Kelly with confidence

**Wrecktangle** · 01-21-10, 09:22 AM

Been in and out on Kelly. Not sure what the right answer is, but here's what I use:

Firstly, you have to be religious in calcing your win%. Carry 30, 60, 90, and all game win loss averages per sport/technique. Sides and totals of same sport are different techniques. Use the lowest win% of your 30-60-90 game win%s as your "expected" win % for your Bayesian Prior horizon.

If you bet more than 1/2 Kelly you'll wreck your BR just due to variance. As Kelly assumes independent events and sports events are correlated to some extent (model bias can be involved), going over 1/2 is dangerous. For large BRs you'll probably want to be in the .1 range especially for simultaneous betting.

Your win% is critical. For Kelly to have any chance of working you must be very sure of that number.

**betso** · 01-21-10, 02:54 PM

Thanks Wrecktangle. I totally agree with you that sides and totals differ and also between sports.

I am not so worried about the kelly fraction being to high, rather I am more concerned about relative bet sizes. E.g. say I have two angles applied to different sports

a, Have produced 5% edge and I am 90% confident it has a positive expected return.
b, Have produced 3% edge and I am 95% confident that it has a positive expected return.

How much should I bet (if anything) on b relative to a? And if possible how could this be tied back to a kelly fraction (because kelly gives you a definitive answer if you are 100% confident about your edge if I am not misstaken)?

**roasthawg** · 01-21-10, 03:45 PM

Bet a set PERCENTAGE of your bankroll on every play... 1/40th is what I use. Estimating winning percentages in sports gambling is very difficult... why not take the guess work out of the equation?

**icsky3** · 01-21-10, 04:10 PM

Originally posted by roasthawg

Bet a set PERCENTAGE of your bankroll on every play... 1/40th is what I use. Estimating winning percentages in sports gambling is very difficult... why not take the guess work out of the equation?

yeahh roast im right there with ya brother.

**gLonghorn** · 01-21-10, 05:17 PM

roasthawg,

i agree that certainty of expected win% is a fools errand and i understand your simplified 1/40 decision. at the same time, i don't like the 1/40th approach because it doesn't consider the relative merit of each opportunity.

if you pick totals at 56% and lines at 54%, there is tremendous value in the long term of having wagered a larger unit size on the totals. it is an easy cop out to just use 1/40th, but part of maximizing profit is not taking the easy way out.

fwiw, i understand that it's impossible to know your true expected win%, but understanding your relative win% (sides vs totals or NFL vs CBB or whatever) is not hard to understand, and that is enough to make this worthwhile.

**Dunder** · 01-21-10, 07:38 PM

Originally posted by betso

I want to bet a kelly fraction of my bankroll but I have some doubts about my edge.

Say I am only 90% confident I have a positive edge and that my estimated edge to date is 5%.

How should I size my bets under these circumstances? It would be nice to have some adjustment factor (so that the bet ranges from full kelly to 0) depending on how certain one is regarding the edge.

Your thoughts are appreciated.

If you have a high degree of confidence on your EV calculations, then go with something around one third to one half-Kelly would be my advice.

If your degree of confidence is lower such that your EV is basically an educated guess, that flat stakes would be more appropriate.

Keep records exactingly to establish whether your EV calculations are working out or not. If so a move up to full Kelly is the way to go - if that fits your appetite for risk.

**JohnAnthony** · 01-21-10, 08:58 PM

Originally posted by gLonghorn

roasthawg,

i agree that certainty of expected win% is a fools errand and i understand your simplified 1/40 decision. at the same time, i don't like the 1/40th approach because it doesn't consider the relative merit of each opportunity.

if you pick totals at 56% and lines at 54%, there is tremendous value in the long term of having wagered a larger unit size on the totals. it is an easy cop out to just use 1/40th, but part of maximizing profit is not taking the easy way out.

fwiw, i understand that it's impossible to know your true expected win%, but understanding your relative win% (sides vs totals or NFL vs CBB or whatever) is not hard to understand, and that is enough to make this worthwhile.

Referring to the emphasized line, why not try to bet in units

1 unit - 1/40th..
2 units - 1/20th..

