Average Stake as % of Bankroll?

[wantitall4moi]

bottomline is kelly is a formula. The numbers you plug in are based on what you know or what you think you can know. which is where it doesnt work. Or if you dont know the formula it wont work either.

Math ALWAYS works, it is the numbers people plug into the formulas that are wrong.

But like i said people take a formula or try to get others to give them a formula, and try and work it backwards with known results to get numbers that will plug in and make it work every time. Which isnt possible.

As for my example of kelly. I used 2-1 as pay out odds, just to get extreme. but a 2-1 pay out with a 50% chance of success would call for a 25% bankroll bet with kelly. If I had used 3 to 1 then 33.3% would be the amount of each bet, 4-1 would be 50% and so on. Thats the formula, and how what numbers and odds plugged in work.

I suppose I could have used +105 for odds but that would make people think you could never lose, mostly because it might take along long time to eat away at the profits, but with enough losing streaks it would. And the irony is, as all probability guys know the more flips you make the more certain a losing streak long enough to break you will happen. I just plugged in easy odds and a realistic losing streak to show kelly doesnt do what it claims to do, even when you know everything for certain.

With a sports bet all you know for certain is the pay out. Win expectation is not a known value. regardless of what you might think. No matter what you use it is a prediction, either yours or a bookmakers or a programs. Nothing mathematically verifiable.

[wiffle]

4-1 would be 50% and so on.

wat

[wiffle]

And the irony is, as all probability guys know the more flips you make the more certain a losing streak long enough to break you will happen. I just plugged in easy odds and a realistic losing streak to show kelly doesnt do what it claims to do, even when you know everything for certain.

lol, if you "plug in" a losing streak, then of course you're going to lose money

also, you resize your bets, obv

what do you think kelly claims to do?

[wantitall4moi]

lol, if you "plug in" a losing streak, then of course you're going to lose money

also, you resize your bets, obv

what do you think kelly claims to do?

your bets only resize due to % of your shrinking bankroll not in terms of win probability. Lessmoney equals less bet size. Which is simple math you dont need a formula to tellyou that.

Your probability in a coin flip never changes, it is a constant, just like the amount of the pay off. All numbers in the coin flip scenario are constant, yet still using full kelly it is possible to go broke, or at least lose enough so you cant bet.

All partial kelly or whatever bastardized forms of the equation does it limit risk size, which any monkey can do. But kelly supposedly has an optimal bet size, which the formula clearly spells out.

Basically when guys find success betting kelly it is because of other factors, so they do well in spite of it not because of it. Which generally means they are winning more than they are losing. But also not betting enough overall to have a risk of ruin. Which I stated. For the most part anything over 2% bankroll percentage on any one play is the absolute max, a one side play anyway (where one side wins you collect the other side wins you lose). You really dont need kelly to figure that out.

All kelly does in terms of theory is give people an amount that is corollary their expected advantage. Which again you dont really need kelly for. Because the percentages kelly will give out are just as flawed as some number you could pull out of your ass.

As for your 'wat' comment. yeah should be 37.5.

But it actually shows the odds you get are something to be mindful of as well, especially in sports betting. odds that are 'too good' will actually cause you to over bet more than 'truer' odds would.

What people always fail to realize with math and sports is if you are using past results to determine a future outcome youre already screwed. And that those results dont equal probability or chance or odds or whatever you want to call it.

The best example, the -3 in the NFL, for years people claimed it had a 12 or 13% chance to come in, which according to some databases of closing lines it did. That is assuming all people agreed what games were lined -3 in the first place. But as time went on the 'odds' dropped because fewer games lined 3 were landing on 3, so eventually it got all the way down to 10%, that was around 2007 (going back to 1994). Now it is somewhere around 9% (1994-2010) at least on the most generic of database searches, thats a huge drop over the past 10 years or so, So pre 2001 it was 12%, and since 2001 (2001-2010) it is around 8%. Does the chance of a coin flip ever change? Do any odds change, especially that much over a short time period? Or is it like the US dollar worth less now than it used to be?

