Why the Kelly Criterion is not a good bankroll management system

[Dark Horse]

If I am 99% confident that my range of picks falls between 51.5% and 57%, how does that help me kelly-stake my bets? How can anybody say they pick 53.1%? That is ridiculous.

It doesn't help you. The range you give is a classic confirmation of Wrecktangle's suggestion that Kelly isn't realistic for most players.

You can develop a precise understanding of how each of your angles produces in the long run. But it will take years.

[eidolon]

I've made thousands upon thousands using Kelly. But i also lost my house.

[BigCap]

In reverse order- you didn't understand it- there are things that get stupid where you can place a max stake on a decent roll. Props is an example everybody understands, but it comes with the limit problem. Not everything has the limit problem.

I did understand - your specific text was in regards to "Props". The overwhelming majority of the time prop bets will cap out for any serious bettor's bankroll. This is why it adds basically nothing to the Kelly conversation and the max bet is simply correct. Enough said.

Nobody can be 100% confident that they pick 53.1% (well, that they will continue to pick 53.1% going forward). You assign 99% of the probability inside 51.5 and 57. How much do you assign to each? How about between 51.5 and 52% between 52% and 53%? Etc. If you do those numbers, you can figure it out.

This does not get the bettor any closer to a usable kelly fraction when making a wager. Either they will be likely over-staking, or under-staking. This is the crux of the OP. Rarely will they be staking anwhere close to correct, e.g. complex blackjack counting.

[RickySteve]

I find the whole premise of maximizing bankroll via sportsbetting ridiculous due to the shallowness and the shadiness of that market. Kelly is a fine theoretical framework that has no application in sportsbetting as in an investment venue.

On a side note, it has been discussed many times in this forum that most people, including Kelly followers, do not realize that their Kelly bankroll is two to four orders of magnitude larger than what they think it is.

You know a lot of multi-millionaires gambling with an edge on a $500 bankroll, do you?

[Data]

You know a lot of multi-millionaires gambling with an edge on a $500 bankroll, do you?

I neither know a lot of multi-millionaires nor how is this relevant to what I said.

[Flying Dutchman]

bumping this for all the Kelly threads by the pointy heads.

Kelly will burn your house down.

[roasthawg]

Kelly is beyond reproach... the problem comes from quantifying your edge. Picking 53% winners is easier imo than actually quantifying your edge in any given game. This is what makes kelly not as successful a tool for sports betting than for other endeavors where the edge is a known.

The main thing to take from Kelly I think is that you should bet a percentage of your bankroll rather than a set unit size.

[Flying Dutchman]

C'mon, nobody can do Kelly in sports.

Name one guy who provably did it.

You gotta a lotta guys who did it in BJ where you can quantify your edge with razor precision, but nobody is good enough to do it in sports.

NOBODY.

So Kelly is a dead issue in sports betting.

[20Four7]

I utilize kelly in sports betting but I'm also conservative in my estimates. So in the end I"m probably doing 1/4 to 1/2 kelly max

[roasthawg]

I just use a set percentage of my bankroll for all bets. 2.5% per play works great for me... still risky as I make a ton of plays but it's also large enough for me to double my roll every 6 months if things are going well.

The main thing to take from Kelly imo is that you should be wagering a percentage of your bankroll if you are a +ev bettor rather than a standard unit size. With the standard unit you miss out on the compounding factor.

[u21c3f6]

C'mon, nobody can do Kelly in sports.

Name one guy who provably did it.

You gotta a lotta guys who did it in BJ where you can quantify your edge with razor precision, but nobody is good enough to do it in sports.

NOBODY.

So Kelly is a dead issue in sports betting.

Kelly is a tool and its concept can absolutely be applied to sports betting. I gave a brief description of how I use Kelly in post #17 in this thread.

Joe.

[Justin7]

C'mon, nobody can do Kelly in sports.

Name one guy who provably did it.

You gotta a lotta guys who did it in BJ where you can quantify your edge with razor precision, but nobody is good enough to do it in sports.

NOBODY.

So Kelly is a dead issue in sports betting.

I use Kelly.

