placing several bets at once using Kelly criterion

[allin1]

Recently I have been using Kelly criterion but I have a question for others who use it.

For example I want to place a bet, I calculate the stake with the formula and now it is pending.

Is it best to wait for it to be graded until placing a new bet?

Can the long term evolution of the bankroll suffer if while I have a pending bet I place another one with the stake calculated for the remaining available bankroll?

What about several different bets placed at once? How do you use Kelly criterion in such a situation?

Or is it better to place just one bet at once in order to see if it is a winner to calculate the stake for the next bet according to the new available bankroll?

[allin1]

I think I found the answer here http://www.albionresearch.com/kelly/default.php

[zonedave]

can someone link or explain.

why kelly is better than flat betting?

[allin1]

can someone link or explain.

why kelly is better than flat betting?

[roasthawg]

Simultaneous kelly is done by taking the product of 1-kelly for all simultaneous wagers and then multiplying each kelly amount by that product. So if I have two simultaneous bets, one calling for a full kelly wager of 2.5% and another calling for a 2% wager then I would bet (1-.025)*(1-.02)*2.5% = 2.387% on the first wager and (1-.025)*(1-.02)*2% = 1.91% on the second wager.

[allin1]

roasthawg thanks a lot for the answer. This is new info for me.

[allin1]

just found this: http://www.sbrforum.com/betting-tools/kelly-calculator/

great tool

[Cutler'sThumb]

Everything I've read regarding Kelly is that it is a somewhat viable theory, but completely impractical in real life. This is basically because if you place 1000 bets, based on probability theory, you will have 2 or 3 runs of 12 loses in a row and your ante to win the one unit back will be astronomical. Most people will not be able to handle that for one reason or another (bankroll limitations, human nature, ect.).
I'm no sage, but a scientific mind says find another system that tickles you.

[LT Profits]

Everything I've read regarding Kelly is that it is a somewhat viable theory, but completely impractical in real life. This is basically because if you place 1000 bets, based on probability theory, you will have 2 or 3 runs of 12 loses in a row and your ante to win the one unit back will be astronomical. Most people will not be able to handle that for one reason or another (bankroll limitations, human nature, ect.).
I'm no sage, but a scientific mind says find another system that tickles you.

This thread is about Kelly, not Martindale

[bztips]

This thread is about Kelly, not Martindale

To be picky, it's Martingale not Martindale! (wink, wink)

[FourLengthsClear]

Everything I've read regarding Kelly is that it is a somewhat viable theory, but completely impractical in real life. This is basically because if you place 1000 bets, based on probability theory, you will have 2 or 3 runs of 12 loses in a row and your ante to win the one unit back will be astronomical. Most people will not be able to handle that for one reason or another (bankroll limitations, human nature, ect.).
I'm no sage, but a scientific mind says find another system that tickles you.

LOL.
There is so much wrong with this 'paragraph' I don't know where to start. Suffice to say I would suggest more reading and that you pay more attention this time.

[LT Profits]

To be picky, it's Martingale not Martindale! (wink, wink)

DOH! Guess I watch too many game shows.

[Cutler'sThumb]

Fair enough, as I said I'm no expert. I made my first sports bet about a month ago, and have only read a limited amount and am on the very bottom of the learning curve. Am I wrong about what I said regarding the Martingale system? The inevitable long losing streak mixed with human nature being the Achilles Heel was what I came away with. The 12 losses in a row being inevitable was in a thread I read regarding probability theory over several thousand outcomes. I haven't gone deeper than that, but it seemed to make sense.
I'm clueless on Kelly at this point, but I understand that there is great debate on either side and look forward to learning.

[Cutler'sThumb]

Being an admitted newbie,I'm in total sponge mode now. Plus when I posted last night I was 1/2 asleep trying to get to 20 posts so I could PM this guy back wanting to know how to get Adam Meyer's outfit to stop humping him (I put up a post telling of my horrible experience with them...the beginning of my education in this world).

[mebaran]

Fair enough, as I said I'm no expert. I made my first sports bet about a month ago, and have only read a limited amount and am on the very bottom of the learning curve. Am I wrong about what I said regarding the Martingale system? The inevitable long losing streak mixed with human nature being the Achilles Heel was what I came away with. The 12 losses in a row being inevitable was in a thread I read regarding probability theory over several thousand outcomes. I haven't gone deeper than that, but it seemed to make sense.
I'm clueless on Kelly at this point, but I understand that there is great debate on either side and look forward to learning.

