[donjuan]

Did I read this right? One coin flip at -110/+150 odds? Clearly, you would not risk 50% of your bankroll on either outcome, when you can guarantee yourself the easy profit. When people are giving away money there is no need to bet.

This isn't about your random musings. This is about the highest EG scenario.

[donjuan]

Huh...interesting. Thanks for that bit DH...Guess no one here knows wtf they are talking about.

Im not mad at Juan, he just needs to learn to read, as the thing I was initially talking about was the 25% br coin flip at +100 where the guy was NOT talking about betting two sides. Then wiffle chimed in saying something similar with the 50% and I thought he was saying the same thing the other guy said. And no, none of the websites I've ever read on Kelly have mentioned betting -EV senarios. And yes, Juan, I have read a LOT of Ganch's posts, thanks.

DH has no clue about Kelly or really anything. His personal risk aversion ≠ Kelly.

Ask him about streaks. Your head will explode.

Do the math. The highest EG is from betting 50% on both.

[wantitall4moi]

DH has no clue about Kelly or really anything. His personal risk aversion ≠ Kelly.

Ask him about streaks. Your head will explode.

Do the math. The highest EG is from betting 50% on both.

Thats what I was trying to get at as well, but I get caught up in the disdain I dont always finish my thoughts on it.

kelly is still a formula and it goes both ways so all you need to do is plug in the numbers into the formula. So it is the same as 1+1=2 and 2-1 =1, it just has a couple other extra parts to it. The place where people make mistakes is with probability, but if probability is a known and undisputed fact then the results should be the same for everyone. The other part where people can differ is pay out odds, but that isnt a mistake simply a result of getting a different price on your personal bet/wager.

[Dark Horse]

This isn't about your random musings. This is about the highest EG scenario.

Musings. lol How about reality versus dogma?

Single coin flip. 100K bankroll, no risk. Only a fool would not take the free 8% profit. If the coin would be flipped many times, that would be a different story, but you never mentioned anything beyond a single coin flip. Perhaps you didn't express yourself clearly.

Ask him about streaks. Your head will explode.

I haven't talked of streaks here in years. But I'm glad you still remember. As to the image of your head exploding, not very hard to imagine. Careful with that one.

[donjuan]

Musings. lol How about reality versus dogma?

Single coin flip. 100K bankroll, no risk. Only a fool would not take the free 8% profit. If the coin would be flipped many times, that would be a different story, but you never mentioned anything beyond a single coin flip. Perhaps you didn't express yourself clearly.

Please try to keep up. The scenario was directly in response to the statement that full Kelly says you would never bet 50% of your roll on a -110 bet. Again, your personal risk preferences have nothing to do with the highest EG scenario and you only serve to confuse others by saying they do.

[HeeeHAWWWW]

Single coin flip. 100K bankroll, no risk. Only a fool would not take the free 8% profit.

Totally depends on the wealther of the individual. Bill Gates wouldn't hedge.

[donjuan]

Totally depends on the wealther of the individual. Bill Gates wouldn't hedge.

Since it's a Kelly discussion, the assumption is that it is a Kelly bankroll.

[wantitall4moi]

here lets try and all get on the same page, or maybe end this.....


http://www.sportsbookreview.com/betting-tools/kelly-calculator/

[tomcowley]

Musings. lol How about reality versus dogma?

Single coin flip. 100K bankroll, no risk. Only a fool would not take the free 8% profit. If the coin would be flipped many times, that would be a different story, but you never mentioned anything beyond a single coin flip. Perhaps you didn't express yourself clearly.

Only an idiot wouldn't withhold some amount of money from the arb and wager it at +10.2% ROI (instead of +8.2% of the arb).

[podonne]

yeah that was after about 12 or more hours of poker and at 430 am so sue me for being less than lucid. but if a guy goes 0-10 using kelly and he has the 'optimal' bet size going he is going to be down about 45% so the 5% mistake I made wouldnt even be worth mentioning.

Kelly doesnt work, end of story. It assumes way too much. And where one part of a formula is wrong it makes the whole formula invalid.

