Who pays the vig !

[greg]

j.r. miller on his web site states that winners pay vig not the losers. While this guy is a pro, it seems to me that while all wagerers risk the vig and that remains true for the winning wagerer only the losing wagerer actually pays the vig. assuming were not talking semantics here what is correct ?

[thedevil]

Greg,

He's wrong....

Both the winner and loser pay the vig.....

The loser pays laying 11/10...

The winner pays by getting only 10/11 not 11/11.

THE DEVIL

[greg]

thanks devel, however how does the winner pay? yes they both risk vig. If they both actually pay then the vig would end up being way higher than the 4.55% or the9.1 % from the winner as jrmiller writes or 9.1% from the loser as I contend.

[wintermute]

Depending on how you argue your case you can make it look as if either the winner pays all the vig or the loser pays it all.

Case 1: Loser pays.

The book says to you. You wager $100 to win $100 on the Titans. You also leave with us a handling fee of $10. If you win, you get back your wager, the handling fee and the win. If you lose we keep your wager and the handling fee.

Case 2: Winner pays

The book says to you. You wager $110 to win $110 on the Titans. If you win we withhold a handling fee of $10. If you lose there's no handling fee.

JR Miller has a very good site and is very knowledgable about most things but he and everyone else in the world that thinks that either one side or the other pays a disproportionate share of the vig is wrong.

The Devil is right.

'mute

PS JR also thinks that proportionate betting a la Kelly is suicidal. Not true.

Ronbets

You see I do post when something of interest to me comes up.

[wintermute]

Depending on how you argue your case you can make it look as if either the winner pays all the vig or the loser pays it all.

Case 1: Loser pays.

The book says to you. You wager $100 to win $100 on the Titans. You also leave with us a handling fee of $10. If you win, you get back your wager, the handling fee and the win. If you lose we keep your wager and the handling fee.

Case 2: Winner pays

The book says to you. You wager $110 to win $110 on the Titans. If you win we withhold a handling fee of $10. If you lose there's no handling fee.

JR Miller has a very good site and is very knowledgable about most things but he and everyone else in the world that thinks that either one side or the other pays a disproportionate share of the vig is wrong.

The Devil is right.

'mute

PS JR also thinks that proportionate betting a la Kelly is suicidal. Not true.

Ronbets

You see I do post when something of interest to me comes up.

[wintermute]

Sorry - I posted once and it didn't come up so I tried again. Then both appeared.

[irish]

Any notion of common math sense tells you the winner pays the vig.,Listen to how the 11/10 bet takes place.Lay 110 to win 100= 200 returned to winner,looser lays 110 =0 returned, extract the broker,both lay110 ,returned to winner is 220.i.e the looser always looses 110 the winner makes 220 except when there is a broker involved then it costs him the 10 or 4.55% VIG

[AussieVamp2]

'mute

got an optimal number of bets or some sort of modified Kelly plan then?

For instance, if I am looking at 500 soccer games on a weekend, and find 48 of them of interest, all at the same level, and said Kelly formula says 'bet 3% of stake' then you are in a bit of trouble

It obviously has to be modified to be workable.

[greg]

ok irish, if i book a 110 wager on team a and that is the only action i have on this event my client wins then i collect the vig out of his winnings henceforth my vig is in the form of a reduced payout. On the other hand if my client loses then i collect zero vig? or do i not collect vig also from him in that he risked 110 to win 100 on a even money prop.

[This message has been edited by greg (edited 07-04-2000).]

[zippy]

I came across that claim of Miller’s years ago. It seemed wrong to me then, and it seems wrong to me now. It is indeed just a matter of semantics. Consider it an $11 bet where the winner gets shorted $1 of what he would have won if there were no vig; or consider it a $10 bet where the loser pays an extra $1 above what he would have paid if there were no vig. Neither description is more accurate than the other.

[AussieVamp2]

or the bookmaker view

theoretically, in the long run they are brokers who give money from losers to winners and charge a handling fee, so to speak

[wintermute]

AV2

Kelly's formula applies as is only if you bet one game at a time. It has to be modified if you're betting more than one game simultaneously. Edward Thorp of blackjack fame worked out the math for making two wagers at a time. The formula says that you bet slightly less for each of the 2 wagers than you would bet optimally for one wager.

I doubt that there's any algebraic solution for the problem when you have more than 2 wagers but I did a numerical solution a couple of years ago. As you make more and more wagers the amount per wager continues to decrease. You come closer and closer to the point where you are betting all of your bankroll but you never quite get there.

For example, say that Kelly says you should bet 5 percent of your bankroll per wager. If you have twenty five wagers to make then the Kelly amount will drop to something of the order of 3.9 percent per wager.