Now I could actually go as far as to say with line shopping and betting early or late one could reduce that 8% to something close to 6% since 2001. Sometimes for a little more vig, but not enough to overcome that big a change, certainly less than buying off the 3 after the fact. So basically guys who used to bet pre 2001 used to determine the price of moving a line of the 3 at around 25-30 cents, actually guys still argue it is worth that much because that is what most books still charge. Which they used 'math' to back up. So at a 6% push rate what does that make buying off worth now?

That is where math fails in sports.

Again way off topic but it also goes to the heart of the underlying thought process of sports bettors.

Its ok to use it to some degree but you cant marry yourself to it.

While it is nice to have databases with subsets and results, and I have databases with openers, closers,and all the moves in between in all the major sports going back almost 10 years, it still doesnt help predict the future. But it gives insight into how games moved, how many times they move, when they moved back and forth, and all the nuances lines and odds take. And ultimately what it shows is that there isnt enough data to gain an advantage. Because line moves are 'right' as much as they are 'wrong'. And in some subsets line moves are wrong versus the opening line, despite the fairy tale that beating a closer is a sure fire way to win. Only way that is the case is if you bet the extremes both ways and hit enough middles. I have run that data as well, and there arent enough winners there either. At least not enough to make it profitable, it does reduce risk a lot obviously, but isnt a money maker.

Sorry to completely derail this thread, but I think the answer to the original question has been answered in quite a few ways.

[HeeeHAWWWW]

As for my example of kelly. I used 2-1 as pay out odds, just to get extreme. but a 2-1 pay out with a 50% chance of success would call for a 25% bankroll bet with kelly. If I had used 3 to 1 then 33.3% would be the amount of each bet, 4-1 would be 50% and so on. Thats the formula, and how what numbers and odds plugged in work.

At this point the gambler would be advised to stop for a reality check. +50% EV?

Kelly will screw you if your method is this far off base, no doubt. It demends a certain discipline and accuracy of estimate. If you don't have both, it shouldn't be used (and especially not full kelly).

[ballahollic2]

I figure 2% for my top wagers and 1% for my middle wagers. My small plays I only make .5%

[CrimsonQueen]

Better luck next time.

Alright wise one, please explain to the class Kelly, and how you should bet 50% of your Bankroll at -EV.

[wantitall4moi]

At this point the gambler would be advised to stop for a reality check. +50% EV?

Kelly will screw you if your method is this far off base, no doubt. It demends a certain discipline and accuracy of estimate. If you don't have both, it shouldn't be used (and especially not full kelly).

you can use full kelly if you want, but dont put your expectation above 53%. If youre getting -107 then you can lower it to 52%, basically give yourself as close to a break even probability as possible. To me that is the only way kelly 'works'. Because that is the time it actually tells you to play a realistic percentage. But at 52% and -107 your bet would be about .6%, so you could maybe step it up a little over that. but 53% at -107 calls for a 2.6% size bet. The major problem I have with kelly is how fast the percentage of the bankroll rises for such small increases in perceived advantage.

[CrimsonQueen]

your bets only resize due to % of your shrinking bankroll not in terms of win probability. Lessmoney equals less bet size. Which is simple math you dont need a formula to tellyou that.

Your probability in a coin flip never changes, it is a constant, just like the amount of the pay off. All numbers in the coin flip scenario are constant, yet still using full kelly it is possible to go broke, or at least lose enough so you cant bet.

All partial kelly or whatever bastardized forms of the equation does it limit risk size, which any monkey can do. But kelly supposedly has an optimal bet size, which the formula clearly spells out.

Basically when guys find success betting kelly it is because of other factors, so they do well in spite of it not because of it. Which generally means they are winning more than they are losing. But also not betting enough overall to have a risk of ruin. Which I stated. For the most part anything over 2% bankroll percentage on any one play is the absolute max, a one side play anyway (where one side wins you collect the other side wins you lose). You really dont need kelly to figure that out.

