[RabidGolfer]

That's an interesting, I honestly hadn't looked at it like that (too few games to make a decent profit). While my theory applied broadly to every game may only yield a 53% or 54% win percentage (If that), I would like to think using a little bit of discrimination based on the way the teams are playing might yield me a better percentage. For instance, I would never bet this way if the dog was Sacramento, since we've all seen them choke and give up huge amounts of points. Also, some games are legitimate 3 point spreads, And I don't think you can write any formula or criteria to differentiate Which ones contain value and which ones don't. Since this is the case and it's kind of a subjective system, the only way I'm going to be able to determine the success percentage is by tracking my results day forward. I'm already monitoring my picks in a thread on the NBA forum, so I will simply add a subset of my picks called public biased picks. Let's track this for a month or so and see how it does.

[brettd]

With that being said, Golfer, what is enough of an edge in terms of points per game where you would foresee being profitable long term? If a fair line should theoretically be -3 and you get it at -1 every game, is that going to be enough of a margin to win decent money long term, or are you more likely to hit 53 or 54% and scrape out small profits since your subset of games is so small?

If you got two points better long term than the closing line, you'd be flying around in a private jet in a few years time by betting to edge. An average one point deviation from the closing spread, and you are making very solid returns against the NBA/NCAAB point spread in the long term.

The million dollar question is whether you can achieve even a one point closing line value. KenPom is not going to help you do that for NCAAB.

[EXhoosier10]

If you got two points better long term than the closing line, you'd be flying around in a private jet in a few years time by betting to edge. An average one point deviation from the closing spread, and you are making very solid returns against the NBA/NCAAB point spread in the long term.

The million dollar question is whether you can achieve even a one point closing line value. KenPom is not going to help you do that for NCAAB.

I'd love to hear the winning percent over the last few years (I suppose for home teams) when beating NBA closers by 1 point compared to using closing numbers.

[wrongturn]

I'd love to hear the winning percent over the last few years (I suppose for home teams) when beating NBA closers by 1 point compared to using closing numbers.

Winning percentage increases by about %4 in average.

[ask]

When you guys do backtesting of a model/system, do you use opening or closing betting lines? And how much does a closing line deviate from its opening line in general?

[RabidGolfer]

When you guys do backtesting of a model/system, do you use opening or closing betting lines? And how much does a closing line deviate from its opening line in general?

Great question! I'll probably use closing lines, as I typically place my bets and make my decisions 1 - 2 hours prior to the game. I've been using Statfox.com, and I assume those are closing lines.

[RabidGolfer]

The more we discuss this, the more I realize that backtesting this theory will be next to impossible. Everything hinges on public perception, which we all know is fluid. So for example, the Memphis @ Atlanta game fits my criteria perfectly tonight except that neither team is playing all that well. This game is probably a legit 2.5 point game and not a trap line. So it would be really hard to go back over years of data and find games where they public was lured into taking a popular team on a low spread, because we don't know who was actually popular at that point in time. I may just have to track this for a couple of seasons on a day forward approach. Thoughts?

[flsaders85]

The more we discuss this, the more I realize that backtesting this theory will be next to impossible. Everything hinges on public perception, which we all know is fluid. So for example, the Memphis @ Atlanta game fits my criteria perfectly tonight except that neither team is playing all that well. This game is probably a legit 2.5 point game and not a trap line. So it would be really hard to go back over years of data and find games where they public was lured into taking a popular team on a low spread, because we don't know who was actually popular at that point in time. I may just have to track this for a couple of seasons on a day forward approach. Thoughts?

This approach probably worked 5-10 years ago when linemakers juiced certain lines based off public perception, but what you're seeing today is linemakers attempt to create accurate/efficient lines that are pounded by sharps. Public money is nothing compared to synidcate/sharp money or whatever you want to call it. Now, square books do exist where they will shade a favorite like the Heat....meaning if Pinny lays Heat -3, square book will lay -3.5. But as you will experience and hear from others, any square book cuts your limits if you play the steam. I know you said you are a trader. NBA sides is one of the most efficient markets out there. You can't beat it long term without a model that incorporates predictive data that no one else use.