Even if a sports bet was posted at true odds a person would have difficulty in winning long term with kelly. With vig it is impossible to win with kelly, unless you assume your power at predicting winners outweighs whatever vig you have to pay. So then you have to determine your personal win probability versus a 'fair' line. So now you have two things to worry about, is the line fair and if so what are my chances to beat it? This is all underlined by risk of ruin by betting way too much or even way too little. Winners wont be big enough and losers will be too big, and you eventually go broke.

Even with a coin flip and following kelly it is possible to lose following a truly controlled experiment. where you know your 'true' odds and thus exactly what to bet. because even a coin flip is not mathematically guaranteed to follow a 'perfect' enough sequence to avoid going broke. Nor is it guaranteed to go 50/50 even after many many flips. And the truth is if kelly worked beyond a shadow of a doubt every person that followed it or subscribed to it would all be making the same exact bets for the same exact percentage of money because they would have it figure out.

I told the guy what worked for me, in reality. Not some formula that people have tried for ever to make work out in sports betting, and failed time after time to get to work. I bet sports for 15 years as my only source of income (with a little poker mixed in) and I did pretty well. Following what worked for me, things I learned from practical use and trial and error. I didnt have anything else to fall back on at one point, so if I went broke I was broke for real. Not like guys waiting for their next pay check or welfare check to come into reload. But that is another thing, most guys dont have a true bankroll they have an active amount of money they feel they can lose and not get hurt. So most guys bankrolls arent big enough in the first place. If you have a million bucks use it, dont put in 250K and call that a bankroll, go broke and reload with another 250K. So that is another fatal flaw, misidentifying what a bankroll is in the first place.

Did I do it 'perfectly', I doubt it, but I made a hell of a lot of money and lived a pretty good lifestyle and still do. So that is good enough for me. It was certainly better than every other person I know personally who bets on sports or plays cards. At least in the actual gambling facet of it.

When it all comes down to it the way to be successful at 'gambling' is to become the book maker. Which is the same thing as saying arbitrage your ass off with as much volume as you can get. And balance it out the best you can. Forget +ev or thinking you have an advantage with certain plays. Give me 8-10 cents on every bet and thats all I need. When you eliminate the need to win all the theories are moot anyway.

I'm not sure I follow all the math in these two posts, but I have to agree with wantitall4moi that Kelly is not appropriate for sports betting.

Kelly requires you to know your exact edge in any given betting situation, not an estimated edge based on your past results with this type of situation, which is all you can ever get with sports betting. Kelly works wonderfully only in mathmatically-defined games like a card counter in blackjack.

I would set a $$$ betting level based on a % of bankroll and keep this $$$ bet for a period of time, say a month. They readjust based on what your bankroll has done. That way you don't decrease your bet after each loss or increase after each win, but rather adjust your bet size over time based on your performance.

[Chipp]

I'm not sure I follow all the math in these two posts, but I have to agree with wantitall4moi that Kelly is not appropriate for sports betting.

Kelly requires you to know your exact edge in any given betting situation, not an estimated edge based on your past results with this type of situation, which is all you can ever get with sports betting. Kelly works wonderfully only in mathmatically-defined games like a card counter in blackjack.

Actually, I'd argue the two are almost exactly the same. Unless you're using computer software when playing BJ or have access to Rain Man, you're most likely using one of several card counting systems. These systems only provide estimates of the expected true deck probabilities, which in certain cases may be off substantially.

The same dynamic exists with sports betting. We don't know the exact probabilities of a given outcome occurring, but nevertheless, through a combination of quantitative and qualitative analysis we can still determine (hopefully) unbiased estimates of event win/push probabilities.

Compare the probabilities implied by whatever counting system you use with the true prior probabilities over a few hundred million deck runs, and what you'll probably see are that your system's predictive figures possess a similar degree of accuracy as that which might be expected of an advantage sports bettor.

...

[Darkside Magick]

kelly criterion =

[podonne]

Actually, I'd argue the two are almost exactly the same. Unless you're using computer software when playing BJ or have access to Rain Man, you're most likely using one of several card counting systems. These systems only provide estimates of the expected true deck probabilities, which in certain cases may be off substantially.

