1. #1
    Bsims
    Bsims's Avatar SBR PRO
    Join Date: 02-03-09
    Posts: 765
    Betpoints: 36

    New system development

    I am starting this thread based on the purpose of this forum; “Handicapping theories, betting systems, tips, tricks, odds and math”. I am currently beginning work on a new system and will post some about it here. First, I hope to provide some ideas or concepts to beginners or those who might not have considered these ideas. Second, I will explain on how I have addressed some problems and provide my solutions. Finally, I hope some of you might have different approaches to solving these problems.
    Points Awarded:

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


  2. #2
    Bsims
    Bsims's Avatar SBR PRO
    Join Date: 02-03-09
    Posts: 765
    Betpoints: 36

    The basis of the system is using the book makers odds to compute what they say is the expected score of the event in question. The simplest situation is basketball. The books post a point spread and over/under totals. From these we can easily compute what their expected score is. For example, say Team 1 is an 8-point favorite and the total is 200. That implies the final score expected from the odds is Team 1 to win 104 to 96.

    The reason basketball is the simplest form is that their spreads and totals are almost always even on both sides (-110 or -105). Take baseball for instance. The problem is the books give us money lines to pick the winner of the games. The totals offer is a total and differed odds for the over and under. The question is how do we convert these to expected runs per game? I’ll begin to deal with how I addressed this problem in following posts.

  3. #3
    Bsims
    Bsims's Avatar SBR PRO
    Join Date: 02-03-09
    Posts: 765
    Betpoints: 36

    I’ll start the approach with some recent baseball lines from Japan and Korea. The first step is to convert the Las Vegas style lines to European style. This is done with the following Excel formula.

    =IF(B2>0,+(100+B2)/100,1-100/B2)

    Bookmaker Las Vegas Format Lines Team1 Team2 Total Over Under
    Saitama Seibu Lions-Chiba Lotte Marines -115 -105 9 -105 -115
    Yomiuri Giants-Hanshin Tigers -140 120 7 -110 -110
    NC Dinos-SK Wyverns -175 155 10.5 -120 100
    Converting to European Stye Lines Team1 Team2 Total Over Under
    Saitama Seibu Lions-Chiba Lotte Marines 1.87 1.95 9 1.95 1.87
    Yomiuri Giants-Hanshin Tigers 1.71 2.20 7 1.91 1.91
    NC Dinos-SK Wyverns 1.57 2.55 10.5 1.83 2.00

  4. #4
    Bsims
    Bsims's Avatar SBR PRO
    Join Date: 02-03-09
    Posts: 765
    Betpoints: 36

    One can use the European odds to compute expected probabilities. This is done by using the reciprocal of the European odds (1/Odds). When you do the first conversion to probabilities, thay add up to more than 1 (because of the juice). These probabilities can then be normalized by dividing each by the sum of the probabilities (i.e. =+B13/(SUM(+$B13+$C13)) in Excel). Following are the conversions of our example lines.

    Computed Probability Team1 Team2 Total Over Under
    Saitama Seibu Lions-Chiba Lotte Marines 53.5% 51.2% 9 51.2% 53.5%
    Yomiuri Giants-Hanshin Tigers 58.3% 45.5% 7 52.4% 52.4%
    NC Dinos-SK Wyverns 63.6% 39.2% 10.5 54.5% 50.0%
    Normalized Computed Probability Team1 Team2 Total Over Under
    Saitama Seibu Lions-Chiba Lotte Marines 51.1% 48.9% 9 48.9% 51.1%
    Yomiuri Giants-Hanshin Tigers 56.2% 43.8% 7 50.0% 50.0%
    NC Dinos-SK Wyverns 61.9% 38.1% 10.5 52.2% 47.8%

  5. #5
    Bsims
    Bsims's Avatar SBR PRO
    Join Date: 02-03-09
    Posts: 765
    Betpoints: 36

    The first problem is to use the total line and odds to estimate the expected total runs. In the first game above, the total is 9, the over is -105 while the under is -115. This implies that the expected total is a bit less than 9. The second game has the odds at -110 both ways implying the expected total is actually 7. The odds on the third game imply that the expected total is greater than 10.5. I use the following formula to make these adjustments.

    New Total = Old Total-(10*(0.5*Over Pct))

    Normalized Computed Probability Team1 Team2 Total Over Under
    Saitama Seibu Lions-Chiba Lotte Marines 51.1% 48.9% 9 48.9% 51.1%
    Yomiuri Giants-Hanshin Tigers 56.2% 43.8% 7 50.0% 50.0%
    NC Dinos-SK Wyverns 61.9% 38.1% 10.5 52.2% 47.8%
    Adjust Total For Odds New Total = Old Total-(10*(0.5*Over Pct))
    Saitama Seibu Lions-Chiba Lotte Marines 8.89
    Yomiuri Giants-Hanshin Tigers 7.00
    NC Dinos-SK Wyverns 10.72

  6. #6
    Bsims
    Bsims's Avatar SBR PRO
    Join Date: 02-03-09
    Posts: 765
    Betpoints: 36

    The final step is allocating the estimated total runs for each team based on their win probabilities. This is done with a complicated formula that I developed a few years ago.

