Proposed alternative to middling/scalping

Salain · Oct 19, 2005

Everyone's heard of middling and scalping as profitable ways to bet both sides against the middle.

- Scalping is just a direct form of arbitrage and there really isn't much to say about it. A nice safe investment, or as safe as any money with a book can be. Winning % range is not a factor.

- Middling is based on percentages winning percentages in the 10 percentage points around %50. It's not as safe since you can lose money, but if you have deep pockets it's a reasonable long term strategy.

--------

What i haven't seen is a system based on highly variable factors from season to season that are relatively consistent within the system. (There are factors like this, you can check a thread I've got going in the NBA forums about a predictive program that hits %70 on 143 games for the '04 NBA season.) This program seems to hold up for the data I've entered so far on the '03 season, but it very well may not hold up overall...but I believe there's a way to bet where it won't matter.

The key is finding any system where you've got a greater than ~55% chance of winning OR losing in a significant number of games. You're looking for a factor that lets you decide games that is HIGHLY variable, it doesn't matter what direction the betting leads you in(lots of wins or lots of losses), so long as you can predict it will be extreme.

------
Taking this factor, you post up two bankrolls (at different sites) and kelly bet at %10 or so, one bankroll going with your indicator and the other going against it. You'll keep betting more on the positive side and less on the negative side.
-------

I haven't run any models on this but I don't really need to, so long as the variable takes a heavy favor to one side or the other you'll be profitable (perhaps EXTREMELY so). The enemy of this system is definitely mediocrity. The great thing about it is that you stand the potential (because of kelly betting at high %s) to make an extremely large amount of money. (if you hit 50% 9 in 10 times and 70% the other on 130 bets or so, you'd be up 500X your money, rough estimate) You also don't need to identify winners, just variables that are highly volatile between seasons and yet consistent within a season. They DO exist.

It's an approach to betting both sides of a game that I haven't seen around, and it opens up a new set of systems to profitability (volatile ones).

This is all new to me, what do you veteran betters think?

Fishhead · Oct 19, 2005

The above example is about as close to what I do for a living as you will find anywhere.

D2bets · Oct 19, 2005

Maybe I'm just dense today, but for the life of me I cannot figure out what you're talking about. Why would you bet both sides of the same game unless it was a profitable middle in itself. Are you saying you would take, let's say +8 at Book A for 3 units and then -8 at Book B for 1 unit? That can't be it. I think I have no clue what you're talking about. I must be dense today.

Salain · Oct 19, 2005

D2bets said:
Maybe I'm just dense today, but for the life of me I cannot figure out what you're talking about. Why would you bet both sides of the same game unless it was a profitable middle in itself. Are you saying you would take, let's say +8 at Book A for 3 units and then -8 at Book B for 1 unit? That can't be it. I think I have no clue what you're talking about. I must be dense today.

Yeah I'm not very good at explaining. Basically you pick one or a set of factors on which you're going to decide who the winner is going to be. For example alphabetical order, lowest team wins according to the system. You have looked at past years and know that betting this way may net you %70 wins, but you can't be sure that retrograde fitting will work...so what are you left with? The answer is a system that you know can give you stats that are way off %50 for a year...it's obviouslly a ridiculous system for choosing winners year after year, so you can't be sure it won't give you %30 next year. So you use this system, betting both sides of the game using kelly betting lets the side that wins more have the largest bets on them, making it profitable. What you're betting on here isn't being able to predict a winner correctly, it's being able to predict a stat stands a good chance to give you a non-%50 winning percentage.

Obviouslly of no use without those kind of stats.

CapnHandi · Oct 19, 2005

Very interesting Salain. I'm doing similar sorts of things with football. The only thing I'd caution you on is to not get overly excited about a system that "predicts" the results of last year's games. A good approach to take is to develop a model that seems to work for 2 to 5 years ago and then test it on last year's games.

GL

Salain · Oct 19, 2005

CapnHandi said:
A good approach to take is to develop a model that seems to work for 2 to 5 years ago and then test it on last year's games.

Yeah, would if I had the data...I designed it on '04-'05 and it gives 65% on the portion of '03-'04 I've inputted(through january), still I couldn't agree more about being very cautious about any optimism.

Think anyone around here has finals and closing lines from the NBA pre '03-04? That's as far back as I can find, will post asking for them in the NBA forum later. lunch now though.

D2bets · Oct 19, 2005

Salain said:
So you use this system, betting both sides of the game using kelly betting lets the side that wins more have the largest bets on them, making it profitable .

This is the part where you lost me. You are betting both sides of the same game? Are you only doing this when there are significant price differences between books to make the middle worthwhile? :icon_conf

TheNuts · Oct 19, 2005

Salain said:
------
Taking this factor, you post up two bankrolls (at different sites) and kelly bet at %10 or so,
-------

What do you mean by kelly betting at 10%. I have been using the kelly bet size calculator as a trial betting method for some sports. I thought there was only 1 formula, the one i have been using is:

Bet Size = [ (O*P - 1) * B ] / ( O -1 )

Where O is the odds available ie, -115 = 215/115
P is probability of winning
B is bankroll

What does the 10% mean?

royalfan · Oct 19, 2005

Not sure what confuses me more, the fact one thinks something like this can work when I cannot even understand it, or the fact fishead would bet both sides of a game at two places just for the sake of bettiing both sides of a game unless there is a scalp or middle opportunity which isnt what this system is all about.

