[ChuckyTheGoat]

This thread has potential, but it's going in different directions. When I say ROI, I'm talking about your Estimated ROI based on your projection. Figure out what that # is. +10% ROI play > +5% ROI play.

I'll introduce a concept of Risk-Adjusted Return. In financial models, there's normally a tradeoff btw Risk + Reward. Take a Moneyline sport like MLB. Consider a -200 Fav which you project to win 70% of the time. Now, consider a +200 dog which you project to win 35% of the time.

Each has +5% ROI. On the fav, you clear +10 for each 200 Risked. On the dog, you clear +5 for each 100 Risked.

Are the two risks equivalent? My answer is a clear No. The -200 Fav is a MUCH better risk. I say this b/c the Fav can handle the swings better. You'll rarely run into a hugely negative run on the -200 Fav. On the +200 Dog, you can easily run into a string of bad results and go broke (or nearly broke).

[jbayko]

This thread has potential, but it's going in different directions. When I say ROI, I'm talking about your Estimated ROI based on your projection. Figure out what that # is. +10% ROI play > +5% ROI play.

I'll introduce a concept of Risk-Adjusted Return. In financial models, there's normally a tradeoff btw Risk + Reward. Take a Moneyline sport like MLB. Consider a -200 Fav which you project to win 70% of the time. Now, consider a +200 dog which you project to win 35% of the time.

Each has +5% ROI. On the fav, you clear +10 for each 200 Risked. On the dog, you clear +5 for each 100 Risked.

Are the two risks equivalent? My answer is a clear No. The -200 Fav is a MUCH better risk. I say this b/c the Fav can handle the swings better. You'll rarely run into a hugely negative run on the -200 Fav. On the +200 Dog, you can easily run into a string of bad results and go broke (or nearly broke).

What you're describing in this post is essentially the main thrust of the Kelly Criterion. It will recommend higher stakes for opportunities having a higher win probability compared to other opportunities which share the same Expected Value (or ROI) but a lower win probability.

[danshan11]

isnt that what my fixed profit staking method does and one of the main benefits of it?

[tsty]

Maybe you guys misunderstand me.

Whether you "know" your edge on the bet or not, surely increasing your risk amount blindly based on the odds, ie: betting 130 to win 100 or 100 to win 130 anytime the odds are -130 or +130, is not mathematically sound.

You guys seem to saying (assuming?) it is. Do you have numbers to support that?

If the edge is the same then you bet more. It's fairly intuitive.

However that had nothing to do with my post. I was just stating that betting without knowing your edge is retarded. aka system bettors

[tsty]

Also can people stop misusing ROI

If sportsbetting roi was 5% then nobody would be doing it

[Optional]

If the edge is the same then you bet more. It's fairly intuitive.

However that had nothing to do with my post. I was just stating that betting without knowing your edge is retarded. aka system bettors

Thats a co-incidence. Thought your post had zero to do with what I said too!!

You still don't seem to get it.

So you blindly bet 30% over your base unit anytime odds are -130? That is true eh?

I think it's "fairly intuitive" that doing that is -EV

Do you have any math to back up your guess that it is not?

[HeeeHAWWWW]

I'll do a sim later, but already know the results: staking X/(decimal odds-1) will beat flat staking X, and Kelly will beat either of them.

[danshan11]

I'll do a sim later, but already know the results: staking X/(decimal odds-1) will beat flat staking X, and Kelly will beat either of them.

how do you calculate kelly when you dont know your edge before the line closes?

[HeeeHAWWWW]

Here we go, 100k sim runs (-200 to +200, 5% edge, 1k total plays, starting with $10k)

Percentile	5%/(odds-1)	Flat 5%	Flat 2.5%	Flat 7.5%	Kelly (2%-8% EV)
0.100%	$409	$114	$1,547	$4	$398
1.000%	$1,156	$437	$2,993	$32	$1,077
5.000%	$2,781	$1,347	$5,203	$166	$2,743
10.000%	$4,429	$2,489	$7,068	$423	$4,511
25.000%	$9,832	$6,876	$11,783	$1,989	$10,232
33.333%	$13,461	$10,326	$14,407	$3,662	$14,252
50.000%	$23,440	$20,797	$20,430	$10,592	$26,027
75.000%	$56,885	$63,380	$35,406	$55,465	$66,480
90.000%	$128,620	$174,186	$58,645	$260,251	$158,717

I threw in flat 7.5% just to show quite how badly over-staking does.

