Modeling... if interested... here is my approach

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • usma1992
    SBR MVP
    • 08-02-11
    • 1405

    #1
    Modeling... if interested... here is my approach
    Background: Computer Science school teacher... previous life... Army Officer... Simulation expert and presenter. Degree is in Electrical Engineering with Computer Engineering Track.

    Step 1: Collect data. Has to remain pure. I have had people ask me to exchange data and I quickly asked is it pure or did up manipulate it. Data is only valuable in its infancy state. And it has to be current. You can manipulate it at your hearts content after the fact. Pure data believe it or not is a commodity.
    Step 2: Use mathematics to include standard deviations ... z scores and try to determine the variables that at least determine a reasonable score outcome. There are so many variables that impact the score that it is unreasonable to think you will predict the exact outcome but you may be able to get close.
    Step 3: Develop an algorithm to test against past data. This presents a problem because many don't actually have past data that is pure to test against. I do. If you have virgin past data and can develop an algorithm that worked in the past... try it moving forward.
    Step 4: Did you algorithm actually work? Was it close to the predictions or way off? Do you need to make adjustments or is it right but unlucky or is it completely worthless?

    I have spent well over ten years.... just getting to the put that I collect data correcting and do the math correctly. I am now creating algorithms that worked in the past and seeing if they can predict the future.

    As a modeler you are always going to tweak as new data comes in.

    College Football Results:
    Keep in mind these results are just based on the week they occurred if the same algorithms didn't work the previous 3 years they are worthless.

    Week 10 10-8
    Week 11 11-5
    Week 12 4-6
    Week 13 9-5

    You have any questions... send me a message. I'll try to accommodate as much as I can .

    My NFL program has done really well until last week went 1-6. NBA is doing well but last night 4-5. I have losses but everything trending well.

    Dave

    P.S. The irony... is I have been told it can't be done. I actually think I might have already done it for college football... I am onto NFL... but more NBA and NCAAB,
    Last edited by usma1992; 12-01-23, 05:15 PM.
  • Optional
    Administrator
    • 06-10-10
    • 60618

    #2
    Good luck.

    I agree that integrity of your data is vital.

    But I don't think using historical stats only will produce definitive results. Both teams and bookmakers adjust based on that history too, making it out of date immediately really.

    I'd be looking to incorporate a cross check of your output against something current.



    If you like stats and coding I have had a simple idea you may like to try. As I am never going to get around to it by the looks.

    Get accurate birthtime and place data for all players and calculate biorthyms.

    I'd then use it to calc head to head player positions. DEF v OFF and a few other things. You get the idea.

    Back test that to see if it aligns with your stats model and/or historical results.


    Basically looking for an up to the moment data source to filter your normal model against.
    .
    Comment
    • usma1992
      SBR MVP
      • 08-02-11
      • 1405

      #3
      Sounds like a great idea. LOL... my guess u can't spell model.
      Last edited by usma1992; 12-01-23, 07:41 PM.
      Comment
      • VeggieDog
        SBR Hall of Famer
        • 02-21-09
        • 7214

        #4
        "Get accurate birthtime and place data for all players and calculate biorhythms."

        That would be better for one-on-one contests. Tennis or MMA. Maybe even golf.
        Comment
        • usma1992
          SBR MVP
          • 08-02-11
          • 1405

          #5
          Lol... I appreciate the input... I will try to integrate next year. Until then... I'll stay my course.
          Comment
          • Optional
            Administrator
            • 06-10-10
            • 60618

            #6
            Originally posted by usma1992
            Sounds like a great idea. LOL... my guess u can't spell model.
            That went over my head.
            .
            Comment
            • JAKEPEAVY21
              BARRELED IN @ SBR!
              • 03-11-11
              • 29209

              #7
              Originally posted by Optional
              That went over my head.
              Sounds like he thinks you are a bogan
              Comment
              • VeggieDog
                SBR Hall of Famer
                • 02-21-09
                • 7214

                #8
                Originally posted by JAKEPEAVY21
                Sounds like he thinks you are a bogan
                I had to look that up. Gotta try to work it into my vocabulary. Or at least my SBR posts.
                Comment
                • A Quant
                  SBR MVP
                  • 05-14-18
                  • 1357

                  #9
                  Opti can't spell model.

