0
$\begingroup$

I'm struggling a bit with a problem -more specifically on how to tackle the problem- and came for help or pointers. I'm an engineer with very little education in statistics but I'm really motivated in getting a better understanding of the resolution of such problems.

I have two independent players playing darts and I would like to express the probability that player 1 is better than player 2 based on a single measurement which consists of the sum of N throws of each player. For the sake of simplicity, I assumed that the outcome of a measurement depends only on one single parameter and I can already simulate without problem the density function of a N-throws based on that parameter. Intuitively, I already see that players having the same parameter value can generate measurements that are, on average, 50% better than the other player (one wins a match, then the other wins it etc.).

What I would like to do now is to answer the question about the probability of player 1 being better than player 2, or conversely, the probability that they are equally good based on a single match result.

How should I approach this problem ? I'm not necessarily looking for an analytical form as long as I can find a method to populate a table giving confidence in the assertion based on the two scores that can only take bounded values between 0 and 10*N.

$\endgroup$

1 Answer 1

0
$\begingroup$

A simple way to proceed is to do an hypothesis sign-test that is a binomial test. $H_0: p=0.5$ against $H_1: p>0.5$

You reject $H_0$ if your fixed p-value is less than a fixed %

For example, if you have $N=10$ throwns with the following winners

$$\{x,x,y,x,x,y,x,x,x,x\}$$

You set 1 if x and 0 if Y thus your result is 8 and being the probability that a $Bin(10;0,5)\geq 8=0.0547$ you can reject the hypothesis that $X\leq Y$ at any significance level greater than 5.47%

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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