The debate isn't over whether Kelly maximizes growth, that has been established. The debate is whether or not a person can calculate their edge accurately enough to use Kelly, and whether or not the market they are betting into is one that adheres to the strong form efficiency hypothesis. And even then..I think the only debate is whether the markets are efficient, as it's been established that you can calculate your edge at least accurate enough to use fractional Kelly.

[FourLengthsClear]

Fair enough, as I said I'm no expert. I made my first sports bet about a month ago, and have only read a limited amount and am on the very bottom of the learning curve. Am I wrong about what I said regarding the Martingale system? The inevitable long losing streak mixed with human nature being the Achilles Heel was what I came away with. The 12 losses in a row being inevitable was in a thread I read regarding probability theory over several thousand outcomes. I haven't gone deeper than that, but it seemed to make sense.
I'm clueless on Kelly at this point, but I understand that there is great debate on either side and look forward to learning.

Fundamentally you are not wrong about Martingale (or any other form of "chase" system), the issue was that you confused Martingale with Kelly. They are entirely different concepts.

The probabilities you mentioned (2 or 3 streaks of 12 losses in 1000 plays) are way-off if we are talking about sides/totals in the -110 range. The likelihood of 12 straight losses on coin-flips is 1 in 4,096.

[brh75]

rooo0llll tide

[stake]

The debate isn't over whether Kelly maximizes growth, that has been established. The debate is whether or not a person can calculate their edge accurately enough to use Kelly, and whether or not the market they are betting into is one that adheres to the strong form efficiency hypothesis. And even then..I think the only debate is whether the markets are efficient, as it's been established that you can calculate your edge at least accurate enough to use fractional Kelly.

What is the best way to calculate edge? And how do you know if it is accurate? Not criticizing just asking because I'm learning.

[FourLengthsClear]

What is the best way to calculate edge? And how do you know if it is accurate? Not criticizing just asking because I'm learning.

Whatever method you use to make your picks (mathematical modeling, handicapping etc.) you should have a calculated probability of winning your bet. Your edge is the positive expected value of this probabilty relative to the odds you bet at.

To use a simple example, if you are able to bet at +110 on a coinflip, the edge (+EV) would be

50.00% (probability of your bet winning)
divided by
47.61% (implied probability of the odds you bet at)
=
1.0502

minus 1

So +EV of 5.02%

[warriorfan707]

Kelly betting is horrible.

[mebaran]

What is the best way to calculate edge? And how do you know if it is accurate? Not criticizing just asking because I'm learning.

You handicap a game and come up with your own line. You compare the line you came up with to the line the market is offering. If there is a difference in your favor (ie 2%), than that is your edge. But since there are so many variables and misinformation and a billion other variables that I couldn't even begin to list, knowing your edge, absolutely, is impossible, even for the best handicappers in the world.

Hope this helps. I'll add that you should search through the think tank using the advanced search function for anything you're curious about.

[FourLengthsClear]

Kelly betting is horrible.

It is if you are unable to quantify an edge with confidence.

[stake]

You handicap a game and come up with your own line. You compare the line you came up with to the line the market is offering. If there is a difference in your favor (ie 2%), than that is your edge. But since there are so many variables and misinformation and a billion other variables that I couldn't even begin to list, knowing your edge, absolutely, is impossible, even for the best handicappers in the world.

Hope this helps. I'll add that you should search through the think tank using the advanced search function for anything you're curious about.

So kelly does not work unless you get lucky in determining your edge?

[FourLengthsClear]

So kelly does not work unless you get lucky in determining your edge?

Most proponents of Kelly will not be 'handicappers' but those who use mathematical models to produce a fair line.

These are tested/compared to the real closing lines over thousands of matchups so that it is not a matter of 'getting lucky'.

[stake]

Most proponents of Kelly will not be 'handicappers' but those who use mathematical models to produce a fair line.

These are tested/compared to the real closing lines over thousands of matchups so that it is not a matter of 'getting lucky'.

I didnt mean for the getting lucky to be brash just the way my young mind spit it out. So using kelly I dont have to determine my edge because kelly does it for me? If a line is telling me that a team should win 60% of the time but I think they will win 70% of the time thats where my handicapping comes in to determining edge? Using kelly sounds like a tight rope act of figuring edge properly. Sorry for asking so many questions just trying to understand in layman terms.

[mebaran]

I didnt mean for the getting lucky to be brash just the way my young mind spit it out. So using kelly I dont have to determine my edge because kelly does it for me? If a line is telling me that a team should win 60% of the time but I think they will win 70% of the time thats where my handicapping comes in to determining edge? Using kelly sounds like a tight rope act of figuring edge properly. Sorry for asking so many questions just trying to understand in layman terms.

