_Punt_4THEE_Profit_ Posted September 8, 2011 Share Posted September 8, 2011 The Kelly system of betting. I think I have the jist of it but how would it work on something like this; http://forum.punterslounge.com/f19/over-0-5-goals-long-term-papertest-121190/ Can you show me the calculations?Cheers. Quote Link to comment Share on other sites More sharing options...
Shy10ck Posted September 8, 2011 Share Posted September 8, 2011 Re: Can someone explain to me... You need to estimate the winning probality to calculate Kelly stake. You only got odds and bank. The stake is relative to the value (odds x estimated probability) and to the size of the bank. From Wiki where: f* is the fraction of the current bankroll to wager; b is the net odds received on the wager ("b to 1"); that is, you could win $b (plus the $1 wagered) for a $1 bet p is the probability of winning; q is the probability of losing, which is 1 − p. Quote Link to comment Share on other sites More sharing options...
_Punt_4THEE_Profit_ Posted September 8, 2011 Author Share Posted September 8, 2011 Re: Can someone explain to me... Yes but could you accurately use a made up percentage for winning such as 80%? Quote Link to comment Share on other sites More sharing options...
Shy10ck Posted September 8, 2011 Share Posted September 8, 2011 Re: Can someone explain to me... Yes but could you accurately use a made up percentage for winning such as 80%? No, not really. The idea of Kelly staking is to make your bank grow most optimal and that depends on value bets. If you just make up a percentage, you cant be sure of value. No value = you loose. Quote Link to comment Share on other sites More sharing options...
_Punt_4THEE_Profit_ Posted September 8, 2011 Author Share Posted September 8, 2011 Re: Can someone explain to me... Ok can someone verify this for me; For example in my thread linked above I have a treble with odds of 1.19 and a starting bank of £10. (1.19)-(0.20)/(1.19)= 4.95% 4.95% of £10 = 50p That is assuming I have winning probability of 80%. So because the Kelly strategy is actually aggressive and it comes out with a 50p stake my £1 is wrong for the long term and also you are supposed to use a conservative estimate for the probability of winning? Anyone shed any more light on this for me? Quote Link to comment Share on other sites More sharing options...
_Punt_4THEE_Profit_ Posted September 8, 2011 Author Share Posted September 8, 2011 Re: Can someone explain to me... Sorry had the above typed out before I got your reply, so where exactly can you use Kelly in sports betting? Quote Link to comment Share on other sites More sharing options...
Shy10ck Posted September 8, 2011 Share Posted September 8, 2011 Re: Can someone explain to me... You bet against value, 80 x 1,19 = 0,952 You need more than odds 1,25 if win percentage is 80 But (1,19x0,8)-(0,2)/1,19 = 0,63% 0,63% x 10 = 0,063 Quote Link to comment Share on other sites More sharing options...
_Punt_4THEE_Profit_ Posted September 8, 2011 Author Share Posted September 8, 2011 Re: Can someone explain to me... You bet against value, 80 x 1,19 = 0,952 You need more than odds 1,25 if win percentage is 80 But (1,19x0,8)-(0,2)/1,19 = 0,63% 0,63% x 10 = 0,063 Sorry to be a pain mate but can you explain this for me again/ in more detail. You won't believe me when I say I have an A-Level in statistics now lol Quote Link to comment Share on other sites More sharing options...
Shy10ck Posted September 8, 2011 Share Posted September 8, 2011 Re: Can someone explain to me... This is your (1.19)-(0.20)/(1.19)= 4.95% This is mine (1,19x0,8)-(0,2)/1,19 = 0,63% First of all you calculated wrong, (1.19)-(0.20)/(1.19)= 0,83 Second you used a wrong formula :D Quote Link to comment Share on other sites More sharing options...
_Punt_4THEE_Profit_ Posted September 9, 2011 Author Share Posted September 9, 2011 Re: Can someone explain to me... Oh yes of course bp means b.p or b x p ...clearly been a while since I have done this! So can you explain to me what this figure gives me in terms of my GH? Is it telling me I should stake 7p per bet if my bank is £10? Ideally I want odds of 1.10 and I reckon with odds near that success rate should be in the 98% mark (Ignore my selections for this week as it is a treble). Also if you say my bank is just £1 does this produce a near 100% stake each time? To be honest I don't really mind as I will probably be doing this I was just wondering how to relate this to my GH if you could do a calculation for Kelly and my GH I would appreciate it though. Quote Link to comment Share on other sites More sharing options...