                  I told you Opti he was a nutter.

                  Now you're finding out.
                  Comment
                  • Optional
                    Administrator
                    • 06-10-10
                    • 60618

                    #10
                    Originally posted by A Quant
                    Opti can't spell model.

                    I told you Opti he was a nutter.

                    Now you're finding out.
                    I think I have realized he was just being dismissive as he didn't want feedback. Tough for him to deal with time wasters like me when he already knows it all.
                    .
                    Comment
                    • usma1992
                      SBR MVP
                      • 08-02-11
                      • 1405

                      #11
                      First of all, I think you need to have a little crazy in you... if you think that you can beat Vegas on a refurbished laptop using simple math statistics and excel spreadsheet. Optimal... I really thought you were joking regarding biorythms. Who has the ability or time to do that... and would it have any bearing.

                      Right now, I am running four systems NFL, NCAAF, NBA, and NCAAB simultaneously, working a job and a half, editing my son's college papers, and trying to fine tune my results. NCAAB picks start tonight and I hope to have a decent algorithm for over/unders and spreads by XMAS.

                      However, College Football is crushing it. 38-25 the last five weeks. Championship weekend it went 4-1 and only lost the Bama game. More impressive... my system projected Michigan 25-0. It came in 26-0, 1 point from perfection. Keep in mind... my algorithms have to work the past two seasons and predict the future and they have the last 5 weeks. I have sent my picks out to the bigger hitters in my area, they are starting to believe and they are ready to spot me next year. We will be betting 500-2K per game as we currently plan. Trying to keep solid bankroll management with only 2.5% or less bet on each game.

                      I by no means have it all figured out. In fact, this is the first time in 10 years it is working. But as TJ said... The harder I work the luckier I get. I honestly believe that I have a 95% working model for college football. I know ... it can't be done.

                      Dave
                      P.S. Time to edit another college paper... why does he always wait until last minute.
                      Comment
                      • Optional
                        Administrator
                        • 06-10-10
                        • 60618

                        #12
                        Originally posted by usma1992
                        First of all, I think you need to have a little crazy in you... if you think that you can beat Vegas on a refurbished laptop using simple math statistics and excel spreadsheet. Optimal... I really thought you were joking regarding biorythms. Who has the ability or time to do that... and would it have any bearing.
                        Collecting the accurate data might take a little time. But coding to calculate the biorhythms and then how to apply those is actually kind of trivial.

                        And no, I don't know how well it would work. I don't even have an opinion on whether biorhythms have much scientific validity.

                        But if I was a modeler, I think the missing link when analyzing stats is ANYTHING that is current up to the minute to filter against.

                        And biorhythms is one thing that we can calc accurately with code, and is up to the second fresh data.


                        I am not saying it will give you the answers to cap with, but I genuinely think it might provide a filter to refine your stale data modelling.

                        Maybe someone can think of another metric like that if we don't like biorhythms.


                        Just food for thought someone here might find interesting.
                        .
                        Comment
                        • usma1992
                          SBR MVP
                          • 08-02-11
                          • 1405

                          #13
                          Appreciate your input...

                          Since I got a working model with college football or at least I believe I do ... I'm going to keep my approach.

                          Produce expected scores for each team... and competition. Collect data on when my system wins and loses. Identify when my system is strong and create algorithm to leverage those opportunities. That's what I did on NCAAF and it worked. I had a poster that used to break down where my picks were strong and weak, which helped.

                          Not the case anymore, I just get critics. Regardless, I had Alabama score 89-66 it came in 89-65 or vice versa. If I can get that close... I can adjust the algorithms to find the winning bets.

                          NBA ... I think I can do ... NCAAB I know I can do... NFL... so few games. I may start posting NCAAB... when I get a comfortable sample size and algorithm.

