Thought experiment for you "do I have a winning model" types

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • hutennis
    SBR Wise Guy
    • 07-11-10
    • 847

    #71
    Originally posted by mathdotcom
    This is the same old topic you keep bringing up in every single thread.

    Here is the short counterexample to your claim (in my best attempt at hu-english): Vegas model is opener lines, like example RedSox/Yankees -105/-115... but almost always the line is move and so closer lines different from open line and who yu tink move dis line? not vegas.... must be someone... with better Ferrari
    I dont understand the hostility.

    I keep on asking the same question b/c I don't ever get the answer.
    Even now, you did not answer it.
    What happens to a Vegas line after it's opened is an interesting subject and we can come back to it later,
    but it is not what i asked about.

    I asked about the basis for belief that line modeled by inspired handicapper may have an edge over Vegas line.

    Let's say inspired handicapper took 5000 data point on 10 aspects of the game, ran regression analysis and came up with with the fair spread for the match -9.
    Vegas line opens -8.
    Can the inspired handicapper assume that he has an edge over Vegas and, if yes, on what basis?
    What if Vegas ran regression analysis on 10000 data points of 20 aspects of the game and -8 is in fact better line produced by better approach.
    How, without knowing what exactly Vegas did to produce its line can inspired handicapper draw any conclusions about his presumed edge and if he still does how valid those conclusions are?
    Last edited by hutennis; 08-18-12, 12:07 PM.
    Comment
    • Analysis
      SBR Rookie
      • 08-18-12
      • 3

      #72
      Originally posted by hutennis
      I dont understand the hostility.

      I keep on asking the same question b/c I don't ever get the answer.
      Even now, you did not answer it.
      What happens to a Vegas line after it's opened is an interesting subject and we can come back to it later,
      but it is not what i asked about.

      I asked about the basis for belief that line modeled by inspired handicapper may have an edge over Vegas line.

      Let's say inspired handicapper took 5000 data point on 10 aspects of the game, ran regression analysis and came up with with the fair spread for the match -9.
      Vegas line opens -8.
      Can the inspired handicapper assume that he has an edge over Vegas and, if yes, on what basis?
      What if Vegas ran regression analysis on 10000 data points of 20 aspects of the game and -8 is in fact better line produced by better approach.
      How, without knowing what exactly Vegas did to produce its line can inspired handicapper draw any conclusions about his presumed edge and if he still does how valid those conclusions are?
      In the described scenario neither the bookie or handicapper is able to do any sort of assumptions.

      But what if they'd ran 1000, or 10 000 games against each other - with the handicappers model more accurate than the bookies - could they then draw any conclusions about edge?
      Comment
      • hutennis
        SBR Wise Guy
        • 07-11-10
        • 847

        #73
        Originally posted by Analysis
        In the described scenario neither the bookie or handicapper is able to do any sort of assumptions.

        But what if they'd ran 1000, or 10 000 games against each other - with the handicappers model more accurate than the bookies - could they then draw any conclusions about edge?
        If handicapper does not want to use results of his predictions for evaluating his abilities (much better way)
        and wants to look back trying to find conformation in a bell curve, he has to make sure:

        1. He does not twitch model ever, does not fit data ever, you know, things of that sort.
        I have no idea how can average person be so disciplined and not to fall for irresistible urge to make himself look
        better than he really is. Very hard. That's why this type of quality control is usually done independently.

        2. He has to have an absolutely clean record of his implied probabilities.
        I still yet to see something like this posted:" I have such and such W/L record with avg implied of 51.24%.
        Instead, it's always "I have such and such W/L record and my avg odds are somewhere in a neighborhood of -130 or so.
        How good am I?"
        This is ridiculously bad.
        Binomial calculator is a great toy, but if you want to know the truth you must be precise.
        Tiny percentage difference can get you from 192 to 1 to a 47 to 1 or to 9 to 1 in a blink of an eye.

        Plus. Is bell curve even a right method?
        I know it will do a great job analyzing models of air flow produced in a wind tunnel.
        Sock market? Sports betting? I'm not so sure.
        Last edited by hutennis; 08-18-12, 04:41 PM.
        Comment
        • mathdotcom
          SBR Posting Legend
          • 03-24-08
          • 11689

          #74
          hutennis has had problems understanding selection bias for a few years now

          Sports betting and handicapping forum: discuss picks, odds, and predictions for upcoming games and results on latest bets.
          Comment
          • illfuuptn
            SBR MVP
            • 03-17-10
            • 1860

            #75
            Originally posted by Inspirited
            This is kind of like hut's argument that he repeats over and over again in just about every thread. You think you have a significant system/model that beats the books, but it is likely that you're really just one of the lucky ones.