[RabidGolfer]

This approach probably worked 5-10 years ago when linemakers juiced certain lines based off public perception, but what you're seeing today is linemakers attempt to create accurate/efficient lines that are pounded by sharps. Public money is nothing compared to synidcate/sharp money or whatever you want to call it. Now, square books do exist where they will shade a favorite like the Heat....meaning if Pinny lays Heat -3, square book will lay -3.5. But as you will experience and hear from others, any square book cuts your limits if you play the steam. I know you said you are a trader. NBA sides is one of the most efficient markets out there. You can't beat it long term without a model that incorporates predictive data that no one else use.

Good explanation, thank you. With EXhoosier10's help, I just did some backtesting of all games from 2008 - 2010 using closing lines from 5Dimes and applying my criteria. There were 32 games that fit the criteria, and the home dog went 19-12-1 ATS. That's about 59%, but like someone said previously, there are so few games per season that match up, that it would be hard to make a big profit with this system. I may keep an eye out for the obvious ones and pound them, but I probably won't bank on only this long term. Thanks for everyone's input!

[njcardshark]

Good explanation, thank you. With EXhoosier10's help, I just did some backtesting of all games from 2008 - 2010 using closing lines from 5Dimes and applying my criteria. There were 32 games that fit the criteria, and the home dog went 19-12-1 ATS. That's about 59%, but like someone said previously, there are so few games per season that match up, that it would be hard to make a big profit with this system. I may keep an eye out for the obvious ones and pound them, but I probably won't bank on only this long term. Thanks for everyone's input!

I acually agree with you. But not just to the extent of public teams, but rather public perception itself. Throughout the course of the season perception will vary even with the popular teams. I think there are a large number of spots where you can find value in betting dogs against teams in public favor particularly at home. Obviously you can't do it blindly, but you can use it as a baseline to find spots where the popular play would be overvalued, regardless of whether or not it acually a "public team." I think in order to do this its important to be able to create a reasonable line yourself in order get a baseline to recognize the spots where there could be some value in the dog, and where the lines are acually just sharp. I personally use Sagarin ratings as one of the tools I use to create my lines, and if I find a reasonable margin between the lines I can use it to see if there is acually value there.

[HUY]

I've been reading this "public bias" nonsense for years now. Never seen any proof whatsoever that team popularity gives any bias to the odds offered. On the other hand, I've seen numerous studies denying this.

[peacebyinches]

I agree in principle with pretty much every point you’ve made Rabidgolfer (some points more/less than others) but we are on the same page for sure. I too, find it really hard to deny the presence of some skews arising in line setting and odds as a function of team popularity.

The example that always comes to mind is the existence of purely blind bettors that wager on their ‘home-town’ favorite team most games or even each and every game (or in some cases, each “important” game). The logical result of this is that more blind wagering occurs for bigger market teams. I had a friend that was such a diehard cowboys fan that regardless of whether they were favored or not, he would bet any takers that they would win straight up (especially for “important” games, rivalry games, games with playoff implications, etc…) and once said something to the effect of “if you won’t bet on your team no matter what, you ain’t a real fan like me”.

That example is the only way I find the term “public money” actually discernible from “sharp money”. But of course these blind bettors are going to have a very minimal impact in biasing the market, which makes me question how to actually find the cutoff criteria between lines that are/are not skewed.

What factors make a team, a ‘public team’? For example, in the approach we are taking here, whats actually more important in explaining betting market dynamics? Overall popularity, as in more fans = more ‘public’ bias? Or maybe overall perception, as in greater regard of team skill = more ‘public’ bias? (For that last one I am thinking along the lines of public perception being skewed due to people under the mindset of “The Heat are unbeatable! I’m betting the house they’ll win the championship! How could they ever lose??” (or whatever team, the Heat just work best here imo)) Is it some interaction of these variables? How do we go about quantifying these variables? Are there underlying aspects of these variables that we would be remiss to not at least consider, such as the fluid (or static) nature which we define our criterion, year to year, week to week, game to game?


I have my own opinions on much of this, but would like to hear from others first, otherwise this post will go on forever, haha.