1. #1
    Waterstpub87
    Slan go foill
    Waterstpub87's Avatar Become A Pro!
    Join Date: 09-09-09
    Posts: 4,043
    Betpoints: 7236

    Margin needed to bet basketball

    For basketball, both NBA and NCAA, I make bets based off the difference between the line I create and the posted line. For spreads in NBA it would be greater than 2.5 points(ignoring points for 1 to -1), in NCAA it is closer to 3.5. Totals similar, 3.5 and 5 points respectively. I wonder though, is this the best way.

    When I think about it, 2.5 is worth much more on 4 points spread vs 24 points, but I treat them the same. Totals this is less of an issue. Does anyone have a better way of doing this, something like % of difference to the spread?
    Points Awarded:

    semibluff gave Waterstpub87 1 Betpoint(s) for this post.


  2. #2
    KVB
    It's not what they bring...
    KVB's Avatar SBR PRO
    Join Date: 05-29-14
    Posts: 74,849
    Betpoints: 7576

    This is a tough one.

    One thought is to take individual spreads or small ranges and see the difference in success rates, play with it a bit to see if anything is significant, or just more pronounced.

    It isn't always more successful as the difference between you and the market gets larger. Often at the extremes it fails.

    Just like you said, it the difference is more pronounced with the extremes of 2 and 24. But with tighter margins in between, the difference isn't as significant.

    So there's tinkering with the different ranges of differences in your line and the spread. Now take those differences and tinker with the different offered spreads or spread ranges. Is there really a pronounced difference?

    Maybe 2.5 is worth more on a 4 point spread than a 24 point spread, but did you test and see? Test those results with your algo at those different spreads. This could be tedious and maybe even prohibitive but I know you are as solid with excel as the best of them so you can probalby work something out.

    In the end, you may find that the 2.5 isn't really worth more on a 4 point spread vs. 24.

    Percent difference to the spread can be dangerous I think but probably even when optimal, might not cause an adjustment, especially for the first 15 points of a spread or so. When I get some time I'll throw some numbers out there with some ideas, but any testing really should involve your handicapping.

    I suppose we can think of some things and then plug any capping into it.

    I need some time, I'm a little tied up, but I do get what you are asking, I'm just not sure we can get enough out of it. It's a good creativity test. Those push rates are pretty consistent and close with so many NBA numbers and it seems to distribute more evenly than many would think.

    Points Awarded:

    semibluff gave KVB 1 Betpoint(s) for this post.


  3. #3
    deeppckts
    deeppckts's Avatar Become A Pro!
    Join Date: 12-19-12
    Posts: 830
    Betpoints: 394

    1. Given how off your lines are from the market, it's clear you're handicapping and not modeling. So odds are you do not have an edge to begin with, so the optimal bet size is zero. You're just trying to pretend you're sophisticated by making your own number first. Without a real edge, talking about bet size is pointless.

    2. If you think the Knicks should be +2 but they are +3.5, use the dropdown tool to see what price you'd get on Knicks +2. Is it greater than +100? Then you have a positive EV bet there. But again, since you couldn't figure this out, I am sure you're handicapping with a pencil behind your ear and a bunch of useless stats sheets = you have no edge but think you're being sophisticated about your betting approach compared to those who just look at lines and bet when their gut disagrees.

  4. #4
    Waterstpub87
    Slan go foill
    Waterstpub87's Avatar Become A Pro!
    Join Date: 09-09-09
    Posts: 4,043
    Betpoints: 7236

    Kvb,

    Really just a random thought. I find empirically that I am playing big spreads in NCAA basketball. It makes logical sense that this could be from a lower trigger point. From what I understand, the half point calculator is built parametrically, ie that a half point in basketball is almost always worth 8 cents or so, which doesnt make sense. That half point could be worth between like 4 and 9 depending where it is. I dont have the answer to that because I haven't looked at all the data. I would attempt to scrape historical lines and get actual historical push percentages, but this is a lot of effort for a result that I am not completely sold on.

    I agree with it being a dangerous idea. On the ferry, I was thinking about it, and if you go with like 20%, it has to be 4 pts off on 20, but 1 on 3. It isnt going to be a quick fix.

    I could scrap all the lines with python, but it would be a significant undertaking, especially dealing with the errors.

  5. #5
    Fullkelly
    Fullkelly's Avatar Become A Pro!
    Join Date: 02-20-17
    Posts: 52
    Betpoints: 2017

    I would say picking an arbitary number off from market does not work that well, but is an ok starting point. Maybe convert everything to a cover percentage given the spread by using some sort of historical distribution using the SBR half point calc. Have you backtested your model and see if it has CLV? NBA Sides are pretty tough to beat, but totals are still beatable. It is interesting that your cutoff is 2.5 on NBA sides which is a tougher market to CBB which has 3.5 cutoff. BOL

  6. #6
    Waterstpub87
    Slan go foill
    Waterstpub87's Avatar Become A Pro!
    Join Date: 09-09-09
    Posts: 4,043
    Betpoints: 7236

    Quote Originally Posted by Fullkelly View Post
    I would say picking an arbitary number off from market does not work that well, but is an ok starting point. Maybe convert everything to a cover percentage given the spread by using some sort of historical distribution using the SBR half point calc. Have you backtested your model and see if it has CLV? NBA Sides are pretty tough to beat, but totals are still beatable. It is interesting that your cutoff is 2.5 on NBA sides which is a tougher market to CBB which has 3.5 cutoff. BOL
    Yea, tend to do pretty ok on the clv. Been running the same nba model for a couple years, tend to beat nba sides on average by like .85 pts or so. NCAA basketball is over 1 point currently, but it tends to get better as the season progresses. The problem with the calculator is that operationally, it is difficult to push so many things through the calculator. If I am looking at 50 games, punching 20 of them into the website and comparing is going to take awhile.

    The lines are tighter for nba. It gives me roughly 15 to 20% of NBA sides.

  7. #7
    deeppckts
    deeppckts's Avatar Become A Pro!
    Join Date: 12-19-12
    Posts: 830
    Betpoints: 394

    lol how can you have a model and not know how to calculate EV when your model says Team A should be +5 and the posted spread is +8.

  8. #8
    TommieGunshot
    TommieGunshot's Avatar SBR PRO
    Join Date: 03-27-12
    Posts: 1,555
    Betpoints: 6350

    In general, beating the line by one point on the side, or 1.5 on the total is profitable. That is a little bit stronger in the NBA than college

    Points are generally worth the same regardless of spread. They are worth less the higher the total goes. Beating the side by 2.5 points is about the same whether it's a four point spread, or a 24 point spread. But beating the side by 2.5 points is going to be stronger when the total is 197 instead of 247.

  9. #9
    Waterstpub87
    Slan go foill
    Waterstpub87's Avatar Become A Pro!
    Join Date: 09-09-09
    Posts: 4,043
    Betpoints: 7236

    Quote Originally Posted by TommieGunshot View Post
    In general, beating the line by one point on the side, or 1.5 on the total is profitable. That is a little bit stronger in the NBA than college

    Points are generally worth the same regardless of spread. They are worth less the higher the total goes. Beating the side by 2.5 points is about the same whether it's a four point spread, or a 24 point spread. But beating the side by 2.5 points is going to be stronger when the total is 197 instead of 247.
    TommieGunshot,

    Thanks for posting this. This is an interesting idea. Definitely something to consider.

Top