            It seems like the way mathy put it, this needs to be expressed in a bayesian form. Given the knowledge that mathy has, his posterior leads him to believe the coin are fair, but given the knowledge that Mr. Ed has, Mr. Ed believes the coin likely to be unfair.

            ???
            hutennis, why you rip this post? He's absolutely right. MDC didn't present this problem in the clearest of ways and I think that led to some confusion.

            What are the odds in Mr. Ed's mind that his coin is unfair? Well that's whatever all that jazzy math works out to be.

            Odds from my perspective, given that I know all coins identical AND nobody but Eddy thinks they're biased? Pretty damn close to 0.

            Just a difference in a priori knowledge...
            Comment
            • mr.ed
              SBR High Roller
              • 09-07-07
              • 211

              #76
              I believe Mathy is stating that Mr. Ed's claims on another thread are most likely bogus because hundreds of bettors come up with models/systems and very, very few are actually successful. He can comfortably make this conclusion on his hypothetical scenario with the hundreds of coin flippers because the coins are identical.

              But in the real world, Mr Ed's coin is unique and nobody else is using his coin, so if the W/L numbers quoted are accurate, Mathy's and Mr. Ed's determination of whether or not it he has a profitable model/system should be identical.
              Comment
              • CanuckG
                SBR Posting Legend
                • 12-23-10
                • 21978

                #77
                Is hutennis a ghost of Justin7? No way one person can be this dumb.
                Comment
                • hutennis
                  SBR Wise Guy
                  • 07-11-10
                  • 847

                  #78
                  Originally posted by CanuckG
                  Is hutennis a ghost of Justin7? No way one person can be this dumb.
                  I just looked at your latest posts.

                  Judging by that alone, when people like you call me dumb I can not be happier.
                  Comment
                  • Wrecktangle
                    SBR MVP
                    • 03-01-09
                    • 1524

                    #79
                    MDC = MF = Hutinnis
                    Comment
                    • cyberbabble
                      SBR Wise Guy
                      • 08-30-10
                      • 772

                      #80
                      The end is near. hutennis is now in charge of the HTT.
                      All pretenders and challengers now cower in fear of the wrath of mighty hutennis.
                      Comment
                      • GunShard
                        SBR Posting Legend
                        • 03-05-10
                        • 10031

                        #81
                        Let's see if I can figure this out.

                        1. 1 person out of 100 comes back.
                        2. That 1 person, did 200 coin flips, 130 heads, 70 tails. That coin landed 65% on heads.
                        3. You said 100 people given 100 normal quarters. I'm assuming the quarters are normal and not rigged.
                        4. 99 people will have 50% coins and the 1 person that came back has a 65% heads coin.

                        100 people overall probablity of everyone's coin tosses is 50.15% that will land heads.

                        If more people come back with their results the percentage will either change or stay the same.
                        Last edited by GunShard; 08-27-12, 03:50 AM.
                        Comment
                        • hutennis
                          SBR Wise Guy
                          • 07-11-10
                          • 847

                          #82
                          Originally posted by GunShard
                          Let's see if I can figure this out.

                          1. 1 person out of 100 comes back.
                          2. That 1 person, did 200 coin flips, 130 heads, 70 tails. That coin landed 65% on heads.
                          3. You said 100 people given 100 normal quarters. I'm assuming the quarters are normal and not rigged.
                          4. 99 people will have 50% coins and the 1 person that came back has a 65% heads coin.

                          100 people overall probablity of everyone's coin tosses is 50.15% that will land heads.

                          If more people come back with their results the percentage will either change or stay the same.

                          It is like some kind of weird joke.

                          All your percentage calculations make no sense at all.
                          The results of flipping by each person are independent of each other.
                          No matter what we know or dont know about rest of the coins, after one coin out of 100 came back with such a bias result the initial assumption that all coins are fair can be safely dismissed.
                          It is simply too unlikely that we would observe 1 in 100000 coins result within 100 coins sample.
                          Now the valid assumption is that, no matter what other results will be, at least one coin is not fair.
                          Total number of coins (100) plays a major role. If we would have 1000000 coins then it would be a different story.
                          Now knowledge about total number of coins involved in experiment does make difference.

                          The lottery example in another post is a perfect analogy.

                          If 1 out 100 mil lottery players wins it is totally expected and the fact that winner got so lucky will not raise any questions about possible cheating.
                          Too many players involved. Somebody just had to.

