1. 11-06-11 01:42 PM #1
    flsaders85
    flsaders85's Avatar Become A Pro!
    Join Date: 11-06-11
    Posts: 68
    Betpoints: 5253

    College Hoops Modelers

    So I started modeling college hoops which is extremely tough in itself, but it looks like I have found an edge in picking sides, but not totals. My question is: my previous assumption was that it would be easier to find an edge in picking totals than sides...is this true? If I am constructing a model to compute a score for two teams, why would it have successful results picking sides but not totals?
    Reply With Quote Reply With Quote
  2. 11-06-11 02:38 PM #2
    tukkk
    ★★★★★
    tukkk's Avatar Become A Pro!
    Join Date: 10-04-10
    Posts: 391
    Betpoints: 1

    human brain tends to find patterns too easily, consider your sample size first
    Reply With Quote Reply With Quote
  3. 11-06-11 05:35 PM #3
    mathdotcom
    mathdotcom's Avatar Become A Pro!
    Join Date: 03-24-08
    Posts: 11,689
    Betpoints: 1943

    The way some model it, they are basically modeled together so that neither is more difficult than the other. Ie. they predict total pts for Team A and total pts for Team B and then the difference is the spread and the sum is the total.

    But I would still answer it should be a bit easier to go after totals for no other reason than the sharpest guys are more focused on the bigger markets, and totals markets are smaller.
    Reply With Quote Reply With Quote
  4. 11-06-11 06:45 PM #4
    Justin7
    Justin7's Avatar Become A Pro!
    Join Date: 07-31-06
    Posts: 8,577
    Betpoints: 1506

    How many seasons' worth of data are you using for backtesting?
    Reply With Quote Reply With Quote
  5. 11-06-11 07:47 PM #5
    flsaders85
    flsaders85's Avatar Become A Pro!
    Join Date: 11-06-11
    Posts: 68
    Betpoints: 5253

    I used three seasons of data for my model (2005-2007) and backtested it on those years and 2008-2010.

    Mathdotcom: The model you speak of is what I am doing (predicting two teams scores). The sides outperformed totals by a lot.
    Reply With Quote Reply With Quote
  6. 11-07-11 10:45 AM #6
    mathdotcom
    mathdotcom's Avatar Become A Pro!
    Join Date: 03-24-08
    Posts: 11,689
    Betpoints: 1943

    You shouldn't backtest it on 2005-2007 because that's the data you used to create the model.
    Reply With Quote Reply With Quote
  7. 11-07-11 08:06 PM #7
    flsaders85
    flsaders85's Avatar Become A Pro!
    Join Date: 11-06-11
    Posts: 68
    Betpoints: 5253

    Regardless, the 2008-2010 has the same results which makes me question the model and my database. Or maybe the model is only only successful with sides.
    Reply With Quote Reply With Quote
  8. 11-07-11 08:43 PM #8
    mathdotcom
    mathdotcom's Avatar Become A Pro!
    Join Date: 03-24-08
    Posts: 11,689
    Betpoints: 1943

    There is no one way to model anything, so if you have something that seems to work for totals then stick with it. If it's not working for sides then change that one up. If you're modeling anything reasonably complicated you always have to be re-evaluating your model.
    Reply With Quote Reply With Quote
  9. 11-08-11 08:14 AM #9
    Wrecktangle
    Wrecktangle's Avatar Become A Pro!
    Join Date: 03-01-09
    Posts: 1,524
    Betpoints: 3209

    Quote Originally Posted by mathdotcom View Post
    There is no one way to model anything, so if you have something that seems to work for totals then stick with it. If it's not working for sides then change that one up. If you're modeling anything reasonably complicated you always have to be re-evaluating your model.
    It's a rare thing that I agree with this guy...
    Reply With Quote Reply With Quote
  10. 11-08-11 07:56 PM #10
    flsaders85
    flsaders85's Avatar Become A Pro!
    Join Date: 11-06-11
    Posts: 68
    Betpoints: 5253

    Thanks guys
    Reply With Quote Reply With Quote
Top
« Previous Thread | Handicapper Think Tank | Next Thread »