                          Just pumped... I got won hitting well.

                          Dave
                          Comment
                          • Optional
                            Administrator
                            • 06-10-10
                            • 60618

                            #14
                            I wasn't meaning to sound like a critic. Was mostly just trying to help encourage discussion.

                            Looking forward to seeing your good results going forward
                            .
                            Comment
                            • usma1992
                              SBR MVP
                              • 08-02-11
                              • 1405

                              #15
                              6:00 PM C CONN ST 70 BOSTON COLLEGE 80 150 -15 142.5 Too Close Over
                              6:00 PM LEMOYNE 71 ARMY 70 141 -3.5 135.0 Too Close Too Close
                              6:30 PM QUEENS U - CHAR 70 WINTHROP 77 147 -6.5 156.5 Too Close Under
                              7:00 PM NAVY 65 GEORGE WASHINGTON 78 143 -11.5 145.5 Too Close Too Close
                              7:00 PM UNC-ASHEVILLE 77 KENNESAW ST 82 159 -2.5 166.0 Too Close Under
                              7:00 PM MISS VALLEY ST 46 LIBERTY 93 139 -32 131.0 Too Close Over
                              7:00 PM LAFAYETTE 61 COLUMBIA 76 138 -8 135.5 Home Minus Too Close
                              7:00 PM W CAROLINA 77 HIGH POINT 79 155 -1.5 154.0 Too Close Too Close
                              7:00 PM STETSON 67 CHARLOTTE 66 134 -9 132.5 Too Close Too Close
                              7:00 PM NC CENTRAL 53 VIRGINIA 74 127 -22.5 123.5 Too Close Too Close
                              7:00 PM MERRIMACK 64 FLORIDA 84 148 -21 147.5 Too Close Too Close
                              9:00 PM ARK-PINE BLUFF 62 GONZAGA 97 159 -34 166.0 Too Close Under
                              6:30 PM FLA ATLANTIC 70 ILLINOIS 75 145 2 147.5 Home Plus Too Close
                              6:30 PM DELAWARE 67 XAVIER 75 142 -11.5 149.0 Too Close Under
                              6:30 PM GEORGE MASON 63 TENNESSEE 77 140 -16 135.0 Too Close Too Close
                              7:00 PM BRADLEY 73 AKRON 72 145 -1.5 138.0 Too Close Over
                              7:00 PM PROVIDENCE 66 OKLAHOMA 76 142 -5 144.5 Home Minus Too Close
                              Comment
                              • usma1992
                                SBR MVP
                                • 08-02-11
                                • 1405

                                #16
                                7:00 PM WISCONSIN 60 MICHIGAN ST 71 131 -5.5 131.5 Too Close Too Close
                                7:00 PM VILLANOVA 74 KANSAS ST 75 149 2.5 142.5 Too Close Too Close
                                7:00 PM W MICHIGAN 63 NOTRE DAME 69 132 -11 129.5 Too Close Too Close
                                7:00 PM DREXEL 62 PRINCETON 68 130 -8 128.0 Too Close Too Close
                                7:30 PM MISSOURI ST 69 MIDDLE TENN ST 64 133 1 132.5 Too Close Too Close
                                7:30 PM GEORGIA TECH 69 GEORGIA 71 140 -5.5 147.0 Too Close Under
                                7:30 PM LOUISIANA TECH 71 SF AUSTIN ST 71 142 -4 139.5 Too Close Too Close
                                8:00 PM INDIANA ST 86 N ILLINOIS 76 162 7.5 166.5 Too Close Under
                                8:00 PM MOUNT ST MARYS 63 OLE MISS 78 141 -15 134.5 Too Close Too Close
                                8:00 PM MISSOURI-KC 59 KANSAS 81 141 -26 141.0 Too Close Too Close
                                8:00 PM CORNELL 78 SYRACUSE 81 159 -5.5 166.5 Too Close Under
                                8:00 PM WEBER ST 65 UTAH VALLEY ST 61 126 2.5 127.5 Too Close Too Close
                                8:30 PM LOYOLA-IL 68 TULSA 71 139 1 141.5 Home Plus Too Close
                                8:30 PM BUFFALO 59 BUTLER 92 151 -23 149.0 Too Close Too Close
                                9:00 PM N TEXAS 66 BOISE ST 62 128 -5 125.5 Too Close Too Close
                                9:00 PM OKLAHOMA ST 69 S ILLINOIS 73 142 1 132.0 Home Plus Over
                                9:00 PM N CAROLINA 69 CONNECTICUT 78 147 -5.5 152.5 Too Close Under
                                9:00 PM SOUTHERN UTAH 159 UTAH 1 159 231 153.5 Too Close Too Close
                                9:00 PM INDIANA 69 MICHIGAN 79 148 -7 146.0 Too Close Too Close
                                9:00 PM PACIFIC 66 IDAHO 73 139 -2.5 143.0 Too Close Under
                                Comment
                                • usma1992
                                  SBR MVP
                                  • 08-02-11
                                  • 1405