The same dynamic exists with sports betting. We don't know the exact probabilities of a given outcome occurring, but nevertheless, through a combination of quantitative and qualitative analysis we can still determine (hopefully) unbiased estimates of event win/push probabilities.

Compare the probabilities implied by whatever counting system you use with the true prior probabilities over a few hundred million deck runs, and what you'll probably see are that your system's predictive figures possess a similar degree of accuracy as that which might be expected of an advantage sports bettor

I was just attempting an illustration where the odds are known and fixed. An expert, knowing the number of cards in the deck and each of the cards that have been seen, could calculate an exact edge. A sports bettor can never have that level of knowledge.

Kelly needs the edge to be exact because its an equation. Sports bettors can only estimate an "uncertain" edge, a range, maybe a normal curve with a mean estimated_edge +/- 1%. The uncertainty has to be maintained throughout the equation and in the result, but no one does that, they just take thier "estimated edge" as an absolute number and finish the equation. The error rate skyrockets once you do that, such to be useless.

I'm not saying the concept behind Kelly can't be used, just that Kelly-as-we-know-it doesn't work when the edge is an uncertain value. How do I know? There is no dispersion variable (standard deviation, variance, skew, etc...) just a precise edge.

Maybe someone can tackle what the Kelly equation looks like when used with an uncertain edge and also how to evaluate a Kelly result of " bet 4.5% +/- 0.3453% of your bankroll" (you can't bet a range after all)...

[bztips]

I was just attempting an illustration where the odds are known and fixed. An expert, knowing the number of cards in the deck and each of the cards that have been seen, could calculate an exact edge. A sports bettor can never have that level of knowledge.

Kelly needs the edge to be exact because its an equation. Sports bettors can only estimate an "uncertain" edge, a range, maybe a normal curve with a mean estimated_edge +/- 1%. The uncertainty has to be maintained throughout the equation and in the result, but no one does that, they just take thier "estimated edge" as an absolute number and finish the equation. The error rate skyrockets once you do that, such to be useless.

I'm not saying the concept behind Kelly can't be used, just that Kelly-as-we-know-it doesn't work when the edge is an uncertain value. How do I know? There is no dispersion variable (standard deviation, variance, skew, etc...) just a precise edge.

Maybe someone can tackle what the Kelly equation looks like when used with an uncertain edge and also how to evaluate a Kelly result of " bet 4.5% +/- 0.3453% of your bankroll" (you can't bet a range after all)...

It would be nice if you provided even the slightest amount of logical and/or empirical evidence that Kelly is "useless" because your edge is uncertain; simply asserting that it's so doesn't make it true.

[probettor1]

Just saying:
1 Never bet flat betting. Different edge, different bet size.
2 Only bet large bets if you are a amateur gambler or a genius. If you are not a pro you will lose money on the long run for sure, but if you bet all your money in one bet then you have a 50-50 to win or lose; although you return might be slightly lower that you bet.
3 This is gambling, no one has the truth in his hand. Your next 7 picks might be all winning picks and your next 600 might be the worst streak you ever had.
4 Even though betting 1-2% sound great, when you have a small bankroll it might makes no sense. If you spend 8 hours handicapping to bet 50 bucks you better get a real job. And yes, betting most of your money in one bet sound crazy but that was what Billy Walker did when he started.

[p19101]

I have 2% flat bet unless it's a dog of +250 or more, then I bet 1% to reduce swings. I don't recalculate how much 2% is more than roughly every 3 months.

[wiffle]

I have 2% flat bet unless it's a dog of +250 or more, then I bet 1% to reduce swings. I don't recalculate how much 2% is more than roughly every 3 months.

cool story bro

[jolmscheid]

Do you re-calculate wager size daily?

[Derty D]

I believe in the process not the end result. Hence going with 1-2% for every unit wagered is a responsible money management theory. You will weather huge swings down in that case. Recalculating total bankroll once a week to either raise my 1-2% or lower my 1-2% is paramount.

Just as you would raise your 401k with every raise in salary, one should raise their stake.

[podonne]

It would be nice if you provided even the slightest amount of logical and/or empirical evidence that Kelly is "useless" because your edge is uncertain; simply asserting that it's so doesn't make it true.

