[tsty]

Thats not what he is saying

[Alfa1234]

Do you understand how stupid you sound?

If I lose 30% of my bankroll every day then I end up with 0%?

RoR on kelly is zero

No you'd end up with 24%. A quarter of the amount you started with. Obviously that makes it a much better idea.

What Semibluff said absolutely makes sense, you are extremely likely to lose a large portion of your initial roll by using full Kelly (assuming your edge is not huge). That means you'll have to reload to continue betting with some kind of a decent stake, because let's be honest, you will not have the mental discipline to continue betting with small bets to overcome the long downswing. Once you reload, you are no longer using full Kelly as your initial roll was only a portion of the actual roll you had at your disposal.

[danshan11]

to use full kelly you are using your complete bankroll and as your bankroll is dropping so is your bet size, now if you are using half of your assets and will add the other half later, you are betting half kelly but WHO cares Kelly sucks and yes I know people love it but there are better variations like the fixed profit betting. tsty I am not sure you have a good understanding of kelly by some of your comments which is a good thing because no one should be using it anyway!

[Barrakuda]

lol there is literally a plethora of papers out there dissecting kelly to it's core yet you guys keep posting this garbage

anyone who has read through those papers would realise that full kelly is the only way to go

kelly maximizes ev to it's fullest potential

Only if you know your precise edge with 100% certainty. That certainty cannot exist. Hence, full Kelly is always too aggressive in sports betting.

[Barrakuda]

I was not talking about edge

and finding 30% edge isn't that rare

finding a few every year isn't that uncommon
mayweather vs connor was a lot more than that

So what was the Mayweather edge precisely? If you cannot answer down to the tenth of a percent, full Kelly is not at all optimal. Kelly assumes the edge is known, not estimated.

[Barrakuda]

and to be honest on a staking method, I am firmly a believer in fixed profit
simply divide your roll into 100 pieces and place each bet to win 1 unit
-200 bet 2 to win 1
100 bet 1 to win 1
+200 bet .5 to win 1 and so on
its magic, you want to have more money on bets that are more likely to win. its an ugly cousin of kelly but if you backtest it can show an edge over flat and there is no way really to backtest kelly so who knows if it is better than kelly

I don't agree. You're saying that, assuming equal edge, you should risk $1,000 to win $1 on 1-1,000 favorite, and risk $0.001 to win $1 on a 1,000-1 dog? The dog bet will have zero impact on your bankroll, even if you make several thousand of them, whereas the favorite bet has the potential to crush you. In other words, if you had both bets on, you'd completely ignore the dog bet and worry solely about the favorite, which suggests the respective sizes of impact are not close to equal.

Do you have a sim showing that this staking strat maximizes reward/risk?

[tsty]

No you'd end up with 24%. A quarter of the amount you started with. Obviously that makes it a much better idea.

What Semibluff said absolutely makes sense, you are extremely likely to lose a large portion of your initial roll by using full Kelly (assuming your edge is not huge). That means you'll have to reload to continue betting with some kind of a decent stake, because let's be honest, you will not have the mental discipline to continue betting with small bets to overcome the long downswing. Once you reload, you are no longer using full Kelly as your initial roll was only a portion of the actual roll you had at your disposal.

No I wouldn't

I would always have 100% lol

just stop while you're ahead

I have not reloaded for over 5 years

[Bsims]

An ongoing debate, my bet sizing is better than yours. Since I'm working on a CBB system I have a file with 2017-18 bets with expected returns based on the model. So, I decided to try the four most touted approaches discussed. The first set of results with all the expected returns >$1.00. It's a losing set of data, hence all lost money. In this situation, both Kelly and half Kelly were disasters. I then filtered out all wagers with expected returns < $1.20, leaving a winning set of wagers. Here full Kelly managed to lose money while half Kelly showed a profit. But both were out performed by the flat betting systems. That's disappointing to me because I really like Kelly. So here are the results .

All bets with ExpRet >$1.00
Strategy	Games	Bet	Net	Ret/$	minBet	maxBet	avgBet	minKelly	maxKelly
Bet $1 on every pick	8,011	$8,011.00	-$92.33	$0.988	$1.00	$1.00	$1.00
Return $1 on every pick	8,011	$8,002.66	-$103.39	$0.987	$0.07	$2.95	$1.00
Full Kelly	7,814	$47,681.06	-$999.95	$0.979	$0.00	$994.43	$6.10	$0.02	$6,627.68
Half Kelly	7,814	$2,356.93	-$999.97	$0.576	$0.00	$86.23	$0.30	$0.03	$1,000.00
All bets with ExpRet >$1.20
Strategy	Games	Bet	Net	Ret/$	minBet	maxBet	avgBet	minKelly	maxKelly
Bet $1 on every pick	401	$401.00	$29.88	$1.075	$1.00	$1.00	$1.00
Return $1 on every pick	401	$186.34	$18.67	$1.100	$0.07	$1.15	$0.46
Full Kelly	374	$81,359.10	-$206.24	$0.997	$2.85	$3,324.64	$217.54	$48.11	$13,913.23
Half Kelly	374	$539.81	$16.37	$1.030	$0.05	$13.01	$1.44	$2.25	$108.59