Variable staking will be even more superior relative to flat if you bet bigger dogs, as will Kelly (the gap grows very big quite quickly).

[danshan11]

variable staking is what I do?
1.xx to win 1 and .XX to win 1 dogs

and for Kelly what did you use as your edge?

[Optional]

Here we go, 100k sim runs (-200 to +200, 5% edge, 1k total plays, starting with $10k)

Percentile	5%/(odds-1)	Flat 5%	Flat 2.5%	Flat 7.5%	Kelly (2%-8% EV)
0.100%	$409	$114	$1,547	$4	$398
1.000%	$1,156	$437	$2,993	$32	$1,077
5.000%	$2,781	$1,347	$5,203	$166	$2,743
10.000%	$4,429	$2,489	$7,068	$423	$4,511
25.000%	$9,832	$6,876	$11,783	$1,989	$10,232
33.333%	$13,461	$10,326	$14,407	$3,662	$14,252
50.000%	$23,440	$20,797	$20,430	$10,592	$26,027
75.000%	$56,885	$63,380	$35,406	$55,465	$66,480
90.000%	$128,620	$174,186	$58,645	$260,251	$158,717

I threw in flat 7.5% just to show quite how badly over-staking does.

Variable staking will be even more superior relative to flat if you bet bigger dogs, as will Kelly (the gap grows very big quite quickly).

Can you explain the significance of the percentile ranges for me please? And whats going on when betting flat 2.5% to produce such a different result up to 50%?

[danshan11]

I agree I am just dumb or something but I dont understand the chart either

[Miz]

Also can people stop misusing ROI

If sportsbetting roi was 5% then nobody would be doing it

I disagree. How do you define ROI in your world?

i view it as total profit divided by total cumulative number of dollars risked, meaning that if you profited $1000 after 20 bets where you risked $1000 on each bet the ROI of that series of bets is 5%. I still calculate it that way even if my bankroll were $10k.

I view the denominator in terms of total amount risked irrespective of time although everyone is more familiar with annualized ROI

[Miz]

do you view it as (ending bankroll - starting bankroll ) / (Starting bankroll) ?

[Optional]

I agree I am just dumb or something but I dont understand the chart either

We don't need to be dumb to ask a question. End up dumber if we don't.

I assume it's how often you will come out with that result after 100k attempts.

I just don't really understand why he chose .1%, 1%, 5% and made larger steps later. But I'm sure there is a reason.

[danshan11]

heehaw please come back and explain.

[tsty]

do you view it as (ending bankroll - starting bankroll ) / (Starting bankroll) ?

Yes that's how it works

otherwise it's yield

[tsty]

how do you calculate kelly when you dont know your edge before the line closes?

this is wrong

[jbayko]

Also can people stop misusing ROI

If sportsbetting roi was 5% then nobody would be doing it

I don't agree that we're misusing the term. The stat which everyone is using here is analogous to ROI as used in finance, not yield. I realize that some people have misappropriated the term "yield" in sports betting, and I can't figure out why they've done that.

But I have a bigger problem with your second statement since most bettors lose money in the long term yet they keep on betting....

[jbayko]

how do you calculate kelly when you dont know your edge before the line closes?

If you don't know your edge until the line closes, then how (why) do you ever bet?

[tsty]

I don't agree that we're misusing the term. The stat which everyone is using here is analogous to ROI as used in finance, not yield. I realize that some people have misappropriated the term "yield" in sports betting, and I can't figure out why they've done that.

If I bet 50 to win 50 and lose what is my roi?

[HeeeHAWWWW]

Can you explain the significance of the percentile ranges for me please? And whats going on when betting flat 2.5% to produce such a different result up to 50%?

Ahh, sorry - the choice of percentile ranges is simply that those are defaults in the script, they're not particularly meaningful here.

Flat 2.5% is doing well at the "unlucky" end of possible results because it's more conservative than the others - you are trading off top end growth, for a bit of protection from the most severe downturns in the bottom half.

[danshan11]

this is wrong

Please explain how this is wrong, thanks for the replies

[danshan11]

If you don't know your edge until the line closes, then how (why) do you ever bet?

I am saying overall I know I have an edge in the sport or bet type and if I did not I would not bet. What I do not know is which bets will have an edge and which wont and exactly how much that edge is per bet.

Bet 1-1000 avg edge 1%
bet 1001 edge???