..and so on?

**Wrecktangle** · 01-22-10, 07:42 PM

For the average guy, the Century method works pretty well. Basically, 1 unit is 1%, 2% = units, etc. I've write up on it but since I can't get PMs sort of out of luck.

**Desert Tortoise** · 01-23-10, 02:39 AM

There are mathematical articles written on this subject. Look up Bayesian Kelly Criterion.

**Dark Horse** · 01-23-10, 02:46 AM

Decided on this some years ago:

Kelly can produce astronomical bet sizes that many aren’t comfortable with. There is such a
thing as peace of mind.

Playing around with the numbers and my own comfort level I came up with these personalized
values. I would -conservatively- set 60% as the maximum attainable winning expectation for any
system (even though that is not true), and bet 1/3 Kelly. That would mean that, at -105 odds, any
betting system of 60% or higher would automatically result in a 6% bet size. If I never bet
anything with a lower winning expectation than 54%, and always bet at -105 odds, that would
give me seven bet sizes:

54% - 1.89%
55% - 2.58%
56% - 3.26%
57% - 3.94%
58% - 4.63%
59% - 5.31%
60% and above - 5.99%

The difference per percentage point is roughly 0.7% of bankroll:
6.0 - 5.3 - 4.6 - 3.9 - 3.2 - 2.5 - 1.8.

Err on the side of caution. If my expectation was between 58% and 57%, I would take the lower of the two.

For comparison, for full Kelly a 54% winning expectation would produce a 5.68% (!) bet size
(with -105 odds). That may work for some, but not for me.

======================= ===============

If my maximum bet size was not 6% but 5% of bankroll, then I would simply use 2.0 - 2.5 - 3.0 -
3.5 - 4.0 - 4.5 - and 5.0 % of bankroll for winning expectations 54% through 60+%.

**AliceCooper** · 01-23-10, 09:56 AM

Originally posted by gLonghorn

roasthawg,

i agree that certainty of expected win% is a fools errand and i understand your simplified 1/40 decision. at the same time, i don't like the 1/40th approach because it doesn't consider the relative merit of each opportunity.

if you pick totals at 56% and lines at 54%, there is tremendous value in the long term of having wagered a larger unit size on the totals. it is an easy cop out to just use 1/40th, but part of maximizing profit is not taking the easy way out.

fwiw, i understand that it's impossible to know your true expected win%, but understanding your relative win% (sides vs totals or NFL vs CBB or whatever) is not hard to understand, and that is enough to make this worthwhile.

you make a good point, being a math teacher myself, I actually like the simpler approach. the fact is that it is impossible to know the true expected win % (yes, we can make good educated guesses), and your way might be slightly better, but in the scheme of things, I think it is largely irrelevant. For one thing, one could probably make educated guesses as to modifying the 1/40 rule (which I primarily go by and with the assumption that I have a 55% chance of winning any even money bet).

assume for a sec,you have a 60% chance of winning a bet, and you would like to know in theory how much of your bankroll to bet.if you adjust to say 3%, you are probably pretty close to what any mathematics would tell you.

the way I look at it life is so dynamic that the 1/40 rule is somewhat irrelevant i.e., you want to buy a car or house, so you dip into your bankroll. now things have changed. remember, what Shakespeare said about "the best laid plans of mice and men"....in this case, just keep it simple.

i

**Ruifgalmeida** · 01-23-10, 10:02 AM

never bet more that 10%,if you cant calculate your edge just go flat betting 1% and your bankroll will thank you.sports betting is a marathon not a sprint

**roasthawg** · 01-23-10, 07:44 PM

Originally posted by Ruifgalmeida

sports betting is a marathon not a sprint

Couldn't agree more with this line... I would give anything to take the ups and the downs out of sportsbetting and just watch my bankroll gradually increase. In reality it doesn't work like that... there'll be stretches where you can't lose followed by stretches where you can't win. The highs and the lows will take you for a ride in the beginning until you come to expect it... and even then the lows still SUCK.

**DOMINATER** · 01-29-10, 06:48 AM

Great information how a system actually works