[danshan11]

I don't agree. You're saying that, assuming equal edge, you should risk $1,000 to win $1 on 1-1,000 favorite, and risk $0.001 to win $1 on a 1,000-1 dog? The dog bet will have zero impact on your bankroll, even if you make several thousand of them, whereas the favorite bet has the potential to crush you. In other words, if you had both bets on, you'd completely ignore the dog bet and worry solely about the favorite, which suggests the respective sizes of impact are not close to equal.

Do you have a sim showing that this staking strat maximizes reward/risk?

I never have bet on anything that was 1000-1 so I have no idea what would be the benefits in an extreme scenario like that. I say I like the fixed profit betting on the major sports like NBA and the MLB where the line ranges from +300 to -300 usually.

[HeeeHAWWWW]

I then filtered out all wagers with expected returns < $1.20, leaving a winning set of wagers. Here full Kelly managed to lose money while half Kelly showed a profit.

The overall results probably suggest you're over-estimating your edge (no offence, it's almost the default of most gamblers). You might actually be at a stage where full-Kelly is -EG.

[danshan11]

thinking about it help me through this thought
if you stake
flat bet 1% of current BR and start to lose you need to even hit at a higher % than ATS to get even or caught up right since your bets are smaller?
flat bet 1% of starting bankroll and if you start to win, you are actually going to lower your ROI because you are winning 1% of starting not 1% of current bankroll right?

my wording might not be correct but the idea is close I think, can someone elaborate

[Bsims]

The overall results probably suggest you're over-estimating your edge (no offence, it's almost the default of most gamblers). You might actually be at a stage where full-Kelly is -EG.

I used $1.20 because it showed a profit on that particular set of data. As we've discussed before, expected results from a model that are very high should be viewed with suspicion. In reality, I never use any suggestions with expected returns over $1.15, and these are rare.

[Bsims]

if you stake
flat bet 1% of current BR and start to lose you need to even hit at a higher % than ATS to get even or caught up right since your bets are smaller?
flat bet 1% of starting bankroll and if you start to win, you are actually going to lower your ROI because you are winning 1% of starting not 1% of current bankroll right?

Your assessment is correct. I've done lots of work in the past looking at various bet sizing schemes. In fact I devoted 11 posts on my blog starting on 06/20/18 dealing with this issue. I started using Kelly in my current CBB system. But after my analysis, I've decided to go back to flat bets.

Your first point is right on. I won't lower the wager amounts because I get behind. That would essentially giving up on the system. In that case I'll just quit using the system.

The opposite result is more problematic. If the system is profitable then there will be a tendency to want to increase the wagers. I fallen into that trap in the past, not successfully. It'll be tough, but hopefully I won't do that again .

[zorba74]

lol there is literally a plethora of papers out there dissecting kelly to it's core yet you guys keep posting this garbage

anyone who has read through those papers would realise that full kelly is the only way to go

kelly maximizes ev to it's fullest potential

You don't want to optimize profits; crazy but true. If one wanted to optimize profits, they would employ Full Kelly but it's not "fit for purpose" when it comes to sports gambling. If you are skilled at id'ing edges, then a modified version of Kelly would be my reccommendation.

[Alfa1234]

Bsims, your results indicate higher odds would be more profitable with your system. I have found the same thing to be true in my DB, whilst favourites were also profitable, the higher priced dogs outperform them and long term flat unit betting beats a "to win" amount stake...although the volatility goes way down by using the latter system.

@ Danshan, I would not adjust your 1% bet size after each bet but only adjust the actual stake every few weeks or every month, that way a prolonged losing streak would not instantly drop your ROI by having a lower betsize once you start winning again.

[danshan11]

thanks alfa can you elaborate on the method more clearly to overcome the issues either way

[Optional]

@ Danshan, I would not adjust your 1% bet size after each bet but only adjust the actual stake every few weeks or every month, that way a prolonged losing streak would not instantly drop your ROI by having a lower betsize once you start winning again.

+1.

[Alfa1234]

thanks alfa can you elaborate on the method more clearly to overcome the issues either way

Not sure what you mean. If you bet 1% of your roll regardless of the odd you are taking, your volatility will be up because you could be betting 5 +800 dogs in a row with a large probability of losing all 5 bets. If you then adjusted your bet size after each loss (because your roll would be lower) you would be betting a lower amount once you start winning again. It's better to wait a full "cycle" to adjust the betsize.

