Kelly Re-Visited

[danshan11]

sorry I dont type well
if you use kelly,COMMA to calculate your edge you need to know your edge so if your edge is X you pop X into kelly every single time you bet so X is not a factor anymore since it stays basically the same and that is flat betting, I could be wrong been wrong many times before, so no issue there but please clarify where I am missing what you are doing because I feel like my understanding of kelly is fairly sound, please advise thanks again!

[danshan11]

You use Kelly to calculate your wager size not your edge. First you must have an edge and at least a close estimate of what that edge is before you can use Kelly. Then your wager size as a % of your bankroll is determined by Kelly as your edge divided by the odds. The actual dollar amount of your wagers will fluctuate with changing edges, odds and/or bankroll size.

Joe.

how do you calculate an edge you have 2 options right?
1 guess what the closing line will be cause no one knows even Pinny
2 calculate the edge you had on your last 5 million games and use that which is not kelly because you are now flat betting again
how else can you do it? please clarify

[danshan11]

You use Kelly to calculate your wager size not your edge. First you must have an edge and at least a close estimate of what that edge is before you can use Kelly. Then your wager size as a % of your bankroll is determined by Kelly as your edge divided by the odds. The actual dollar amount of your wagers will fluctuate with changing edges, odds and/or bankroll size.

Joe.

every edge is different, I have never seen anyone be able to calculate their edge on an individual bet before the line closes. if you use the current bet line as implied win% that can change as well, kelly is great for many things but is not designed for sports betting in anyway

[danshan11]

now if we somehow knew the closing line in advance Kelly would be the best sports betting tool ever made but if you bet
as if you have an implied probability based on betting line and closing line closes way off that you are getting yourself into the unknown and our goal as sports bettors is to keep the unknowns as minimal as possible. I am sure we will not agree and that is OK some very smart and educated bettors love kelly. I think it is personal choice really but for me nothing compares to fixed profit betting which technically is a form of kelly at least a cousin, so I am kinda contradicting myself from the beginning, when I say screw kelly I am saying the most common used method where inexperience cappers use their dream edge and or the implied at betting when often times that is not close to the actual implied.

[danshan11]

I guess my true argument is why add another variable to our job of trying to minimize and account for variables, less variables we have as bettors the clloser we get to winning, which very very few of us do anyway. People think a few people win and a few is super high, you dont hear about number of people who didnt win the lottery you hear about the guy who won 1000000000000 for $1 dollar

[u21c3f6]

how do you calculate an edge you have 2 options right?
1 guess what the closing line will be cause no one knows even Pinny
2 calculate the edge you had on your last 5 million games and use that which is not kelly because you are now flat betting again
how else can you do it? please clarify

First, in order to clarify, if I cash 53% of my -105 wagers can you tell me what my edge is? (and notice that this has nothing to do with what the actual closing line is)

Joe.

[danshan11]

lets say a guy bets overnights on mlb and uses implied when he bets kelly and we all know the lines move a ton before close so lets assume he is 10% off the implied by closing average.
he is now underbetting or overbetting and now he is winning or losing not because of his picks but because of his staking and that is not good!

90% of guys dont know how to calculate an edge
9% of guys dont know how to bet it
.99% of guys dont have discipline to ride it out long term
and the winning .00001% use it and swear by it !

[danshan11]

you would have a 1.8% edge at 53% cashed at -105

[u21c3f6]

you would have a 1.8% edge at 53% cashed at -105

I need you to try again to make sure you just didn't make a calculation error.

Joe.

[danshan11]

I need you to try again to make sure you just didn't make a calculation error.

Joe.

here is my math -105 is 51.22% no margin
if you win 53% of the time you have a 1.8% margin
now if you include some sort of margin that would drastically change but we dont know the margin just knowing the -105

[danshan11]

Joe buddy no reason to try to make me prove I am dumb, I will gladly admit I dumb upfront to avoid the cat and mouse game, if you think I am wrong dont hint just come out and say "just the facts Joe, just the facts" dragnet couldnt resist LOL

[u21c3f6]

here is my math -105 is 51.22% no margin
if you win 53% of the time you have a 1.8% margin
now if you include some sort of margin that would drastically change but we dont know the margin just knowing the -105

I do not mean to be unkind but you do not understand edges or Kelly and therefore have no basis for most of what you have written. I would suggest reading many articles online to get a better understanding of Kelly and edges. I would have to write too much to get you up to speed. Again, not trying to be unkind but you clearly do not understand.