                                  #17
                                  9:00 PM SETON HALL 67 BAYLOR 84 152 -11 150.5 Home Minus Too Close
                                  9:00 PM SAN DIEGO ST 73 GRAND CANYON 68 141 2.5 142.0 Too Close Too Close
                                  9:00 PM EVANSVILLE 61 BYU 93 154 -24.5 150.5 Too Close Too Close
                                  9:15 PM KENT ST 75 S DAKOTA ST 77 151 -1.5 145.0 Too Close Too Close
                                  10:00 PM IDAHO ST 73 FRESNO ST 70 142 -9.5 130.0 Too Close Over
                                  10:00 PM CLEVELAND ST 61 ST MARYS-CA 75 135 -15 134.0 Too Close Too Close
                                  11:00 PM MONTANA ST 66 WASHINGTON 81 147 -18.5 144.5 Too Close Too Close
                                  Comment
                                  • usma1992
                                    SBR MVP
                                    • 08-02-11
                                    • 1405

                                    #18
                                    These were last nights games. Only 2 days in with existing data and algorithms.

                                    But 2-0 first night... and over 60% second night on 20+ games.

                                    Tonight, I have 30 games that I am tracking. It is a lot but it is working so far and will fine tune it by last December. NBA still collecting data and plan to have that finalized for Version 3.0 by Xmas also.

                                    I provide projected numbers to every team and final score.

                                    Enjoy
                                    Dave

                                    Dave
                                    Comment
                                    • comet52
                                      SBR Hustler
                                      • 10-16-19
                                      • 59

                                      #19
                                      Originally posted by Optional
                                      Collecting the accurate data might take a little time. But coding to calculate the biorhythms and then how to apply those is actually kind of trivial.

                                      And no, I don't know how well it would work. I don't even have an opinion on whether biorhythms have much scientific validity.

                                      But if I was a modeler, I think the missing link when analyzing stats is ANYTHING that is current up to the minute to filter against.

                                      And biorhythms is one thing that we can calc accurately with code, and is up to the second fresh data.


                                      I am not saying it will give you the answers to cap with, but I genuinely think it might provide a filter to refine your stale data modelling.

                                      Maybe someone can think of another metric like that if we don't like biorhythms.


                                      Just food for thought someone here might find interesting.
                                      Interesting take Opti. As an astrologer myself I can tell you that obtaining accurate birthtimes and data for people you don't know isn't very easy. But I can also say, if you can find an edge that others don't have in any sport, you can profit. If biorythms work and you can compile meaningful data on them then that could be such an edge.

                                      The current betting markets are flooded with stats modelers and the more they find the same edges/angles, the more the market grinds those initial winners into long term losers due to the volumes they end up getting played at. If one guy had an actual working edge based on something esoteric that no one else has, he'd have a potential goldmine.
                                      Comment
                                      SBR Contests
                                      Collapse
                                      Top-Rated US Sportsbooks
                                      Collapse
                                      Working...