You're absolutley correct, but it's not on me to prove that it is useless, it is on the Kelly backers to prove that is is useful.

Without knowing the precise numbers, but knowing how the equation works, there is a measure of dispersion in the estimated edge (how good it is) in which Kelly works better than flat or proportional betting. If the dispersion is 0 (every calculated edge is precisely exact), then Kelly is the optimal betting system. If dispersion is very large (you are just picking random values for your estimation of edge), then Kelly will be suboptimal. Somewhere in between is a magic value, but to my knowledge no one advocating Kelly has ever calculated it (essentially, how good you need to be at estimating your edge for Kelly to be better than flat or proportional betting).

I truley don't know that the magic value is, what to call it, or how to calculate it, but I know that its greater than zero and not everyone can calculate an edge with dispersion=0. Since its not proven that Kelly is better, it is not better. And since Kelly is known for much wilder swings than the other two, I would recommend a flat or fixed proportional system over Kelly.

[podonne]

I will also repeat this quote from Wikipedia:

... as Kelly's original paper demonstrates, the criterion is only valid when the investment or "game" is played many times over, with the same probability of winning or losing each time, and the same payout ratio.

Sports betting clearly does not meet any of these qualifications, with the exception of being played many times.

[byronbb]

I'm not sure I follow all the math in these two posts, but I have to agree with wantitall4moi that Kelly is not appropriate for sports betting.

Kelly requires you to know your exact edge in any given betting situation, not an estimated edge based on your past results with this type of situation, which is all you can ever get with sports betting.

????? If Pinnacle is closing an NFL game and taking 140k bets a side at -104/-104 it is not reasonable to assume that that game at that spread is a 50/50 game and that getting +104 would give you an advantage you can quantify??

[statictheory]

Without taking this way off track, you need to consider volume, if this hasn't been mentioned. Im not a pro but Ive bet
over 400 cbb and over 150 nfl games so far this year. If you can find yourself a method that allows you too increase the volume
instead of looking for those "" Super high probability"" bets only, the rest will start to fall into place.

[135steward]

I also have up to 40 simulataneous wagers open on a weekend because of the football calendar, so I guess I have to be extra cautious.

Betting all of your money on 40 bets is about the same as betting 2% on one bet. The only variable here is how much time you have to handicap.

[135steward]

.... and seriously if you were good enough to determine a 1% advantage and utilize it in the real world you would need to worry about betting with a formula anyway.

That's the simple truth, wantit. I drifted from the think tank because it seems more like the overthink tank. But for what its worth, I appreciate your practical insight. You must have a lot of grit, too, to keep coming back in spite of the disrespect so many smartasses heap your way.

[Inkwell77]

Good stuff

[marky85marky]

1% and no more. Anything else is too high. 90% of the time you can bet 1.5 or 2 but when you have a losing streak you'll go down hard. I followed a handicapping service for 2 years before finally getting money and giving it a proper go using 1.5%. They lost 60 units in a few months and I was gone and dusted. This is a service that's been around for more than 10 years but my poor money management cost me. If u can't afford downswings no more than 1%. If you want to maximize returns but are prepared to go go bust once every 5-10 years than go 2%.

[probettor1]

If you dont know what percent of your money to bet, then you probably wont even know what team to pick either. This is not about how much to bet, is about winning. As long as you are winning you can bet as much as you want but remember your chances of winning are just a 2% if you are a professional who has study this bussiness for years and happens to be very smart. So 2 dollars per bet will probably fit you well. I would bet a great amount of money that after 100 bets your balance is goint to be negative. Of course i might be wrong but I love stats and the stats are telling me you have over 98% of losing in the long run, so betting against you is a pretty good bet.

[Marginalis]

Kelly IF you can quantify your edge is the optimal staking amount period. Most utilize fractional kelly to reduce risk/overbetting. If you have no clue what your edge is then simply 1-2% of your bankroll per play would be optimal. You'll do better on average results flat betting.

LOL, You still broke?

[ODDmaster]

I'm beting 2.5-5%/per bet ,and mostly enough bets for all bankroll.