                          But if only 100 people would buy tickets and one of them would win, an investigation would be open in two minutes.
                          Last edited by hutennis; 08-27-12, 12:30 PM.
                          Comment
                          • uva3021
                            SBR Wise Guy
                            • 03-01-07
                            • 537

                            #83
                            Comment
                            • MonkeyF0cker
                              SBR Posting Legend
                              • 06-12-07
                              • 12144

                              #84
                              Originally posted by hutennis
                              It is like some kind of weird joke.

                              All your percentage calculations make no sense at all.
                              The results of flipping by each person are independent of each other.
                              No matter what we know or dont know about rest of the coins, after one coin out of 100 came back with such a bias result the initial assumption that all coins are fair can be safely dismissed.
                              It is simply too unlikely that we would observe 1 in 100000 coins result within 100 coins sample.
                              Now the valid assumption is that, no matter what other results will be, at least one coin is not fair.
                              Total number of coins (100) plays a major role. If we would have 1000000 coins then it would be a different story.
                              Now knowledge about total number of coins involved in experiment does make difference.

                              The lottery example in another post is a perfect analogy.

                              If 1 out 100 mil lottery players wins it is totally expected and the fact that winner got so lucky will not raise any questions about possible cheating.
                              Too many players involved. Somebody just had to.

                              But if only 100 people would buy tickets and one of them would win, an investigation would be open in two minutes.
                              So, if a slot machine (which uses a random number generator) comes out of the box, onto the casino floor, and pays two jackpots in the very first two spins of live action, the machine MUST be broken. Right? It can't possibly happen in two spins. That RNG must be biased.

                              What if it happened in two of the first four, five, six spins?

                              The Big Bang Theory can't possibly be true. Can it? Do you know the odds?!?!?!?!?
                              Comment
                              • hutennis
                                SBR Wise Guy
                                • 07-11-10
                                • 847

                                #85
                                Originally posted by MonkeyF0cker
                                So, if a slot machine (which uses a random number generator) comes out of the box, onto the casino floor, and pays two jackpots in the very first two spins of live action, the machine MUST be broken. Right? It can't possibly happen in two spins. That RNG must be biased.

                                What if it happened in two of the first four, five, six spins?

                                The Big Bang Theory can't possibly be true. Can it? Do you know the odds?!?!?!?!?

                                Who said that machine MUST be broken?
                                Who said that 1 winner out of only 100 who bought lottery tickets IS a cheat?
                                Who said that one coin out of 100 that delivered 130 heads in 200 flips IS NOT fair for sure?

                                Those are just most reasonable assumptions?
                                Anyone with different assumptions MUST be out of his mind.

                                And based on these assumptions machine must be taken out of the floor and checked out, lottery player must be investigated and coin
                                flipped a few hundreds times more.

                                As far as Big Bang Theory...
                                Here we are dealing with a sample space that is a bit larger than a few coins, few lottery players and a few slot machines.
                                In fact it's large enough to kinda make anything possible, no matter what.
                                Comment
                                • mathdotcom
                                  SBR Posting Legend
                                  • 03-24-08
                                  • 11689

                                  #86
                                  Monkey

                                  Slot machine of floor break impossible number spin jackpot many time sample size bayesian hah cannot be expected hypothesis test Palms casino or Wynn different story latitude longitude SBR no one central limit theorem violate coins security cameras
                                  Comment
                                  • MonkeyF0cker
                                    SBR Posting Legend
                                    • 06-12-07
                                    • 12144

                                    #87
                                    Originally posted by hutennis
                                    Who said that machine MUST be broken?
                                    Who said that 1 winner out of only 100 who bought lottery tickets IS a cheat?
                                    Who said that one coin out of 100 that delivered 130 heads in 200 flips IS NOT fair for sure?
                                    Backtracking?

                                    Originally posted by hutennis
                                    No matter what we know or dont know about rest of the coins, after one coin out of 100 came back with such a bias result the initial assumption that all coins are fair can be safely dismissed.
                                    It is simply too unlikely that we would observe 1 in 100000 coins result within 100 coins sample.
                                    Comment
                                    • hutennis
                                      SBR Wise Guy
                                      • 07-11-10
                                      • 847

                                      #88
                                      Yes.

                                      after one coin out of 100 came back with such a bias result the initial assumption that all coins are fair can be safely dismissed.