Say you estimate your edge to be 5% it would be best to only adjust your betsize once your bankroll has actually grown 25% or so after the number of bets it would take at that edge to grow the roll (that way you know you are not in a dip or a peak re returns at that point).

E.g. If you have a 10% edge and have placed 100 bets with 1% of your roll, depending on the odds you are taking, you would only adjust your stake after those 100 bets and an exact 10% bankroll growth...or at least untill after the number of bets you placed corresponds with the bankroll growth you should have (120 bets would be 12% growth and so on, you could easily be in a dip after 100 bets with only 5% growth and then get back "on course" after 140 bets with 14% growth. This avoids over or under staking at "over growth" or "undergrowth" periods of your volatility curve. Hope that makes sense.

This also beats Kelly because you do not need to know your exact edge for every individual bet, but only over a larger sample size. It avoids overstaking and understaking on individual games where your estimated edge is too high or too low compared to what it actually is.

Doing this with a "to win" stake works the same way.

[KVB]

Alfa has a good way for practically every bettor out there.

I've posted about it before and it's best to adjust your bet size the least amount as possible as it will cost you vig. In fact, in sportsbetting, you can efficiently take a bankroll from 1k to 100k with only about a dozen bet size changes.

This is a secret of professional bettors and those who desire to take any income at any rate from their endeavors.

Even half Kelly is dangerous for those and best left to long term funds, if you can beat the market.

[danshan11]

Alfa has a good way for practically every bettor out there.

I've posted about it before and it's best to adjust your bet size the least amount as possible as it will cost you vig. In fact, in sportsbetting, you can efficiently take a bankroll from 1k to 100k with only about a dozen bet size changes.

This is a secret of professional bettors and those who desire to take any income at any rate from their endeavors.

Even half Kelly is dangerous for those and best left to long term funds, if you can beat the market.

what do you mean it will cost you vig

[Miz]

If you bet full kelly with a secret vault of money in case ruin occurs, you aren't betting full kelly

If your bankroll doesn't fluctuate 10-30% per day you're doing something wrong? LOL

[Okocha]

Everybody that doesnt know how to quantify their edge and consequently cant use fract. kelly is beeing outperfomed exponentially by those who can. This is a mathematical fact

[danshan11]

Everybody that doesnt know how to quantify their edge and consequently cant use fract. kelly is beeing outperfomed exponentially by those who can. This is a mathematical fact

can you please show some data or a link that shows that

[Okocha]

This is one of the basic main premises of Kelly

I strongly suspect that all this hours you are wasting spamming nonsense on forums if you could actually put those hours to study the ins and outs of SB you would know this


https://www.stat.berkeley.edu/~aldou..._Bad_Kelly.pdf

[danshan11]

This is one of the basic main premises of Kelly

I strongly suspect that all this hours you are wasting spamming nonsense on forums if you could actually put those hours to study the ins and outs of SB you would know this


https://www.stat.berkeley.edu/~aldou..._Bad_Kelly.pdf

this paper is very good but it assumes an edge that very very few bettors have or know how to calculate and the rude knock on me is not necessary

Kelly is perfect if you know your edge when you bet and the fact is very few people know their edge. I am not getting in a kelly debate with you or anyone because "momma said" is always the old argument on SBR. Momma said alligators were mean because the devil was in them, do you believe that too?

[danshan11]

99.97% or more people on this forum including me
A don't know what are edge is
B most dont even know how to calculate it
C have an edge, period!

yes I know you are on your 100th acct and know everything but I dont know you and you already said you had an empire earlier, so I dont know why you are wasting your time chatting with us when you have an empire, is this Jung Un?

FACT 99% of people who bet if they used kelly they are going to seriously bankrupt themselves reading crap by people like you!

[danshan11]

Everybody that doesnt know how to quantify their edge and consequently cant use fract. kelly is beeing outperfomed exponentially by those who can. This is a mathematical fact

and speak english who is this Young Sheldon, I dont understand all those big words so I am gonna take them as disrespect

[Okocha]

if you read carefully im not advocating kelly criterion for every bettor,im just stating the facts

Any bettor who knows how to quantify their edge with a reasonable margin of error and use some sort of a decent kelly fraction is in a great position to reach market cap/limits in the markets they attack way sooner than with other staking method

Believe me almost every bettor in this forumor 2plus2 lurking around or active users ( i dont know any names here aside you from betting talk )with 7 digits rolls are using Kelly criterion or used it until they reach max bet roll due to market limits


Thats the end goal anyway to have a max bet roll(not easy to reach that level obviously in the most liquid markets like NFL ,EPL due to a miryad of reasons)