Joe.

[danshan11]

you are saying my math is wrong? if so show me the error please. also if you can just give me 10 words on where I am wrong on kelly, keep it as simple as possible since you are very busy. what is your edge on 53% winners at -105 telling me this will clarify a ton?

[danshan11]

I think its unfair to call someone incompetent and dont even explain where the math is wrong? "you are so wrong that I wont even explain it and just not respond again, sounds fishy at best, my math seems very solid to me if someone disagrees with my math please fill in for Joe, I would love to learn from my mistake not just going on believing wrong things or doubting something that is right.

[HeeeHAWWWW]

Edge is decimal odds * probability, so 100/105 +1 * 0.53 =1.035.

As with a lot of things, decimal odds are wayyyy easier to work with. I can't help but think that US odds (and UK too actually) are just designed to confuse.

[danshan11]

actually edge has nothing to do with wins and losses as he said 53% cashed . Edge has to do with how you do against the line, another strong argument against kelly is this here where even some seasoned people think different things. I cant believe he called me dumb and would not say what is dumb about it. People wins and losses have nothing to do with your edge. WIns losses do not mean anything.
if you flipped 10000 coins a few people by deviation would be awesome and a few people would be horrible at it. the only true measure of skill in sports betting is line movement, dont listen to the tout hype. Edge is how much you beat the line by over the margin it has ZERO to do with wins and losses. example
wins 600 losses 400
avg line paid 51.22% (true line 50%) what you paid
avg line closed 51.22% (true line 50%) what the game closed at
your edge is not 0% your edge assuming a pinny margin at -105 -105 is (NEGATIVE) -1.22%
example B
avg paid line 51.22% (true line 50%) what you paid
avg closed line 54.22% (true line 53% roughly) what it closed at
edge 3%
wins losses mean nothing only look at closed line value
paid implied win percentage compared to closed line implied win percentage
this is the only way to win and of course calculate your edge if you have one and believe me most dont!

[u21c3f6]

actually edge has nothing to do with wins and losses as he said 53% cashed . Edge has to do with how you do against the line, another strong argument against kelly is this here where even some seasoned people think different things. I cant believe he called me dumb and would not say what is dumb about it. People wins and losses have nothing to do with your edge. WIns losses do not mean anything.

Danshan11, I really don't know how to respond to you. First, I did not call you dumb. I wrote that you don't understand edge and Kelly, meaning that you needed more information to understand edge and Kelly and therefore I suggested that you do some research and reading.

As to the rest of what you wrote, wins and losses are very much at the heart of my edge and therefore I don't know what else I could write that would show you what I mean. In addition, beating the line has nothing to do with my edges. You and I are not only not on the same page, we are not even in the same book. Again, I sincerely suggest that you do some more research and reading on these subjects to get a better understanding.

Joe.

[danshan11]

Joe
just tell me what your edge is on 53% wins on -105 tickets?

[danshan11]

Joe I never like to tell someone they dont know what they are talking about it seems rude but I am afraid your vagueness is leading me to believe you might have bought off more knowledge than you know. I am not sure about that but the vague and elusive responses lean towards that, you dont have time to clarify but you got time to repeat yourself, that is suspect as well. Please clarify your position you are looking a little lost at this point, please clarify, thanks

[danshan11]

Joe, answer this simple question
say you flip a coin 1000 times
do you think it will land on heads 500 times and tail 500 times probably not, more like 527 heads and 473 tails
does that mean to you that heads has an edge? come on brother think!

[u21c3f6]

Joe
just tell me what your edge is on 53% wins on -105 tickets?

HH already answered this above. There are several ways to go about calculating your edge but all (correct) ways will come up with the same answer.

53 wins @ 2.05 divided by 100 (wagers) @ 1.05 or 108.65/105 = 1.034762 in other words if you hit 53% @ -105 you have a 3.4762% edge.

Your full Kelly wager is edge/odds or in this case 3.4762%/(100/105) = 3.65% of your bankroll.

Joe.

[danshan11]

HH already answered this above. There are several ways to go about calculating your edge but all (correct) ways will come up with the same answer.

