[genyes]

I am currently using a system to bet on soccer underdogs. I am betting very modest amounts at this time until I am very confident that the system is worthy of larger bets.
My results so far are: 30 wins and 54 losses (betting the same amount each time) with a 12.7% ROI. The wagers range from slight underdogs up to 4-1 underdogs.


I'm familiar with how to calculate Z-scores--except in this type of situation. How would you determine that the system is a good one i.e. worthy of your usual betting size?

Thanks

[the_mathman]

Do you bet on 1X2 market or in the AH ?

[bztips]

The variance formula for a single bet is:

Variance = betsize^2 * (decimal odds-1)

Compute this for all your bets, then add them up. Then take the square root of the sum to get the std. dev. for your z-score calculation.

[the_mathman]

Z score is not enough.
do you need to totalize a sample of at least 2000 bets distributed along the 12 month of the year.

do you can try to use historical data (e.g.:http://www.betexplorer.com/) for test your method (if is possible)

[genyes]

1X2

[Justin7]

I don't agree with mathman.

If you have a Z score of +2, that is very interesting. At +3, I would definitely bet it, especially if I understood and agreed with the logic behind it.

[Maverick22]

@Justin7 So you calculate this Z-Score. What does that end up telling you? What does it buy you?

[bztips]

@Justin7 So you calculate this Z-Score. What does that end up telling you? What does it buy you?

It tells you the probablility that the results you're getting are just by chance, given your current sample.

[the_mathman]

1X2

so, do you bet on underdog only or sometimes do you bet the X2 double chance (lay favorite)?

[the_mathman]

I don't agree with mathman.

If you have a Z score of +2, that is very interesting. At +3, I would definitely bet it, especially if I understood and agreed with the logic behind it.

unfortunately, the betting world isnt so matematically 'clear' and for validate a selection system do you need a big sample with match distributed along all the year.....

(this is my opinion, of course, based on my esperience on the soccer betting)

[brettd]

Z scores are useful when you're betting on the line. 1x2 with the large variety of odds that one would take odds at would complexify Z score taking. I guess you would sort of have to place all of your bets in odds 'brackets' and wait to you got a sufficient sample in any given odds bracket. But even that would be rough (depending on how narrow you would bracket odds together).

I have the same problem as most of my betting is 1x2. I'm going to resort to Monte Carlo simulation to come to a solution.

[Data]

I am currently using a system to bet on soccer underdogs... How would you determine that the system is a good

I just use my crap detector program. It tells me immediately that the system you described in the first sentence is crap.

[greva]

I just use my crap detector program. It tells me immediately that the system you described in the first sentence is crap.

How is this helpful?

[Peregrine Stoop]

I'd run a monte carlo and see where your results fall upon the distribution of possible outcomes

[Data]

How is this helpful?

That was an attempt to point out that the prerequisite for using z-scores is formulating a rational theory for the source of +EV. "An underdog system" is one of the more frequently used names for yet another variant of coin flipping. There are zillions of these systems and some of them have the Z-scores that are as high as you want. Those systems are still 0EV. Using valid methods in inappropriate ways guarantees drawing false conclusion. Hope this helps.

[specialronnie29]

I don't agree with mathman.

If you have a Z score of +2, that is very interesting. At +3, I would definitely bet it, especially if I understood and agreed with the logic behind it.

are you serious

didnt you write a book on gambling. how do you make any money when you believe this. if you look hard enough you can find a z-score of 20 by betting on teams with 5 vowels in their name on days when there is a full moon and your pet shit himself

[Justin7]

are you serious

didnt you write a book on gambling. how do you make any money when you believe this. if you look hard enough you can find a z-score of 20 by betting on teams with 5 vowels in their name on days when there is a full moon and your pet shit himself

I think you missed the second requirement: I understand and agree with the logic. If there isn't a logical basis, the z-score is mostly useless.

[specialronnie29]

no justin

you wrote this: If you have a Z score of +2, that is very interesting. At +3, I would definitely bet it, especially if I understood and agreed with the logic behind it.

you said you would bet anything at +3

[Justin7]

no justin

you wrote this: If you have a Z score of +2, that is very interesting. At +3, I would definitely bet it, especially if I understood and agreed with the logic behind it.

you said you would bet anything at +3

I stand corrected.

For the record, I would not bet on any Z score unless I had a logic that explained the results.

[specialronnie29]

the phrase z score should be banned from this subforum

half the shit people are applying it to isnt even normally distributed

[bztips]

the phrase z score should be banned from this subforum

half the shit people are applying it to isnt even normally distributed

Uh, Central Limit Theorem...

[the_mathman]

do you remember always that sports aren't (totally) casual events!

[specialronnie29]

Uh, Central Limit Theorem...

N = 74 from binomial distn

can argue about whether that N is large but most guys come on here with a 10-2 record and want to know if it means theyve found the holy grail;