People have problems with Kelly for two reasons. 1) they are too optimistic with their potential return. If you figure that you can win at a rate of 5 percent over time and bet accordingly and then achieve only three percent then your bankroll will dwindle quite dramatically. 2) they have no b-a-l-l-s. Even if you can call games at a rate of 5 percent, betting Kelly at 5 percent will lead to incredible fluctuations in your bankroll.

Even though JR says that Kelly is suicidal he follows it. He advocates increasing the size of your wager after you've reached a certain plateau. It's a very conservative Kelly but it's still Kelly.

If you're making 3 percent on soccer wagers and have on the order of 50 games per weekend - say 25 per day, I wouldn't bet more than 1 percent of capital per game. In fact I wouldn't bet more than 1 percent unless my return was getting up close to the 10 percent level.

'mute

[AussieVamp2]

[QUOTE]Originally posted by wintermute:
[B]AV2

Kelly's formula applies as is only if you bet one game at a time. It has to be modified if you're betting more than one game simultaneously.

-- Yeah, therein lies the trick.


I doubt that there's any algebraic solution for the problem when you have more than 2 wagers but I did a numerical solution a couple of years ago.


-- Not one I want to bother trying to find
Glad you have done this already.

As you make more and more wagers the amount per wager continues to decrease. You come closer and closer to the point where you are betting all of your bankroll but you never quite get there.

-- I guess you could do so some sort of formula approximation, but not fun to dig out when you have 2 minutes


For example, say that Kelly says you should bet 5 percent of your bankroll per wager. If you have twenty five wagers to make then the Kelly amount will drop to something of the order of 3.9 percent per wager.

-- Yeah, almost the whole lot.


People have problems with Kelly for two reasons. 1) they are too optimistic with their potential return.


-- Well, don't have that problem, I'm on the pessimistic side.

If you figure that you can win at a rate of 5 percent over time and bet accordingly and then achieve only three percent then your bankroll will dwindle quite dramatically. 2) they have no b-a-l-l-s. Even if you can call games at a rate of 5 percent, betting Kelly at 5 percent will lead to incredible fluctuations in your bankroll.

-- scary stuff for most.


Even though JR says that Kelly is suicidal he follows it. He advocates increasing the size of your wager after you've reached a certain plateau. It's a very conservative Kelly but it's still Kelly.

-- Of a sort, yes. Is fairly much what I have done.

If you're making 3 percent on soccer wagers and have on the order of 50 games per weekend - say 25 per day, I wouldn't bet more than 1 percent of capital per game. In fact I wouldn't bet more than 1 percent unless my return was getting up close to the 10 percent level.

-- No, I don't. Less even. Even less when beginning on new leagues, etc.

The even more of a pain part with this is the individual bets.

For example, one of the Swedish Division 2 groups had a bet that was looking to me from ratings like the price should be around 1.65-1.70, but was actually paying 2.20. Do we know if this 15+% edge on this game is real or not? No way to tell, and with a large number of games going on you could not bet the massive amount on each game, even with the approximation. One bad day, and see you later.

Then there is the game with the 3% edge which is sometimes indulged in for home teams. For those games out the edges of the distributions it is even harder, as the smaller samples lead to more uncertainty about what the realistic type percentage is. 10% of games are even rated and around a 50-50 shot for the home team.

Using the Tyrol-Innsbruck Admira Wacker Modling example from last night, best price on Innsbruck was 1.40. Rating had 2.65 goals. Now the right price on this might be 1.35, or 1.30 or even 1.25, hard to tell, as only about 1 in 200 games are in this range, and less than 200 game sin an Austrian season. So very hard to determine what the actual edge might be there - apart from the other factors in its favor like Austria reasonably consistent and one of the top teams versus a newly promoted team that was likely to get its head kicked.

A bit of an ugly problem indeed. One that for speed and perhaps psychological security is more easily handled with a set sort of betsize - and if you have 40 bets staggered over an 8 hour period or something, how do you recalculate the size of what you should be betting?

This is something the plateau thing handles for you as well. Get to a certain level, put it up a bit in actual size, keeping in a sane percentage range as you have suggested there.

[wintermute]

AV2

I don't remember the numbers but I ran some Kelly simulations where the individual game expectations ranged from about 1.02 to 1.20 with an overall average expectation of about 1.05. Betting a constant amount based on the average expectation gave you a smaller return than constantly adjusting the bet size but it was a fairly small difference.

'mute

[AussieVamp2]

'mute,

Thanks. Good to know. Saves work and a bit of peace of mind I guess