53 wins @ 2.05 divided by 100 (wagers) @ 1.05 or 108.65/105 = 1.034762 in other words if you hit 53% @ -105 you have a 3.4762% edge.

Your full Kelly wager is edge/odds or in this case 3.4762%/(100/105) = 3.65% of your bankroll.

Joe.

lets laugh order a beer and say to each other at the same time "you are wrong" and talk about something else, you are very correct me and you are on different betting planets. wins have nothing to do with your edge brother refer to the coin flip example I gave you above. good luck and please dont use that method and PS dont forget the juice in your magic formula!

[danshan11]

for everyone else accept Joe, WINS and LOSSES have NOTHING, I mean NOTHING to do with edge.

[danshan11]

sorry Joe *except

[MonkeyF0cker]

actually edge has nothing to do with wins and losses as he said 53% cashed . Edge has to do with how you do against the line, another strong argument against kelly is this here where even some seasoned people think different things. I cant believe he called me dumb and would not say what is dumb about it. People wins and losses have nothing to do with your edge. WIns losses do not mean anything.
if you flipped 10000 coins a few people by deviation would be awesome and a few people would be horrible at it. the only true measure of skill in sports betting is line movement, dont listen to the tout hype. Edge is how much you beat the line by over the margin it has ZERO to do with wins and losses. example
wins 600 losses 400
avg line paid 51.22% (true line 50%) what you paid
avg line closed 51.22% (true line 50%) what the game closed at
your edge is not 0% your edge assuming a pinny margin at -105 -105 is (NEGATIVE) -1.22%
example B
avg paid line 51.22% (true line 50%) what you paid
avg closed line 54.22% (true line 53% roughly) what it closed at
edge 3%
wins losses mean nothing only look at closed line value
paid implied win percentage compared to closed line implied win percentage
this is the only way to win and of course calculate your edge if you have one and believe me most dont!

So, you're saying that you can beat the closing line consistently without estimating your own edge on a particular matchup? I really doubt that's true over a large sample.

[danshan11]

I dont see the relation to beating the closing line and estimating your edge. I dont know my edge, hence you cant use Kelly for sports betting, it is impossible to know your edge until the line closes and you compare the closing line implied to your implied probability from the bet.
I strive to beat the closing line consistently but currently I beat the closing line 80% of the time and about 65% of the time above the margin.
why are you estimating your edge, what does that tell you, I mean I guess when I model a line I am doing that,
example I do my little fancy math and I get Yankees -170 and the line comes out -140 and I get excited and click bet, I am assuming the line will move closer to -170 but I am not assuming it will go to close -170 that would not be smart to do because it does not happen often. so I dont technically calculate my edge but I know if I got -170 and its -140, 8 out of 10 times that will go my way and 6 or 7 times it will go enough to cover the margin or more! Does that make sense I hope it does, I ramble on like a 3 year old and have the writing skills of a 11 year old who skipped english class every day but Friday!

[danshan11]

you can even see here in my small sample I have been putting up my picks here on sbr that I am at about those numbers.

Date	Team	League	Close	Bet At	Bet Amt	Result	Line D	Notes	Potential	Est CLV%	Odds	Actual CLV
7/20	Yankees	MLB	104	-105	1.05	L		U8.5				-2.20%
7/20	Storm	WNBA	-110	-107	1.07	W		2.5				0.69%
7/20	Sky	WNBA	-114	-108	1.08	L		U181				1.35%
7/20	Sun	WNBA	-119	-107	1.07	W		U177				2.65%
7/21	Calgary	CFL	-104	-106	1.06	L		-19.5				-0.48%
7/21	Cardinals	MLB	-106	110	0.91	L						3.84%
7/21	Brewers	MLB	104	102	0.98	W		U8				-0.49%
7/21	Mariners	MLB	-115	-105	1.05	W		U9				2.27%
7/21	Atlanta	MLS	-219	-238	2.38	W						-1.76%
7/21	Liberty	WNBA	-120	-107	1.07	W		O163				2.85%
7/22	Athletics	MLB	108	111	0.90	L		U8				0.68%
7/22	Astros	MLB	-119	-120	1.20	L						-0.21%
7/22	Dream	WNBA	-116	-108	1.08	L		O163				1.78%
7/23	Phillies	MLB	-110	-100	1.00	L		U8.5				2.38%
7/23	Marlins	MLB	-113	-100	1.00	L		U8				3.05%
7/23	Nationals	MLB	-110	-104	1.04	L						1.40%

[danshan11]

this small sample has a very slight edge assuming an avg of 1.1ish margin because different types have slightly different margins!

