the betting sphere

Post by BigLeagueInsights » Fri Dec 05, 2014 12:30 am

I took a look at 2014 Totals Betting Trends: Thus far, teams playing 3 games in 4 nights on a B2B have gone over 67% of the time.

Analysis and other trends here:
http://bigleagueinsights.com/2014-total ... ng-trends/

willguo
Posts: 26
Joined: Mon Nov 03, 2014 7:18 am

Re: Lessons from the betting sphere

Post by willguo » Sat Dec 06, 2014 10:48 am

BigLeagueInsights wrote:Obviously, the oddsmakers don't want people to know how the lines are made, so we have to guess what's inside the black box algorithms that make the lines.
Most oddsmakers tell you *exactly* how the lines are made if you ask them. Almost everyone makes their lines two ways: top-down and bottom-up. Using very rough numbers because I don't want to look anything up, and because I don't want to give away too many of my numbers, for tomorrow's GSW@CHI game (7 Dec 2014)

Top down:

1) Power rate both teams in the preseason. (Let's say +6 for both teams - 55 to 60 win range)
2) Look at current SRS-like measure. (+8 for GSW, +2 for CHI)
3) Combine the two, factoring in injuries, for current power rating. +7 for GSW, +4 for CHI? GSW about a 3 point fav on neutral ground?
4) The game is Chicago, and CHI has an extra day of rest. Roughly GSW -1 then?
5) Sense check with money-line. 1 point spread is usually -115/+105. So roughly 52% chance GSW wins.

Bottom up:

1) Rate every player. Curry I projected +6, so far this year he's been +9, so let's call him a +7. He's been playing lighter minutes this year, about 65% of minutes, but that's because GSW is killing teams. This should be a close game, so let's say he's going to play 75% of minutes.
2) Project a rotation, and sum-product for each team. I get something like GSW is a +10 team, CHI is a +5. This is a lot higher than top-down, because you're not factoring in minutes for Festus Ezeli or Nazr Mohammed. So GSW should be about a 5 point fav on neutral ground? Sounds about right - GSW's towel guys are worse than CHI's.
3) Adjust for rest/HCA. Something like GSW-3?
4) Sense check again with money line. 3 point spread is something like -150/+130, implying a 58% chance GSW wins.

Combine the two, and think about specific match-ups or coaching variables. The spread is probably around 2: 55% GSW wins. That passes the smell test for me. Let's hang GSW-2 and see what happens.

The whole taking small limits and having the market hash out the line for you is mostly crap - if my line moved so predictably, I will see sharps bet $500 on GSW until I move the line to GSW-5, then when my limits are $50,000 they will slam CHI+5. Smaller limits are more so I can protect my book from news overnight - if Steph and Klay get arrested overnight I'm in trouble. I want to wait until shoot around so I get a good feel for who's playing before I really open up the limits.

Crow
Posts: 6250
Joined: Thu Apr 14, 2011 11:10 pm

Re: Learning about / from the betting sphere

Post by Crow » Sun Dec 07, 2014 5:25 am

Thanks for the inside look.

Crow
Posts: 6250
Joined: Thu Apr 14, 2011 11:10 pm

Re: Learning about / from the betting sphere

Post by Crow » Wed Dec 10, 2014 1:26 pm

First week of fake betting done. Rallied to get to a bit over 40% win. Had a plan to try to just bet the 6 teams I thought I knew a decent amount about (6 top teams for hoped for greater reliability), only when
they were home / away, favored / dog v opponent in such a way that the power rating difference and this season ats performance was favorable, lines fell into certain slots and preferably east vs west, but lines are pretty tight and I didn't find this conceived ideal too often. So I strayed a little or too much. It is tough and I am not finding it fun either, so far. But I'll see if my improvement lasts this coming week or fades.

Statman
Posts: 548
Joined: Fri Apr 15, 2011 5:29 pm
Location: Arlington, Texas
Contact:

Re: Learning about / from the betting sphere

Post by Statman » Wed Dec 10, 2014 3:50 pm

Crow wrote:First week of fake betting done. Rallied to get to a bit over 40% win. Had a plan to try to just bet the 6 teams I thought I knew a decent amount about (6 top teams for hoped for greater reliability), only when
they were home / away, favored / dog v opponent in such a way that the power rating difference and this season ats performance was favorable, lines fell into certain slots and preferably east vs west, but lines are pretty tight and I didn't find this conceived ideal too often. So I strayed a little or too much. It is tough and I am not finding it fun either, so far. But I'll see if my improvement lasts this coming week or fades.
I really would guess finding good bets from much larger data sets - like college basketball - would be much easier. It just seems there wouldn't be all that much variance from a personal projected point spread & a Vegas line for any pro league.

Post Reply