                                      And another assumption should be made.
                                      At very least we should assume that we don't know enough about this coin to draw any conclusion and coin should be farther checkedout.
                                      Comment
                                      • wrongturn
                                        SBR MVP
                                        • 06-06-06
                                        • 2228

                                        #89
                                        So nobody should talk about what he would do with his lottery win money, because his assumption of winning can be safely dismissed. OR everybody in USA should talk about what they would do with lottery win money, because it is certain that one will win. Cool concept!
                                        Comment
                                        • hutennis
                                          SBR Wise Guy
                                          • 07-11-10
                                          • 847

                                          #90
                                          Originally posted by wrongturn
                                          So nobody should talk about what he would do with his lottery win money, because his assumption of winning can be safely dismissed. OR everybody in USA should talk about what they would do with lottery win money, because it is certain that one will win. Cool concept!
                                          This is your concept. If it looks cool to you - I'm happy.
                                          Comment
                                          • wrongturn
                                            SBR MVP
                                            • 06-06-06
                                            • 2228

                                            #91
                                            Originally posted by hutennis
                                            This is your concept. If it looks cool to you - I'm happy.
                                            If you can not tell whose concept it is, then you have bigger problem than I realize.
                                            Comment
                                            • cyberbabble
                                              SBR Wise Guy
                                              • 08-30-10
                                              • 772

                                              #92
                                              Several recent good MDT public service posts. Part of some sort of court ordered community service sentence? I've heard that political incorrectness is dealt with harshly in Canada.

                                              Some people understand the point of the thought experiment. Those that don't are probably beyond help anyway.

                                              MDT posts are so good that the mistakes are valuable. This doesn't seem to be very relevant to NFL preseason teasers, but provides something interesting anyway.

                                              I don't have an NFL database, but I just picked a random NFL Sunday (January 1, 2012).

                                              Steelers -7 w/ total 33, ML -325
                                              Saints -7 w/ total 54.5, ML -300
                                              Eagles -7 w/ total 44, ML -320
                                              hint - something about converting pointspread to win percent and the SBR calculator.
                                              Comment
                                              • allin1
                                                SBR MVP
                                                • 11-07-11
                                                • 4555

                                                #93
                                                Originally posted by Wrecktangle
                                                MDC = MF = Hutinnis
                                                Comment
                                                • Spektre
                                                  SBR High Roller
                                                  • 02-28-10
                                                  • 184

                                                  #94
                                                  Originally posted by hutennis
                                                  Thinking helps.

                                                  When p=0.028 the coins have a fair chance to be identical and your knowledge about coins being identical
                                                  can very well be fair.

                                                  If you "know" that coins are identical and one of them comes with p=0.00001 you should examine your "knowledge" pretty fricking hard.

                                                  Agreed.
                                                  Comment
                                                  • Optional
                                                    Administrator
                                                    • 06-10-10
                                                    • 61356

                                                    #95
                                                    Originally posted by hutennis
                                                    It is like some kind of weird joke.
                                                    Yes. Some of these TT guys take a perverse pleasure in driving souls like you crazy.

                                                    I think the thought game was an analogy about the way you look at anyone new who walks in here with a model Mathy. And the game was to catch someone like hutennis who wanted to argue a literal interpretation and taunt them.

                                                    Do I win?
                                                    .
                                                    Comment
                                                    • Optional
                                                      Administrator
                                                      • 06-10-10
                                                      • 61356

                                                      #96
                                                      Although I only give you guys 5/10 this time.

                                                      It's a bit weak catching out a non-English speaker on a reading comprehension point. ;-)
                                                      Last edited by Optional; 08-31-12, 06:00 AM.
                                                      .
                                                      Comment
                                                      • mathdotcom
                                                        SBR Posting Legend
                                                        • 03-24-08
                                                        • 11689

                                                        #97
                                                        It's also weak for a non-English speaker to come in and argue his Engrish interpretation of my own question is the right one.
                                                        Comment
                                                        • Optional
                                                          Administrator
                                                          • 06-10-10
                                                          • 61356

                                                          #98
                                                          True too.
                                                          .
                                                          Comment
                                                          • cyberbabble
                                                            SBR Wise Guy
                                                            • 08-30-10
                                                            • 772

                                                            #99
                                                            The original post provides something useful to think about for new modelers.
                                                            In MDT vs hutennis, MDT seems to be playing some sort of meta-game with hutennis.
                                                            Like a poster building followers for his picks and then posting the opposite side of what he really likes to improve the odds before he bets.
                                                            Then again, maybe MDT makes so much from betting that it has become boring and the posts are for entertainment.
                                                            Comment
                                                            • mathdotcom
                                                              SBR Posting Legend
                                                              • 03-24-08
                                                              • 11689