[danshan11]

you can check out the spreadhseet here if you want to

https://docs.google.com/spreadsheets/d/14fWTAdsOeDJMzOfAeA8SdF_ecn1j8hXRwztGtcNiw-4/edit?usp=sharing

I am not trying to showcase my bets I am trying to show you that edge has nothing to do with wins and figuring your edge especially some edge involving winning is bad business. you think the books figure their edge on winning percentage of course not, they make their money on margin and in turn so should we, books dont kick you out for winning they kick you out for beating the line. you win a 1000 times and never beat the line on popular sports (not some 4th div scrub stuff) and the books will buy you dinner and offer you more signup bonuses, you beat their line more than enough for them to see its not deviation and you are gone, it dont take 1000 beating the lines to get booted!

[HeeeHAWWWW]

example I do my little fancy math and I get Yankees -170

So, you've estimated a probability of the outcome. Multiply it by odds, you have an estimated edge.

You're getting a bit confused here I think :-)

[Bsims]

The issue here is the definition of edge. danshan11 defines it as how well his computed line compares to the closing line, hence can't use Kelly until the game starts.

HeeeHAWWWW uses the most common definition , "Edge is decimal odds * probability, so 100/105 +1 * 0.53 =1.035". Both of these numbers are available throughout the day, hence he can use Kelly.

danshan11 did make a couple of interesting observations. "I guess my true argument is why add another variable to our job of trying to minimize and account for variables, less variables we have as bettors the clloser we get to winning". Kelly does add another variable to the betting experience which can be good or bad.

Also, his bet sizing approach (fixing the to win amount) is a bit unusual. It actually has some Kelly like properties.

[danshan11]

So, you've estimated a probability of the outcome. Multiply it by odds, you have an estimated edge.

You're getting a bit confused here I think :-)

yes but it has nothing to do with previous wins and losses, go back and read Joe thinks that his edge has something to do with his previous wins and losses and that is not a very good way to estimate your future edge.

[danshan11]

The issue here is the definition of edge. danshan11 defines it as how well his computed line compares to the closing line, hence can't use Kelly until the game starts.

HeeeHAWWWW uses the most common definition , "Edge is decimal odds * probability, so 100/105 +1 * 0.53 =1.035". Both of these numbers are available throughout the day, hence he can use Kelly.

danshan11 did make a couple of interesting observations. "I guess my true argument is why add another variable to our job of trying to minimize and account for variables, less variables we have as bettors the clloser we get to winning". Kelly does add another variable to the betting experience which can be good or bad.

Also, his bet sizing approach (fixing the to win amount) is a bit unusual. It actually has some Kelly like properties.

it is a vary Kelly like system for sure, you are now technically using the win probability to determine bet size but in a common sense approach it basically is bet more on things more likely to win and bet less on things less likely to win, makes common sense on the surface, MY BIG HUGE issue with any varying stake method is it can mess up your CLV. example
I bet 200 on the yanks and get 1% net line value
I bet 100 on the Orioles and get 3% net line value
so now I say my CLV is 1.5% but really it is not right?
you got to see it needs to be weighed to get a true CLV now since we have varying bets think of it as stocks
we invest 1000 in GE and get 5% return
we invest 10000 in FB and get 10% return
you could avg them and say I got a 7.5% return but in reality you did not
you got 11000 at 7.5% you made 825
but you really made 1050 and really made 9.5% because of weight
does that make sense?

[danshan11]

the most important thing is we cannot miscalculate our edge, if we make even the slightest mistake in estimating our edge, we can possibly lose money, you got to make sure triple check and quadruple verify you actually have a long term edge or you are losing money!

[HeeeHAWWWW]

the most important thing is we cannot miscalculate our edge, if we make even the slightest mistake in estimating our edge .....

It's a curious (and fortunate) property of the Kelly formula that it's extremely robust to errors.


The problem is when you persistently over-estimate your edge. Admittedly, most fall into that trap.