This system makes no sense whatsoever. Here is why. Say Miami is playing Buffalo. Miami favored by 3 at pinny where one bankroll is and also favored by three at bettrojan and juice is -105 at each both ways. According to this nonsense you would have, say three times as much on Miami at one place than on Buffalo at the other. Thereforse you would have say 3150/3000 on Miami at pinny and 1050/1000 at trojan. This makes no sense to me whatsoever as you would be much better off just playing 2100/2000 at one place which essentially just turns it into a normal bet just like the normal kind. What am I missing here. I cannot believe Fishhead would do something that stupid so I need someone intelligent like Fishhead to explain further.

TheFunkMonkey · Oct 19, 2005

I'm confused too.

WorldNumberOne · Oct 19, 2005

and here i felt like the kid in the back of the class afraid to raise his hand...

now i see the class is full of kids with their hand up...

PokerKingJim · Oct 19, 2005

randyrohm · Oct 19, 2005

Fish I know you have your hands full but you are the one that can clear this thread up :icon_conf

.

Doug · Oct 19, 2005

I don't understand , either !

Cubsmac · Oct 19, 2005

im confused

trytrytry · Oct 19, 2005

He is onto something very valuable, but not describing it quite right. The key here is betting the factors in a Kelly type of criteria (I call them bins not factors but same idea) . You can be sure Fish will not comment anymore on this!

Salain said:
Everyone's heard of middling and scalping as profitable ways to bet both sides against the middle.

- Scalping is just a direct form of arbitrage and there really isn't much to say about it. A nice safe investment, or as safe as any money with a book can be. Winning % range is not a factor.

- Middling is based on percentages winning percentages in the 10 percentage points around %50. It's not as safe since you can lose money, but if you have deep pockets it's a reasonable long term strategy.

--------

What i haven't seen is a system based on highly variable factors from season to season that are relatively consistent within the system. (There are factors like this, you can check a thread I've got going in the NBA forums about a predictive program that hits %70 on 143 games for the '04 NBA season.) This program seems to hold up for the data I've entered so far on the '03 season, but it very well may not hold up overall...but I believe there's a way to bet where it won't matter.

The key is finding any system where you've got a greater than ~55% chance of winning OR losing in a significant number of games. You're looking for a factor that lets you decide games that is HIGHLY variable, it doesn't matter what direction the betting leads you in(lots of wins or lots of losses), so long as you can predict it will be extreme.

------
Taking this factor, you post up two bankrolls (at different sites) and kelly bet at %10 or so, one bankroll going with your indicator and the other going against it. You'll keep betting more on the positive side and less on the negative side.
-------

I haven't run any models on this but I don't really need to, so long as the variable takes a heavy favor to one side or the other you'll be profitable (perhaps EXTREMELY so). The enemy of this system is definitely mediocrity. The great thing about it is that you stand the potential (because of kelly betting at high %s) to make an extremely large amount of money. (if you hit 50% 9 in 10 times and 70% the other on 130 bets or so, you'd be up 500X your money, rough estimate) You also don't need to identify winners, just variables that are highly volatile between seasons and yet consistent within a season. They DO exist.

It's an approach to betting both sides of a game that I haven't seen around, and it opens up a new set of systems to profitability (volatile ones).

This is all new to me, what do you veteran betters think?

D2bets · Oct 19, 2005

trytrytry said:
He is onto something very valuable, but not describing it quite right. The key here is betting the factors in a Kelly type of criteria (I call them bins not factors but same idea) . You can be sure Fish will not comment anymore on this!

But what about this betting both sides of the same game notion? Is that where he didn't describe it quite right? Aside from that, it doesn't look like anything groundbreaking to me -- just betting on trends that appear profitable. I don't see how this is an alternative to middling/scalping. But again, I am dense sometimes, this is probably one of those times.

Salain · Oct 19, 2005

trytrytry said:
He is onto something very valuable, but not describing it quite right. The key here is betting the factors in a Kelly type of criteria (I call them bins not factors but same idea) . You can be sure Fish will not comment anymore on this!

THANKS

. Very glad I went ahead and posted this.

Santo · Oct 19, 2005

Where I struggle is the x< 45 or x > 55 range.. if you have a predictive system that meets either criteria, you're profitable (fading the x<45 obviously)

centextiger · Oct 19, 2005

betting both sides but different amount

Proposed alternative to middling/scalping

New member

Active member

New member

New member

winningtotals every day

New member

New member

New member

New member

New member

Self appointed RX World Champion Handicapper

Respect My Steez

New member

Rx God

New member

Member

New member

New member

New member

RX Fighting Tiger

Forum statistics

Share this page