The market price is X -110/-110 on both sides. I have the opportunity to take X +110. Assuming market terror is accurate, I'm confident I have about a 4.5% edge (vs a -100 no vig line). I might use half-kelly or some other fraction, but I still use it as a starting point.

[Flying Dutchman]

People,

Dr Kelly did not say that 1/2 Kelly or 1/4 Kelly was optimal. He proved Full Kelly.

Everything else is SUB-optimal.

Show me someone who is using (full) Kelly.

[Thremp]

Your BJ example is terrible. Counts are approximations. You could have a count that would show an edge whereas the actual value of the shoe is -EV.

[reno cool]

People,

Dr Kelly did not say that 1/2 Kelly or 1/4 Kelly was optimal. He proved Full Kelly.

Everything else is SUB-optimal.

Show me someone who is using (full) Kelly.

Yes tell em Dutchman

[reno cool]

Your BJ example is terrible. Counts are approximations. You could have a count that would show an edge whereas the actual value of the shoe is -EV.

what the hell are you talking about?

[Fishhead]

what the hell are you talking about?

I'm glad you asked first..............LOL.

[MonkeyF0cker]

C'mon, nobody can do Kelly in sports.

Name one guy who provably did it.

You gotta a lotta guys who did it in BJ where you can quantify your edge with razor precision, but nobody is good enough to do it in sports.

NOBODY.

So Kelly is a dead issue in sports betting.

Billy Walters and the Brain Trust did and continues to utilize Kelly.

[Fishhead]

I just use a set percentage of my bankroll for all bets. 2.5% per play works great for me... still risky as I make a ton of plays but it's also large enough for me to double my roll every 6 months if things are going well.

The main thing to take from Kelly imo is that you should be wagering a percentage of your bankroll if you are a +ev bettor rather than a standard unit size. With the standard unit you miss out on the compounding factor.

..........you may want to set a predetermined date to reevaluate your Kelly gameplan strategy...........say every new year/season.

Continued success.......

[Wrecktangle]

Billy Walters and the Brain Trust did and continues to utilize Kelly.

I have it on decent authority that BW uses fractional Kelly, as full Kelly is just too risky.

[TheLock]

I just wanted to say that I really enjoyed the first page of this thread. Good stuff.


I have a nOObish question though. Lets say I'm currently not using KC but I'm flat betting 2.5% of my BR and it is slowly (obviously) growing. I would continue to wager 2.5% of my current BR.

But what if a starting BR of $2000 was currently at $1800? Should I be risking 2.5% of the current BR or the starting BR?

I realize this sounds (and is) retarded to many of you but what is more "optimal". I am pretty new to this and I don't have anyone to ask.

[Wrecktangle]

2.5% of current.

[donjuan]

Always current. Anything in the past other than any pending wagers is irrelevant.

[Fishhead]

Always current. Anything in the past other than any pending wagers is irrelevant.

Exactly.........

And again, set a timetable as to when to re-evaluate your base percentage.

Treat your sports investments as you would a growth fund........and stay on course and remain disciplined.

[durito]

Exactly.........

And again, set a timetable as to when to re-evaluate your base percentage.

Treat your sports investments as you would a growth fund........and stay on course and remain disciplined.

What is your kelly stake on an NCAAF game at +13 -110 that closes at +12.5 -105/-105

[u21c3f6]

I've made thousands upon thousands using Kelly. But i also lost my house.

You cannot lose your house using Kelly unless you are just not paying attention. Remember, with Kelly your wagers go down in size if you are losing money. Long before one has lost at most half of their bankroll they should be reevaluating if in fact they have a valid edge and/or if they over-estimated their edge (which is one good reason to use half-Kelly).

Joe.

[u21c3f6]

People,

Dr Kelly did not say that 1/2 Kelly or 1/4 Kelly was optimal. He proved Full Kelly.

Everything else is SUB-optimal.

Show me someone who is using (full) Kelly.

Do you agree that a fractional Kelly is a valid way to size one's wagers? If you agree with this, then we really don't have any issue. It's not that it is impossible to use full-Kelly, its just that there are many good reasons why one should not use full-Kelly.

