Why the Kelly Criterion is not a good bankroll management system
Collapse
X
-
Dark HorseSBR Posting Legend
- 12-14-05
- 13764
#36Comment -
eidolonSBR Hall of Famer
- 01-02-08
- 9531
#37I've made thousands upon thousands using Kelly. But i also lost my house.Comment -
BigCapSBR High Roller
- 02-10-08
- 189
#38
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.Comment -
RickySteveRestricted User
- 01-31-06
- 3415
#39I 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.Comment -
Flying DutchmanSBR MVP
- 05-17-09
- 2467
#41bumping this for all the Kelly threads by the pointy heads.
Kelly will burn your house down.
Comment -
roasthawgSBR MVP
- 11-09-07
- 2990
#42Kelly 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.Last edited by roasthawg; 12-06-09, 07:44 PM.Comment -
Flying DutchmanSBR MVP
- 05-17-09
- 2467
#43C'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.
Comment -
20Four7SBR Hall of Famer
- 04-08-07
- 6703
#44I 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 maxComment -
roasthawgSBR MVP
- 11-09-07
- 2990
#45I 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.Comment -
u21c3f6SBR Wise Guy
- 01-17-09
- 790
#46C'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.Comment -
Justin7SBR Hall of Famer
- 07-31-06
- 8577
#47C'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.
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.Comment -
Flying DutchmanSBR MVP
- 05-17-09
- 2467
#48People,
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.
Comment -
ThrempSBR MVP
- 07-23-07
- 2067
#49Your 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.Comment -
MonkeyF0ckerSBR Posting Legend
- 06-12-07
- 12144
#53C'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.Comment -
FishheadSBR Aristocracy
- 08-11-05
- 40179
#54I 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.......Comment -
TheLockSBR Posting Legend
- 04-06-08
- 14427
#56I 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.Last edited by TheLock; 12-07-09, 10:02 PM.Comment -
WrecktangleSBR MVP
- 03-01-09
- 1524
#572.5% of current.Comment -
donjuanSBR MVP
- 08-29-07
- 3993
#58Always current. Anything in the past other than any pending wagers is irrelevant.Comment -
FishheadSBR Aristocracy
- 08-11-05
- 40179
#59
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.Comment -
duritoSBR Posting Legend
- 07-03-06
- 13173
#60What is your kelly stake on an NCAAF game at +13 -110 that closes at +12.5 -105/-105Comment -
u21c3f6SBR Wise Guy
- 01-17-09
- 790
#61
Joe.Comment -
u21c3f6SBR Wise Guy
- 01-17-09
- 790
#62
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.Comment -
MrXSBR MVP
- 01-10-06
- 1540
#63Losing half of your bankroll, no matter your edge, is not an uncommon occurrence with full Kelly.Comment -
u21c3f6SBR Wise Guy
- 01-17-09
- 790
#64
Now, based on your scenario above, can you tell me which wager to make and what is the advantage?
Joe.Comment -
u21c3f6SBR Wise Guy
- 01-17-09
- 790
#65
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.Comment -
MrXSBR MVP
- 01-10-06
- 1540
#66Which 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 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.Comment -
u21c3f6SBR Wise Guy
- 01-17-09
- 790
#67I 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.
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.Comment -
donjuanSBR MVP
- 08-29-07
- 3993
#68
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.Comment -
MonkeyF0ckerSBR Posting Legend
- 06-12-07
- 12144
#69Right. 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?Comment -
TheLockSBR Posting Legend
- 04-06-08
- 14427
#70Thank you for the responses gentleman.Comment
SBR Contests
Collapse
Top-Rated US Sportsbooks
Collapse
#1 BetMGM
4.8/5 BetMGM Bonus Code
#2 FanDuel
4.8/5 FanDuel Promo Code
#3 Caesars
4.8/5 Caesars Promo Code
#4 DraftKings
4.7/5 DraftKings Promo Code
#5 Fanatics
#6 bet365
4.7/5 bet365 Bonus Code
#7 Hard Rock
4.1/5 Hard Rock Bet Promo Code
#8 BetRivers
4.1/5 BetRivers Bonus Code