froment Posted September 9, 2011 Share Posted September 9, 2011 Re: Can someone explain to me... Punt, Shy10ck wanted to say that, at odds of 1.19 and probability of 80% you mentioned, Kelly says it's a no-bet match for you. Kelly is based on edge you (believe to) have over bookie; you must get the odds better than they should be in your opinion; and bigger the edge, higher the percent of the bank you stake. In your case, with probability of 80%, odds are 100/80 = 1.25; but you bet on odds of 1.19, you accept odds lower than they should be by your calculation, which means you don't have edge, and therefore Kelly is unapplicable. At those odds, your selection must win at least 100/1.19 = 84.03 % of time for you to have profit; with 80% strike rate, you'll lose in long run. If you could get, for example, 85% of success at those odds, i.e., if you had 85% confidence in your picks, then Kelly would advise you to stake: (0.85 * (0.19 + 1) - 1) / 0.19 = 6.05 % of your bank. However, if you succeeded to chose your picks with, say, 90% confidence, then stake advised by Kelly would be: (0.90 * (0.19 + 1) - 1) / 0.19 = 37.4 % of the bank. You see, stake largely depends on your ability to properly evaluate fair odds, i.e., your confidence in your picks; that's why it's overcomplicated for me and I don't like it. Note definition of b provided by Shy10ck: b is the net odds received on the wager ("b to 1"); that is, you could win $b (plus the $1 wagered) for a $1 bet So, it's not decimal odds, but (decimal odds - 1), or fractional odds (simply divide the fraction) - that's where 0.19 comes instead of 1.19. Ideally I want odds of 1.10 and I reckon with odds near that success rate should be in the 98% mark So, it's ( 0.98 * (0.10 + 1) - 1) / 0.10 = 78 %. Very big percentage, as in that case you have a tremendous edge: fair odds are 100/98 = 1.02, and you get 1.10, or five times bigger profit. But if it's only 96%, then Kelly stake comes down to 56%; so are you sure it's 98%? That's problem with Kelly - only two matches out of 100 matches can bring it down from 98 to 96, and change stake for one third! Quote Link to comment Share on other sites More sharing options...
Shy10ck Posted September 9, 2011 Share Posted September 9, 2011 Re: Can someone explain to me... GH? It tells you to bet 0,63% of your bank. 10£ = 6,3p 1£ = 0,63P 1000£= 6,3£ But since 1,19 x 80 is under 100, you should not bet at all. Quote Link to comment Share on other sites More sharing options...
_Punt_4THEE_Profit_ Posted September 9, 2011 Author Share Posted September 9, 2011 Re: Can someone explain to me... Punt, Shy10ck wanted to say that, at odds of 1.19 and probability of 80% you mentioned, Kelly says it's a no-bet match for you. Kelly is based on edge you (believe to) have over bookie; you must get the odds better than they should be in your opinion; and bigger the edge, higher the percent of the bank you stake. In your case, with probability of 80%, odds are 100/80 = 1.25; but you bet on odds of 1.19, you accept odds lower than they should be by your calculation, which means you don't have edge, and therefore Kelly is unapplicable. At those odds, your selection must win at least 100/1.19 = 84.03 % of time for you to have profit; with 80% strike rate, you'll lose in long run. If you could get, for example, 85% of success at those odds, i.e., if you had 85% confidence in your picks, then Kelly would advise you to stake: (0.85 * (0.19 + 1) - 1) / 0.19 = 6.05 % of your bank. However, if you succeeded to chose your picks with, say, 90% confidence, then stake advised by Kelly would be: (0.90 * (0.19 + 1) - 1) / 0.19 = 37.4 % of the bank. You see, stake largely depends on your ability to properly evaluate fair odds, i.e., your confidence in your picks; that's why it's overcomplicated for me and I don't like it. Note definition of b provided by Shy10ck: So, it's not decimal odds, but (decimal odds - 1), or fractional odds (simply divide the fraction) - that's where 0.19 comes instead of 1.19. So, it's ( 0.98 * (0.10 + 1) - 1) / 0.10 = 78 %. Very big percentage, as in that case you have a tremendous edge: fair odds are 100/98 = 1.02, and you get 1.10, or five times bigger profit. But if it's only 96%, then Kelly stake comes down to 56%; so are you sure it's 98%? That's problem with Kelly - only two matches out of 100 matches can bring it down from 98 to 96, and change stake for one third! Ah right right cheers Shy10ck you were probably explaining this 100% but I just couldn't get it :\ But understand now with both your efforts :ok Was saying 98% to just see what the different outcomes would be but if you read the GH I talk of Martingale and recovering any losses it was then said ''2 losses in a row from 100 selections'' or to that effect which is where I got it really. Cheers for explaining this guys we got there finally! :rollin Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.