+ Reply to Thread
Results 1 to 12 of 12
  1. #1
    Join Date
    Jul 2006
    Posts
    30

    Meaning and Origin of Kelly System of Betting

    Sometimes on the gaming forums, one sees the mention of "kelly betting" or the "kelly criterion"...

    I'm sure somewhere on the tech section of any one of the better gambling forums, a better account exists than this...

    ...but this is what I recall reading from the late Huey Mahl's Handicapping Journal known as "First Twenty"....

    ..in this first post, i'll explain what the Kelly Criterion actually represents to the bettor

    WHAT THE KELLY CRITERION ACTUALLY STATES:

    for the professional gambler wagering into a positive, quantifiable edge...

    ...it states the OPTIMAL bet size as a percentage of bankroll...

    ...for example, a player who, IN THE LONG RUN, could CONTINUOSLY pick at 57%(4 wins for every 3 losses) @-110.....

    ...should bet 9.7% of his available wad on each discreet(non-simultaneous) betting event.



    I believe it was Armani or Bad Co. at a different gambling forum who stated we all need our heads examined since 94% of all who bet sports are losers.

    Kelly Betting shows the truth of those words...for the winning picker can diminish profitability and even ruin himself by overbetting the percentage...in factttttt....

    UNDERBETTING by the same amount as one OVERBETS by shows the same diminished level of profitability....buttt...

    ...one(presumably those 6% Armani spoke of) increases one's bankroll safety....

    ..better to underbet, Kelly States...not better "profit"-wise, but "safety-wise"



    Kelly Betting will show you how to obtain optimal returns for various positive wagering proficiency levels.

    WINNING PROFICIENCY @ -110

    52.38% WINNING PROFICIENCY....... 0% OPTIMAL BET SIZE

    53.1%............................................. 1.5%

    53.85%............................................ 3.1%

    54.54%............................................ .4.6%

    55.55%............................................ .6.7%

    57%............................................... ....9.7%


    In the next post, i'll relate what Mahl stated about the Kelly Criterion...and his musings on the possible origins

  2. #2
    Join Date
    Jul 2006
    Posts
    30
    For those of you old enough to remember, Huey Mahl was one of the contemporaries of Mike Lee, Mort Olshan, Billy "The Greek" Vassal, Phil Steele, and Kelso Sturgeon...


    John Kelly was an electrical engineer from AT&T assigned to explain to his department's superiors a very technical subject pertaining to the distribution of electrical energy along power lines of varying efficiency...

    He used an analogy...

    He set up a hypothetical gambling shop where bets were past-posted a few minutes after baseball games had started...

    ...since the outcome of the games had not yet been determined... the bettors were working to some advantage with the inside info...

    ..the quirk is that the gambling analogy was perfect for gambling plays...

    Canadian statistician, William Ziemba, used Kelly to look for favorable payouts in SHOW POOLS on prohibitive horse-racing favorites during the later 8th and 9th races....

    ...for even-money payouts...the math is much simpler...you bet in % what your advantage actually is in %


    Mahl mused that at the time Kelly was composing his masterpiece...it had just been a short time since US Senator and losing VP candidate, Estes Kefauver(D-Tenn) had conducted hearings into gambling and organized crime...

    ...wow!...what a great way to come up with the perfect betting system for those fortunate few able to work a small positive edge over the house

  3. #3
    Join Date
    Aug 2006
    Posts
    15
    great post. it's refreshing to view an educated gambler's discourse.

    just thought that i would illustrate the math behind the kelly criterion for anyone interested....

    let's use the theoretical winning percentage expectancy value of 57%. this means that a gambler believes that he has a 57% chance of winning his wager. hence, he has a 43% chance of losing that same wager.
    let's assume that you are laying -110 to win 100.
    this means that, according to your positive expectancy value of 57%, after 100 wages you would win 57 and lose 43. adding the 10% juice for losing wagers would result in a value of minus 47.3 (43 x 1.1)
    57-47.3=9.7.
    kelly states that 9.7% is therefore the ammount of bankroll to risk.

    of course, using reduced vig will increase the amount of bankroll risk as it decreases the negative pull of a theoretical losing wager.

  4. #4
    Join Date
    Aug 2000
    Posts
    91
    A excellent book on the subject is Sportsbetting on the edge by John C Tarbet.

    I think it's virtually impossible to be successful long term betting sports without using some variation of the Kelly System.

    There's an old rule of thumb for sportsbetting money management. Particularly football betting. Which states that a player should wager 5% of their bankroll per play.

    This is arrived at simply using the Kelly Sytem theory that an average to good handicapper should hit 55% of his wagers making his bet size 5.5%.

    It's a good rule of thumb, particularly for weekend warriors or new bettors. Because if you expectation is anything less than 55% then it's hardly worth your time. But yet expecting to do much better than 55% would be unrealistic.