All kelly does in terms of theory is give people an amount that is corollary their expected advantage. Which again you dont really need kelly for. Because the percentages kelly will give out are just as flawed as some number you could pull out of your ass.

As for your 'wat' comment. yeah should be 37.5.

But it actually shows the odds you get are something to be mindful of as well, especially in sports betting. odds that are 'too good' will actually cause you to over bet more than 'truer' odds would.

What people always fail to realize with math and sports is if you are using past results to determine a future outcome youre already screwed. And that those results dont equal probability or chance or odds or whatever you want to call it.

The best example, the -3 in the NFL, for years people claimed it had a 12 or 13% chance to come in, which according to some databases of closing lines it did. That is assuming all people agreed what games were lined -3 in the first place. But as time went on the 'odds' dropped because fewer games lined 3 were landing on 3, so eventually it got all the way down to 10%, that was around 2007 (going back to 1994). Now it is somewhere around 9% (1994-2010) at least on the most generic of database searches, thats a huge drop over the past 10 years or so, So pre 2001 it was 12%, and since 2001 (2001-2010) it is around 8%. Does the chance of a coin flip ever change? Do any odds change, especially that much over a short time period? Or is it like the US dollar worth less now than it used to be?

Now I could actually go as far as to say with line shopping and betting early or late one could reduce that 8% to something close to 6% since 2001. Sometimes for a little more vig, but not enough to overcome that big a change, certainly less than buying off the 3 after the fact. So basically guys who used to bet pre 2001 used to determine the price of moving a line of the 3 at around 25-30 cents, actually guys still argue it is worth that much because that is what most books still charge. Which they used 'math' to back up. So at a 6% push rate what does that make buying off worth now?

That is where math fails in sports.

Again way off topic but it also goes to the heart of the underlying thought process of sports bettors.

Its ok to use it to some degree but you cant marry yourself to it.

While it is nice to have databases with subsets and results, and I have databases with openers, closers,and all the moves in between in all the major sports going back almost 10 years, it still doesnt help predict the future. But it gives insight into how games moved, how many times they move, when they moved back and forth, and all the nuances lines and odds take. And ultimately what it shows is that there isnt enough data to gain an advantage. Because line moves are 'right' as much as they are 'wrong'. And in some subsets line moves are wrong versus the opening line, despite the fairy tale that beating a closer is a sure fire way to win. Only way that is the case is if you bet the extremes both ways and hit enough middles. I have run that data as well, and there arent enough winners there either. At least not enough to make it profitable, it does reduce risk a lot obviously, but isnt a money maker.

Sorry to completely derail this thread, but I think the answer to the original question has been answered in quite a few ways.

I could argue with you all day that math doesn't fail in sports. Read a book like Scorecasting and you'll see where math succeeds. On top of that, I've not watched a single MLB game nor WNBA game this year, yet I have a Z-score of >2 in BOTH sports (admittedly the WNBA has a very small sample size, but MLB has a sample size of 646 games). How? Math. Sure Kelly may be extreme, but to argue that math fails in sports entirely is outrageous.

Also, when you lose 10 bets in a row with Kelly, starting with a $1000 BR, and you estimate you'll win 55%... at -110... you'd be at $567.96 at the end of 10 straight losses, since you always want to start off with 10 L's in a row. Say another person just has flat units of $50 (Or say they vary their bets between $25-$75 and average $50 per) and they lose 10 straight, they are down to $450 after 10 L's to start.

Again, I'm not advocating Kelly, I'm merely pointing out that math in sports is important and a good thing.

[donjuan]

Alright wise one, please explain to the class Kelly, and how you should bet 50% of your Bankroll at -EV.

100k bankroll, 100k limits on a coinflip. One friend offers you +150 and another offers you -110. Kelly says 50% on each.