    Team2 Runs = (2*Team2Pct*Total-SQR(4*Team2Pct*Total^2*(1-Team2Pct)))/(4*Team2Pct2)

    Estimated Runs Team1 Team2
    Saitama Seibu Lions-Chiba Lotte Marines 4.49 4.40
    Yomiuri Giants-Hanshin Tigers 3.72 3.28
    NC Dinos-SK Wyverns 6.00 4.71

  7. #7
    yak merchant
    yak merchant's Avatar Become A Pro!
    Join Date: 11-04-10
    Posts: 107
    Betpoints: 6150

    Well I’ll have to follow this thread. I’m always interested how people explain how they are going take the bookmakers line, do some math on it and then plan on using that to beat the bookmakers line.

  8. #8
    trytrytry
    All I do is trytrytry
    trytrytry's Avatar SBR PRO
    Join Date: 03-13-06
    Posts: 21,843
    Betpoints: 20856

    so are you searching for slighlty weak team total listings at books from this "key" number? maybe but juice on team totals is also quite high so you really have to have an edge to get that working. or are you seeking some other overlay here?

  9. #9
    Bsims
    Bsims's Avatar SBR PRO
    Join Date: 02-03-09
    Posts: 765
    Betpoints: 36

    Once again my health has improved enough to make me interested in resurrecting some of the previous work I’ve done. In particular, my like game system. It was described in the following thread.

    https://www.sportsbookreview.com/for...es-system.html

    The system requires estimated scores for the like games matching routine. It uses the Las Vegas odds to generate these matching scores. I've described how I generate these in my current implementation. I’m hoping for a better method to generate these estimated scores.

  10. #10
    Bsims
    Bsims's Avatar SBR PRO
    Join Date: 02-03-09
    Posts: 765
    Betpoints: 36

    I'm under no illusions that using the like games will give me any edge when it comes to betting money lines or totals, since these are the basis of the matching algorithm. But there are two questions that I want to pursue.

    First, can one use the like game results to support or reject other potential bets? Second, can the like game results be used to make some more obscure bets. For instance in baseball, run lines, the first five inning offers, parlays, etc.

  11. #11
    Bsims
    Bsims's Avatar SBR PRO
    Join Date: 02-03-09
    Posts: 765
    Betpoints: 36

    While I‘d like to address those two questions, the thing that has regenerated my interest is soccer. There are a lot of game and miscellaneous wager types. That makes me wonder about the potential for using the like games to look for good wagers here. That will require a lot of work and questions. The first is how to generate the estimated scores. Soccer has three potential money line results with the inclusion of the draw possibility. Also, the totals are very low and the odds on these vary a lot. This calls for a more complex way to generate the estimated total.

  12. #12
    semibluff
    refuses to update status
    semibluff's Avatar Become A Pro!
    Join Date: 04-12-16
    Posts: 910
    Betpoints: 1187

    Quote Originally Posted by Bsims View Post
    While I‘d like to address those two questions, the thing that has regenerated my interest is soccer. There are a lot of game and miscellaneous wager types. That makes me wonder about the potential for using the like games to look for good wagers here. That will require a lot of work and questions. The first is how to generate the estimated scores. Soccer has three potential money line results with the inclusion of the draw possibility. Also, the totals are very low and the odds on these vary a lot. This calls for a more complex way to generate the estimated total.
    The most obvious way to manually generate totals/handicaps for an individual soccer game is to to take all the odds for each realistically possible game score and manually adjust those odds in line with game outcome odds then take out the juice to have the true odds. You would have to adjust the odds because books add extra juice into game correct-score odds for the underdog. There is also a tendency to add a bit of extra juice into the most likely 1 or 2 score(s) for the favourite. You may also have an issue with different books posting different odds. My advice with this is to copy/paste odds lines for both game outright and correct score odds from multiple books into Excel and then use a weighted average. From there you can use Excel formulas to extrapolate which bits of data you want for your own purposes.

    The above is only going to show data for 1 specific game. Even with short-cuts it's time-consuming to generate and from past experience the clarity it provided didn't show me any edge, although it definitely high-lighted the bad odds on underdog correct-scores & double-result. The above won't help you generate a database of odds and it won't help you determine which games are like games. The only good news I can give you is that if you're determined to pursue such an endeavour I could help provide some pre-prepared soccer odds conversion Excel sheets. I wrote the Excel formulas and layout in them for a different purpose but they're easily customized and it should be much quicker than starting from scratch. I'm sure you're better with Excel than I am but if you do want them pm me with an email address. (I can't post them here because the site doesn't allow attachments).
    Points Awarded:

    Bsims gave semibluff 10 Betpoint(s) for this post.


  13. #13
    Bsims
    Bsims's Avatar SBR PRO
    Join Date: 02-03-09
    Posts: 765
    Betpoints: 36

    Thanks semibluff! Excellent idea. I’ll certainly look into this as an alternative to using an approach that considers the major moneylines and totals. It will be interesting to compare the results. Like you indicated, the problem will be in getting the odds from multiple books for multiple games rapidly. I’d also be interested in historical data to backtest, but this is unlikely. I would do the averaging approach. I write programs to do the calculations, so could handle looking at a lot of games without manual intervention. I think you just added to my workload.

Top