[Bsims]

These techniques size the bets based on the odds and are not dependent on estimating an edge at the time of the wager. Kelly and similar are based on having confidence in estimating the edge.

If you feel you've identified a good wager but won't know your edge until the lines close, then the former is for you.

Once again, it you don't know your edge at the time you make a wager, keep betting the way you are. Forget Kelly.

[tsty]

Once again, it you don't know your edge at the time you make a wager, keep betting the way you are. Forget Kelly.

if you dont know your edge then dont bet at all

its just gambling at that stage

[danshan11]

if you dont know your edge then dont bet at all

its just gambling at that stage

I think we are not on the same page of what an edge is

this is an edge to me
Bet Yankees -164 they close -184 that is an edge
this is not an edge to me
bet yankees -164 they close -164 that is not an edge
this is an edge but not enough to beat the margin
bet yankees -164 they close -170 that is an edge but not large enough to beat the margin

when I bet I dont know the closer so I dont know my edge, how could I?

[Optional]

I think we are not on the same page of what an edge is

this is an edge to me
Bet Yankees -164 they close -184 that is an edge
this is not an edge to me
bet yankees -164 they close -164 that is not an edge
this is an edge but not enough to beat the margin
bet yankees -164 they close -170 that is an edge but not large enough to beat the margin

when I bet I dont know the closer so I dont know my edge, how could I?

Beating the closer is good.

But I think most people talking edge in reference to kelly betting mean you have your own model and produce your own estimated lines and look at that v the current book line to calculate edge. Before you bet.

You may prove to have had an edge in hindsight by seeing you beat the closer, but that does not inform your bet size pre-bet.

[Optional]

Ahh, sorry - the choice of percentile ranges is simply that those are defaults in the script, they're not particularly meaningful here.

Flat 2.5% is doing well at the "unlucky" end of possible results because it's more conservative than the others - you are trading off top end growth, for a bit of protection from the most severe downturns in the bottom half.

Thanks. So basically it says that you will lose less during a bad streak using 5%/(odds-1) but will win more during a lucky streak with flat betting.

And on average 5%/(odds-1) will beat flat betting by about 10%.


Thanks for setting me straight. I thought it was almost opposite to that.

[danshan11]

first your example of an edge must be a joke, how can anyone use the line their model makes to base their bet size on, now that would be gambling (injuries, weather, and so many more things) that is the craziest method I have ever heard. Pinnacle has the best model in the business and they are not even close to the closing line from where they open.

I dont want to get deep on how kelly is dumb as penetrate but anyway its dumb as penetrate cause you do not know your edge not even close
my method you are at least close to the closing line.

[danshan11]

and heehaws chart,
A I have no idea what it means but how would he know how much to bet from past results when all you see is the closing line. we dont bet the closing line very often not close if we did we would be sunk anyway. Maybe he can explain the chart better but if its using the closing lines to adjust bet size it is perfect unfortunately we dont know the closing line when we bet if we did we would be rich

[danshan11]

I bet 41 to try to win 25 with the yankees
when it closes
it costs 46 to try and win 25 with the yankees

this is edge my line was -199 on my model and I just do not see the way of ever thinking the line I make has anything to do with any reality in anyway shape or form.

[Optional]

first your example of an edge must be a joke, how can anyone use the line their model makes to base their bet size on, now that would be gambling (injuries, weather, and so many more things) that is the craziest method I have ever heard. Pinnacle has the best model in the business and they are not even close to the closing line from where they open.

I dont want to get deep on how kelly is dumb as penetrate but anyway its dumb as penetrate cause you do not know your edge not even close
my method you are at least close to the closing line.

Ok, you don't believe it is possible to produce your own market more accurate than the book's openers.

Fine.

But beating the closing line isn't a "staking method" that the thread was supposed to be about. So not sure why you are 1) comparing the two things and 2) getting so angry about it.

Some people really do think they can cap a market more accurately.

Also, there are a lot of soft sports out there. Beach Volleyball for example. Do you consider it impossible to cap that better than the books?

[tsty]

Ok, you don't believe it is possible to produce your own market more accurate than the book's

if you believe this then the answer is easy

just dont bet

[danshan11]

first off I am not mad at all, LOL just how I talk I guess sorry if you got that impression, was not my intent at all
second off soft sports yes tons of them I mean tons of them but the margin is so high beating them is just as hard if not harder but we are talking about staking method in this thread not market effeciency.

I am saying using a model to stake is a mistake, I am in no way trying to combine the 2.