However, if you (or others) are saying that one should not look at Kelly at all, then how do you go about determining wager size?

Joe.

[MrX]

Long before one has lost at most half of their bankroll they should be reevaluating if in fact they have a valid edge and/or if they over-estimated their edge (which is one good reason to use half-Kelly).

Joe.

Losing half of your bankroll, no matter your edge, is not an uncommon occurrence with full Kelly.

[u21c3f6]

What is your kelly stake on an NCAAF game at +13 -110 that closes at +12.5 -105/-105

For myself, my "Kelly" wager would be based on my data for this type of wager and not based on some "supposed" advantage based on line movement. I am not saying that there isn't anyone that can do this, I am just saying that my data has to "prove" to me that there really is an advantage before I would wager on it.

Now, based on your scenario above, can you tell me which wager to make and what is the advantage?

Joe.

[u21c3f6]

Losing half of your bankroll, no matter your edge, is not an uncommon occurrence with full Kelly.

Which is a good reason to use a fractional Kelly.

Because of the possible large fluctuations with full-Kelly I use a half-Kelly but in addition I also cut down on volatility by constructing strategies that have a higher % of success. I prefer to wager on strategies that have a higher win % even if they have a smaller edge. Most times despite the smaller edge you will actually be able to be more profitable because you can wager a larger % of your bankroll on higher win rate strategies. Because of the higher win rate, if your strategy is not valid, it will appear much faster and long before you would lose half your bankroll.

Joe.

[MrX]

Which is a good reason to use a fractional Kelly.

Because of the possible large fluctuations with full-Kelly I use a half-Kelly but in addition I also cut down on volatility by constructing strategies that have a higher % of success. I prefer to wager on strategies that have a higher win % even if they have a smaller edge. Most times despite the smaller edge you will actually be able to be more profitable because you can wager a larger % of your bankroll on higher win rate strategies. Because of the higher win rate, if your strategy is not valid, it will appear much faster and long before you would lose half your bankroll.

Joe.

I agree with all of that. The post I quoted seemed to be in response to a full Kelly scenario.

I still think your assertion about reevaluating "long before" losing half of your bankroll is overstated, even for half-Kelly. If you take 10 people with an accurately predicted edge, all employing half-Kelly, chances are at least one of them will find themselves in the neighborhood of losing half his bankroll before things get rolling.

[u21c3f6]

I agree with all of that. The post I quoted seemed to be in response to a full Kelly scenario.

I still think your assertion about reevaluating "long before" losing half of your bankroll is overstated, even for half-Kelly. If you take 10 people with an accurately predicted edge, all employing half-Kelly, chances are at least one of them will find themselves in the neighborhood of losing half his bankroll before things get rolling.

Yes, you are correct and I agree with you that my point is somewhat overstated. I guess my point was directed at the "losing one's house" scenario. Certainly well before anyone should be close to losing their house there would definitely be some indication that a reevaluation was necessary.

Any tool used blindly and/or foolishly can be dangerous. I just wanted to get across that there is something of value that one can learn from Kelly even if one does not use full-Kelly or Kelly at all. It does not have to be all-in or nothing.

Joe.

[donjuan]

For myself, my "Kelly" wager would be based on my data for this type of wager and not based on some "supposed" advantage based on line movement. I am not saying that there isn't anyone that can do this, I am just saying that my data has to "prove" to me that there really is an advantage before I would wager on it.

Now, based on your scenario above, can you tell me which wager to make and what is the advantage?

Joe.

0 and -.

[MonkeyF0cker]

I have it on decent authority that BW uses fractional Kelly, as full Kelly is just too risky.

Right. But this thread is discussing the merits of Kelly. To completely discount it simply because most are not using full Kelly but rather fractional Kelly is a bit absurd. The fact remains that even fractional Kelly is far more optimal than simple flat betting or any other staking strategy that has been discussed here. What other staking strategies have been proposed that are as effective? If quantification is the issue here, how are you sure that you have an edge or that you aren't overbetting your edge when you are flat betting?

[TheLock]

Thank you for the responses gentleman.