                                                              #100
                                                              Originally posted by cyberbabble
                                                              The original post provides something useful to think about for new modelers.
                                                              In MDT vs hutennis, MDT seems to be playing some sort of meta-game with hutennis.
                                                              Like a poster building followers for his picks and then posting the opposite side of what he really likes to improve the odds before he bets.
                                                              Then again, maybe MDT makes so much from betting that it has become boring and the posts are for entertainment.
                                                              Yeah like SBR posters are going to move a line
                                                              Comment
                                                              • hutennis
                                                                SBR Wise Guy
                                                                • 07-11-10
                                                                • 847

                                                                #101
                                                                Originally posted by Optional
                                                                Yes. Some of these TT guys take a perverse pleasure in driving souls like you crazy.
                                                                They can also take a perverse pleasure of kissing my ass.
                                                                At least there is a chance this task can be accomplished.
                                                                Driving me crazy certainly can not.

                                                                And as far as comprehension.
                                                                What comprehensions? English?
                                                                Grammar? Who gives a shit about grammar?
                                                                Are we professional linguists forum?

                                                                I thought this is the place for sport speculators.
                                                                And as far as a comprehension of what is vitally important
                                                                for sport speculators most of the people who are passing ridicules judgments on me
                                                                are not even on the right continent they should be on. Never mind zip code or neighbourhood.
                                                                Really confused, ignorant and delusional bunch.
                                                                Last edited by hutennis; 08-31-12, 11:53 AM.
                                                                Comment
                                                                • cyberbabble
                                                                  SBR Wise Guy
                                                                  • 08-30-10
                                                                  • 772

                                                                  #102
                                                                  Now that we know what doesn't work, what do you do that does work? I'm not asking for any valuable, secret info. Just a general statement that arbs or whatever are the only thing that you have found that does work. Perhaps bonus whoring? Since I live in the U.S. neither one is practical for me, so I won't be stealing all of your +EV.

                                                                  I'm up around 3% on money at risk since I started trading sports with no losing week for almost a year now (about 1500 trades)
                                                                  Comment
                                                                  • mathdotcom
                                                                    SBR Posting Legend
                                                                    • 03-24-08
                                                                    • 11689

                                                                    #103
                                                                    Originally posted by mathdotcom
                                                                    Suppose I have 100 quarters in a jar and I wonder if they are unfair in the sense that they have an inherent tendency to land on heads or tails significantly more than 50% of the time. I know the coins are all identical. I hire 100 people to go home with one coin each, and ask them to flip their coin 200 times and record the results. I tell them to come back to me only if they think they have an unfair coin. Each of these 100 people does not know I have hired 99 others to do the same.

                                                                    Suppose one of the 100 people I hired, let's call him Mr.Ed, comes back to me and tells me his coin landed 130 heads and 70 tails.

                                                                    1. What is the probability in the head of Mr.Ed that he has an 'unfair' coin?
                                                                    2. What is the probability in my head that Mr.Ed's coin is unfair?

                                                                    I am not asking for answers. I want you to think about the answers in terms of relative values not absolute values.

                                                                    This public service announcement brought to you by MDC Industries.
                                                                    All things highlighted in red are things hutennis could not comprehend because, I guess, he is not a professional linguist.

                                                                    Example of speaking relatively: A is bigger than B
                                                                    Example of speaking in absolute terms: A is 2 and B is 1

                                                                    hutennis if I had to grade your answers I would give you about -400%
                                                                    Comment
                                                                    • mathdotcom
                                                                      SBR Posting Legend
                                                                      • 03-24-08
                                                                      • 11689

                                                                      #104
                                                                      Originally posted by cyberbabble
                                                                      Now that we know what doesn't work, what do you do that does work? I'm not asking for any valuable, secret info. Just a general statement that arbs or whatever are the only thing that you have found that does work. Perhaps bonus whoring? Since I live in the U.S. neither one is practical for me, so I won't be stealing all of your +EV.
                                                                      Jesus are you blind to 95% of the threads in here?
                                                                      Comment
                                                                      • cyberbabble
                                                                        SBR Wise Guy
                                                                        • 08-30-10
                                                                        • 772

                                                                        #105
                                                                        Jesus are you blind to 95% of the threads in here?
                                                                        No.
                                                                        Just giving hutennis an opening to make a decent post instead of his usual stuff.
                                                                        Comment
                                                                        SBR Contests
                                                                        Collapse
                                                                        Top-Rated US Sportsbooks
                                                                        Collapse
                                                                        Working...