[CrimsonQueen]

100k bankroll, 100k limits on a coinflip. One friend offers you +150 and another offers you -110. Kelly says 50% on each.

Obviously that is a +EV spot, but you're not betting on the outcome of a single bet now at -110; you're now arbitraging a -110 line AND a +150, and it's the +150 part that makes it a +EV bet, not the -110 part alone.

[donjuan]

Sorry you have trouble with reading comprehension. Can you show me where anyone said 50% on a bet uncorrelated to any others in your portfolio?

Also it's not pure arbitrage or just a +EV spot. That is the optimal Kelly solution.

[CrimsonQueen]

Wow... you wanna pull reading comprehension out when you're the one who's got trouble reading?

I said this, and only this:
"You do not have an edge on a 50/50 proposition at -110. Even at EVEN odds, you wouldn't make a bet with Kelly on a literal coin flip."

To which you said this:
"Better luck next time."

LOL... I love you think you're right because you took my scenario and said I was wrong, THEN changed the scenario to make yourself correct. What a joke.

Better luck next time.

[CrimsonQueen]

Also, if you must know who initially said betting 25%, that'd be "But if you start with 100 bux and bet full kelly (25%) per flip at +100" wantitall4moi and then talks about losing 10 - 20 straight... in post #18. Then wiffle said 50% on a coin flip. If you want to get all technical, and say that he didn't clarify that it's NOT in conjunction to any other bets, that's like me saying:
"I made this great bet: -400, if the Eagles win the Superbowl, I win."

You:
"That's a terrible -EV bet."

Me:
"You need to learn to read! When did I say it was ONLY on the Eagles to win? It could be the Eagles OR any other team at all. So long as there's a Superbowl winner, I win."

[donjuan]

Nice use of selective quoting. Why don't you go ahead and post the whole thing including the post you quoted that you were responding to?

[wiffle]

Also, if you must know who initially said betting 25%, that'd be "But if you start with 100 bux and bet full kelly (25%) per flip at +100" wantitall4moi and then talks about losing 10 - 20 straight... in post #18. Then wiffle said 50% on a coin flip. If you want to get all technical, and say that he didn't clarify that it's NOT in conjunction to any other bets, that's like me saying:
"I made this great bet: -400, if the Eagles win the Superbowl, I win."

You:
"That's a terrible -EV bet."

Me:
"You need to learn to read! When did I say it was ONLY on the Eagles to win? It could be the Eagles OR any other team at all. So long as there's a Superbowl winner, I win."

not sure if serious

all i said was that it is possible for betting half your roll on a -ev wager to be the optimal kelly solution

[CrimsonQueen]

Nice use of selective quoting. Why don't you go ahead and post the whole thing including the post you quoted that you were responding to?

I posted the post #, and if you wanna go back and reread it, it's ONE PAGE BACK. You are annoying and have included absolutely nothing helpful, so you're now the first person on my ignore list! Cheers!

[Dark Horse]

nm.

[wiffle]

I posted the post #, and if you wanna go back and reread it, it's ONE PAGE BACK. You are annoying and have included absolutely nothing helpful, so you're now the first person on my ignore list! Cheers!

def -ev to ignore donjuan

[CrimsonQueen]

not sure if serious

all i said was that it is possible for betting half your roll on a -ev wager to be the optimal kelly solution

I agree. This is what you said. After other dude posted that you would bet 25% of your roll on a coin flip at +100. He was not implying anything else when he said that.

This thread has degenerated into nothing.

[wiffle]

nm.

ninja edit

[CrimsonQueen]

def -ev to ignore donjuan

It's -EV to listen to the dude... he just tries to catch you on weird technicalities. But that is what the Think Tank has become known for though... a bunch of math nerds fighting each other....

[wiffle]

if you dont love it, leave it

[Pancho sanza]

It's -EV to listen to the dude... he just tries to catch you on weird technicalities. But that is what the Think Tank has become known for though... a bunch of math nerds fighting each other....

