Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Figured you math dudes (and dudettes) could knock this one out of the park!

I've got a website that stores users' game scenario results. Each scenario record can have one of three possible results: Side 1 wins, Draw, or Side 2 wins.

To this end, I'm currently representing balance by showing the simple percentage of total victories for each result. Example:

Result Totals
  Side 1 - 5
  Draw - 1
  Side 2 - 4

Display
  Side 1    Draw    Side 2
   50%       10%     40%

but the astute reader will immediately note that using this methods means a scenario with 5 plays split 2-1-2 and a scenario with 100 plays split 40-20-40 will have the same apparent "balance".

What I would like to do is take into account the total plays for a scenario. Thereby, a scenario with a split of 41-20-39 will rank much higher than one splitting 2-1-2, even though the simple percentage is actually less balanced.

Does that make sense? Ideally, this would map to a scale of some sort, so scenarios could be given discrete balance ratings.

Thanks for any help you can provide.

share|improve this question
    
Do you consider a 50-0-50 split more or less balanced than a 0-100-0 split? –  mhum Feb 24 '11 at 5:55
    
I had not considered it before. But after a minute's thought, my layman's brain seems content with putting 50-0-50 and 0-100-0 in the same pot. Is that wrong? –  Andrew Heath Feb 24 '11 at 6:44
    
It's your model, you get to define it however you like. It helps clarify the problem a little bit for me. –  mhum Feb 24 '11 at 6:52

1 Answer 1

In fact, your measurement of balance with 2-1-2 and 40-20-40 are the same-each player has the same chance of winning. What is different is the assurance that the measurement is reasonably accurate. Under the usual (worrisome) statistical assumptions you can calculate a standard deviation on the number of player 1 wins as $\sqrt{Np(1-p)}$ where $N$ is the number of games and $p$ is the fraction won by player 1. So for 2-1-2 the standard deviation on the number of player 1 wins is $\sqrt{5\cdot0.4\cdot0.6}\approx 1$ So you would say player 1 wins $(40 \pm 20)\%$. In the case of 40-20-40 you would have a standard deviation of about 5, so you would say player 1 wins $(40 \pm 5)\%$. It's not that the second measurement shows more balance, but you have better reason to think it is close.

share|improve this answer
    
"assurance" or I suppose "confidence" was the angle I was looking for. thank you! –  Andrew Heath Feb 24 '11 at 8:01

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.