    No one hit's better than 60% consistently long term. NO ONE.

  5. #5
    Join Date
    Jun 1999
    Posts
    3,461
    Hey guys -

    since I saw the topic Kelly System, I dug up some classic past posts on the subject.

    Over the last decade, many subjects have been discussed and are always accessible via the forum search feature

    Thought these would fit in nice with the topic.

    Here are links to each part of the 7 part series of posts.


    Kelly System Part 1

    Kelly System Part 2

    Kelly System Part 3

    Kelly System Part 4

    Kelly System Part 5

    Kelly System Part 6

    Kelly System Part 7

  6. #6
    Join Date
    Jul 2006
    Posts
    30
    Internet Searching is a skill in itself...excellent work in finding those links....

    I'd too echo the sentiment of the poster who stated that 55% is the "holy grail" in terms of achievement.,..in terms of "is it worth the effort?"

    I'm more of a recreational positive expectation math novice... I'd just like to see how some of the "free pick" guys can do against 53.5%....

    ....which rounds out to 2.4%... a 40-unit bankroll.

    If you've got two games going off simultaneously...you'd be safe betting 1.4% on each game




    Quote Originally Posted by Jeff
    Hey guys -

    since I saw the topic Kelly System, I dug up some classic past posts on the subject.

    Over the last decade, many subjects have been discussed and are always accessible via the forum search feature

    Thought these would fit in nice with the topic.

    Here are links to each part of the 7 part series of posts.


    Kelly System Part 1

    Kelly System Part 2

    Kelly System Part 3

    Kelly System Part 4

    Kelly System Part 5

    Kelly System Part 6

    Kelly System Part 7

  7. #7
    Join Date
    Dec 2001
    Posts
    379
    Isn't the problem with the Kelly method the big 0-7 week? I mean, let's say you have a couple of great weeks back to back and now your unit is size is jacked way up.

    Now you fire away at the increased unit size and lay a big goose egg. Whereas with flat betting, you'd be better of, in this situation.

    I know it's not the way to think. That you're going to go 0-7. But as we all know, it happens.

  8. #8
    Join Date
    Aug 2006
    Posts
    15
    good point brink...0-7 can happen.

    fortunately, the odds of this occurrence are 1 in 128 based upon the average gambler hitting 50%. sharp guys should do better.

    recall that as your bankroll decreases, the amount of money wagered based upon the kelly criterion will decrease as well based upon a percentage of a smaller bankroll.

  9. #9
    Join Date
    Dec 2001
    Posts
    379
    I seem to find a way to defy the odds at least once every season and go 0-7

    But thanks for this thread guys. This is fascinating stuff.

  10. #10
    Join Date
    Jul 2006
    Posts
    30
    Quote Originally Posted by brink
    I seem to find a way to defy the odds at least once every season and go 0-7

    But thanks for this thread guys. This is fascinating stuff.

    ...you could also use a kelly variant to address your concerns...fixed units as a percentage of bankroll...

    i don't like to send you down that path, brink

    yes..your ambivalence about kelly would be right in that regard..

    IF YOU HAVE NOT QUANTIFIED AN EDGE OVER THE VIG.....

    THEN why not just divide your bankroll by 64 units...

    ...if you're betting two simultaneous games...divide your bankroll by 111 UNITS and bet A UNIT ON EACH GAME

  11. #11
    Join Date
    Aug 2006
    Posts
    13
    great thread. you generally need the discipline of a monk to apply this. thats why people lose. i use it in all sports except bases. its hard to back off and minimally increase my wager on a winning streak, though. i tend to greatly increase my wager when winning and greatly decrease it when losing. usually works out for me.kind of the criteria in spirit at least.

  12. #12
    Join Date
    Jul 2006
    Posts
    30
    Quote Originally Posted by dhamma
    great thread. you generally need the discipline of a monk to apply this. thats why people lose. i use it in all sports except bases. its hard to back off and minimally increase my wager on a winning streak, though. i tend to greatly increase my wager when winning and greatly decrease it when losing. usually works out for me.kind of the criteria in spirit at least.
    Sure thing, dhamma...it is tempting to bet more on a streak...

    Since alot of punters say divide bankroll by 40(2.5% which would approximately correspond to a 53.5% win rate....


    I think one on a hot streak could jump up a 1/3 percent for each win achieved beginning AT THE TWO PERCENT Plateau ...

    ...and drop bet size by .36% for each losss sustained...

    if one is truly better than 53.5% winners.... this bet sizing treatment should get you to the right level of Kelly rather quick...

    6 winners in a row would put you only at 4.5% bet size...which would correspond to a win rate of 54.5%.

    Not too many of the scamdicapper pay services have 54.5% lifetime w/l %

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts