[VBOMBER]

I apologize in advance if this is not set up as clearly as one would like or a little repetitive, but any thoughts and comments are GREATLY APPRECIATED.

Scenario:
A capper posts a record of 1100-900 (55%) on the same type of essentially coin flip type bet (odds between even and -110, average of about -106) using the same handicapping methods and analysis.

So I understand how to come up with a Z-score, and determine it would be an extremely small probability that a 50% capper would see those results. But my first question is, how do you determine the Z-score and therefore the liklihood that a 51%, 52%, 53% and so on capper experiences these results?

Next Scenario:
Same capper above with his record over his first 2000 picks. Over his next 200 picks he goes 83-117 (41.5%) using the same handicapping methods and analysis. The capper has experienced bad losing streaks over much smaller samples in his original 2000 plays (such as 10-20, 12-29, etc.) but never over this large of a sample of plays.

My next question is, how do you figure out the liklihood of this happening? Now I believe you have to make some assumptions of the handicapper's ability to be able to figure out the answer, so maybe use the scenarios of a 48% through 55% capper. (I'm guessing this may be (exactly?) similar to my first question above and if shown once for one number can figure out the rest on my own)

General Questions:
1) For winning players with much larger samples than this, have you experienced such streaks and then still bounced back to winning levels? Or is this way out of the norm a capper should expect and it may be time to reevaluate everything and possibly ditch the method altogether? (Obviously looking into it much deeper regardless is something that should be done to see if there is a more obvious explanation).

2) When can one ever feel comfortable about the "%" capper they may be? This capper was 55% after 2000 plays (and pretty consistently throughout at that level over those plays), but dropped to 53.77% after adding just 200 more plays. This is obviously more theoretical and touches upon other questions that have been brought up in other recent threads, but curious if anyone has any thoughts.

Thanks again for any thoughts/comments/input!!!

[Ganchrow]

Scenario:
A capper posts a record of 1100-900 (55%) on the same type of essentially coin flip type bet (odds between even and -110, average of about -106) using the same handicapping methods and analysis.

So I understand how to come up with a Z-score, and determine it would be an extremely small probability that a 50% capper would see those results. But my first question is, how do you determine the Z-score and therefore the liklihood that a 51%, 52%, 53% and so on capper experiences these results?

A Z-score is calculated in general as folllows:

Z = (Actual Results - Expected Results) / (Standard Deviation)

"Actual Results" are self-explanatory.

"Expected Results" are the sum for each bet of bet size * (decimals odds * expected win probability - 1).

"Standard Deviation" is the square root of the sum for each bet of (bet size * decimal odds)^2 * expected win probability * (1 - expected win probability).

Next Scenario:
Same capper above with his record over his first 2000 picks. Over his next 200 picks he goes 83-117 (41.5%) using the same handicapping methods and analysis. The capper has experienced bad losing streaks over much smaller samples in his original 2000 plays (such as 10-20, 12-29, etc.) but never over this large of a sample of plays.

My next question is, how do you figure out the liklihood of this happening? Now I believe you have to make some assumptions of the handicapper's ability to be able to figure out the answer, so maybe use the scenarios of a 48% through 55% capper. (I'm guessing this may be (exactly?) similar to my first question above and if shown once for one number can figure out the rest on my own)

See this post for an example of how to perform this type of Bayesian analysis.