Its not a technicality, Kelly says to bet on negative EV propositions when the amount you stand to win on an open wager is large relative to your bankroll.

[wantitall4moi]

This thread has degenerated into nothing.

its a sports forums what do you expect. Kelly is a formula, there shouldnt be a 'right' or 'wrong' with it per se. Just how it is utilized.

All it does is give people a guideline on how much to risk on something...anything, depending on a few factors. Sometimes known, sometimes not known.

It is basically used to avoid overexposure, while giving an optimal bet size in regards to BR. So that is where the debate begins. betting a 1% flat rate of your BR is obviously safe. Is it optimal? That is where the debate begins. is 3.5% too much? Not if you win it isnt. That is the other debate.

People debating semantics is definitely a degeneration. I just know the guy asked a question. Some guys said kelly, some guys said partial kelly, other guys gave numbers ranging from 1% to 2% for varying reasons. And all of those might be close depending on what numbers they use to plug into the equation. The major problem with kelly is the percentages jump way too much in relation to your confidence (win probability) assumptions. A 1% difference in assumption can meana bet more than 4 times as much as one 1% less confident, 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.

[jolmscheid]

I figure 2% for my top wagers and 1% for my middle wagers. My small plays I only make .5%

If you don't bet a flat 1%, 2%, or whatever on each play, then just keep it easy like ballahollic2....it's better to be consistent more than anything if you ask me...

[donjuan]

Let's go over this:

You:

I wouldn't use kelly myself...but trust me, kelly does NOT tell you to bet 25% of your BR on a coin flip.

Since this is an obviously wrong statement, Wiffle chimes in with the correct

it is possible for kelly to tell you to bet half your roll on a coinflip at -110

then you respond

...??? No! Does no one actually understand Kelly? You only bet if you have an edge. You do not have an edge on a 50/50 proposition at -110. Even at EVEN odds, you wouldn't make a bet with Kelly on a literal coin flip.

Which makes it clear it is you who doesn't understand Kelly. Then we have what follows which results in you throwing a tantrum about being wrong and claiming to put me on ignore for having the audacity to be right.

[CrimsonQueen]

Let's go over this:

You:

Since this is an obviously wrong statement, Wiffle chimes in with the correct



then you respond



Which makes it clear it is you who doesn't understand Kelly. Then we have what follows which results in you throwing a tantrum about being wrong and claiming to put me on ignore for having the audacity to be right.

Why can I still hear you talking about something that everyone else has moved past? You got me on a technicality that wasn't included in the original thing I was even talking about! Congrats! You're King of the Smarties!! Yay!

[wiffle]

Why can I still hear you talking about something that everyone else has moved past? You got me on a technicality that wasn't included in the original thing I was even talking about! Congrats! You're King of the Smarties!! Yay!

[ ] put donjuan on ignore

[wiffle]

it's not really a technicality, though.
you learned an important lesson in optimal bet sizing, but for some reason you're mad at donjuan.

[donjuan]

Why can I still hear you talking about something that everyone else has moved past? You got me on a technicality that wasn't included in the original thing I was even talking about! Congrats! You're King of the Smarties!! Yay!

Just because you didn't previously know that Kelly calls for -EV hedges, doesn't make this a technicality . You pretty clearly need to read more Ganchrow before you go spouting off again.

P.S. I thought I was on ignore.

[Dark Horse]

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.

[CrimsonQueen]

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.

[Pancho sanza]

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.

I guess most people don't follow kelly then, because as was stated, the kelly optimal bet is 50 % of your roll on each side.

[Dark Horse]

I guess most people don't follow kelly then, because as was stated, the kelly optimal bet is 50 % of your roll on each side.

Looks like it.

Not long ago I qualified 'common sense' here as maximum profit for minimum exposure. I don't care if that's considered Kelly or not. But, after I did so, sure enough I was told that is exactly what Kelly is. Fine with me. So my exposure in the aforementioned single coin flip is zero.