1) No. You must figure out your edge and plug it into the Kelly formula.
2) It is a tight rope act, which is why a lot of pros use fractional Kelly (1/2 or 1/4 kelly are common).

[warriorfan707]

It is if you are unable to quantify an edge with confidence.

The main issue is the variables in the equation are purely speculatory.

[overtip]

mmmm

mmmmmm

[mebaran]

The main issue is the variables in the equation are purely speculatory.

That's not the "main" issue. I think you're phrasing this incorrectly. The process of betting on a sport is speculative (as is an options/stock exchange bet). The outcome, and probability of that outcome, are not known to within 100% confidence; that would be impossible. But, you CAN know, to a satisfactory level of confidence (say 95%), that a result will happen. So yes, the formula's inputs are never TRUELY known, but they are known to a high enough degree of confidence that Kelly is worth using. Indeed, then, you are speculating. But it's not the formula, it's the setting in which it is used that is speculative in nature.

The bottom line is that Kelly WILL maximize your growth (more so than almost any other strategy). It is a matter of how comfortable you are with risk, and what level of said risk you are willing to accept to achieve growth. You may be comfortable flat betting, I may be comfortable betting 1/2 Kelly, a high-level prop trader on the CBOE in Chicago may be comfortable using 9/10, or even full Kelly to achieve a return.

Would be willing to wager that 1/2 Kelly would outperform almost any strategy you can come up with. It is very powerful when you know your edge.

[warriorfan707]

Please define "knowing your edge".

[stake]

That's not the "main" issue. I think you're phrasing this incorrectly. The process of betting on a sport is speculative (as is an options/stock exchange bet). The outcome, and probability of that outcome, are not known to within 100% confidence; that would be impossible. But, you CAN know, to a satisfactory level of confidence (say 95%), that a result will happen. So yes, the formula's inputs are never TRUELY known, but they are known to a high enough degree of confidence that Kelly is worth using. Indeed, then, you are speculating. But it's not the formula, it's the setting in which it is used that is speculative in nature.

The bottom line is that Kelly WILL maximize your growth (more so than almost any other strategy). It is a matter of how comfortable you are with risk, and what level of said risk you are willing to accept to achieve growth. You may be comfortable flat betting, I may be comfortable betting 1/2 Kelly, a high-level prop trader on the CBOE in Chicago may be comfortable using 9/10, or even full Kelly to achieve a return.

Would be willing to wager that 1/2 Kelly would outperform almost any strategy you can come up with. It is very powerful when you know your edge.

Thank you mebaran for the response this is a great thread. Looks like I have some reading and studying to do.

[warriorfan707]

The numbers you are supposed to input into the kelly equation are totally subjective.

[mebaran]

Please define "knowing your edge".

Warrior, I suspect you are trying to demonstrate that "edge" cannot be known, but I'm saying it can.

Edge is: Predict the outcome of a game. Compare your prediction with the market's. Difference is your edge.

Obviously way more complicated with plenty of tangents, but the process is the same in the stock market. I valuate a stock price's fair value at $10/share. The market price is $8/share. I believe an opportunity exists, and I act accordingly.

If your methodologies in arriving at your prediction are sound, and is based on a framework of historic results in your sport of choice (or any market), you can know your edge over the market price to a sufficient enough degree to profit.

If your argument is true, and you actually can't compute your mathematical edge, than nobody would win long-term in sports betting/stock picking.

Side note: Kelly is a subset of probability theory. TECHNICALLY, Kelly is only accurate if an individual bet can be "replayed" thousands of times. But our world is based on probabilities...insurance companies that estimate a person's expected lifespan is using probability theory to arrive at their numbers. Do they know how long this person will live with 100% accuracy? No, of course not. BUT, they know the expected lifespan, so they offer a premium accordingly. Last time I checked, these big insurance carriers do pretty well financially.

[warriorfan707]

How do you "compare your prediction of the game with the markets"????

[mebaran]

How do you "compare your prediction of the game with the markets"????

lol...what are you getting at?

I say a team has a 56% chance to win an event. This is based on a framework I have deemed statistically significant. Step 1 complete.

I check the "market" (generally speaking, the best price you can find on both sides...or you can just use Pinny for major sports if you want). In this hypothetical case, let's say the best price I can find on the favorite is -120, and the underdog is +116. This would lead to fair probabilities of each team winning: 45.9% (dog) and 54.1% (favorite). Step 2 complete.

I see there is a discrepancy, and I act. In this case, my edge would be 2.67%.