And my previous post wasnt a jibe,was an advice

[danshan11]

if you read carefully im not advocating kelly criterion for every bettor,im just stating the facts

Any bettor who knows how to quantify their edge with a reasonable margin of error and use some sort of a decent kelly fraction is in a great position to reach market cap/limits in the markets they attack way sooner than with other staking method

Believe me almost every bettor in this forumor 2plus2 lurking around or active users ( i dont know any names here aside you from betting talk )with 7 digits rolls are using Kelly criterion or used it until they reach max bet roll due to market limits


Thats the end goal anyway to have a max bet roll(not easy to reach that level obviously in the most liquid markets like NFL ,EPL due to a miryad of reasons)

And my previous post wasnt a jibe,was an advice

so you mean for the .000001% that A Have an actual long term edge and B know their edge. OK so yes for them a partial Kelly is acceptable for a few weeks until they meet the market caps and then they can just max bet and their income stabilizes and kelly is USELESS for them too!

[Optional]

so you mean for the .000001% that A Have an actual long term edge and B know their edge. OK so yes for them a partial Kelly is acceptable for a few weeks until they meet the market caps and then they can just max bet and their income stabilizes and kelly is USELESS for them too!

Not that it should matter how many people can do it really, if it can be done it's proven.

But I think you are seriously underestimating the number of people who have built a model they are confident does give an accurate enough estimate of edge over a large sample to get an advantage from Kelly.

That said, same as Heehaaaw, I really like the way Alfa suggested doing it if following a Kelly staking plan. Don't bank on the accuracy of each individual bet, look at it over a month and adjust bet sizing every 30 days to smooth the ups and downs.

[zorba74]

so you mean for the .000001% that A Have an actual long term edge and B know their edge. OK so yes for them a partial Kelly is acceptable for a few weeks until they meet the market caps and then they can just max bet and their income stabilizes and kelly is USELESS for them too!

Why are you betting w/out a perceived edge? That doesn’t make any sense.

For entertainment?

Ok, that is perfectly fine, just risk no more than 1% of your Roll on any one play and you should avoid Ruin. I’ve never heard of a dude losing a $1,000 Roll, betting $10 at a time. Obviously, you want to trade your Equity Curve by scaling up/down your bets as your roll increases/decreases while keeping the bet percentage constant (E.G. 1%).

Kelly is fantastic, it normalizes risk, it helps you evaluate bets, etc. This vitriol against Kelly is silly; it is the way forward for many who want to up their game. Full Kelly is insane but you just apply the appropriate multiple to get to the fractional Kelly risk that you are comfortable with and avoids the Risk of Ruin.

[danshan11]

oh perceived edge, I never met anyone who does not think they have a perceived edge.

Kelly is useless again I will say it again
using kelly to bet before the closing line is there is betting on a guessed edge at best. you have no idea what your edge is on a typical bet and adding another if factor to the equation is a mess. Kelly cannot be backtested on sports betting because you dont know what your past perceived edge is.
It is irresponsible to tell people to bet kelly. Nuclear energy is the best and "proven" but I would not want an inexperienced 12 year old putting it in his lawnmower.
99% of the bettors I met dont even know how to calculate an edge and most are lost on what a long term edge really is. yes there is a small portion of people that can model and can come up with a fairly accurate overall edge they have on all their bets but they still cannot determine the actual edge on one specific game BEFORE the closing line.

I am sticking to it is irresponsible to tell people use kelly 1/10 half full and blah blah. anyone that can find an edge, has a real edge probably knows how to stake effectively as well. Kelly in the method accepted is not the best method for sports betting.

[Okocha]

Im new to this forum,how do i put a specific poster on my ignore list

[zorba74]

oh perceived edge, I never met anyone who does not think they have a perceived edge.

Kelly is useless again I will say it again
using kelly to bet before the closing line is there is betting on a guessed edge at best. you have no idea what your edge is on a typical bet and adding another if factor to the equation is a mess. Kelly cannot be backtested on sports betting because you dont know what your past perceived edge is.
It is irresponsible to tell people to bet kelly. Nuclear energy is the best and "proven" but I would not want an inexperienced 12 year old putting it in his lawnmower.
99% of the bettors I met dont even know how to calculate an edge and most are lost on what a long term edge really is. yes there is a small portion of people that can model and can come up with a fairly accurate overall edge they have on all their bets but they still cannot determine the actual edge on one specific game BEFORE the closing line.

I am sticking to it is irresponsible to tell people use kelly 1/10 half full and blah blah. anyone that can find an edge, has a real edge probably knows how to stake effectively as well. Kelly in the method accepted is not the best method for sports betting.

I don't even know what the F**k you are talking about and I don't think you do either.

[zorba74]

Im new to this forum,how do i put a specific poster on my ignore list

No clue but this dude is so far off; he's beyond help.