I am doing a problem from Mark Joshi's Quant Job Interview book, just to have some fun with probability.

The (paraphrased) question is as follows:

Suppose I throw a dart at a circular board of radius $R$. For the sake of the problem, we will assume that we always hit the board, but we are equally likely to hit anywhere on the board. Suppose you win a dollar if you hit 10 times in a row inside a radius of $R/2$, but each time you throw you have to pay 10 cents. If you throw 100 hundred times, how much money will you have won/lost in expectation?

My work and my question:

First, I am assuming that the count of dart throws inside a radius $R/2$ does not reset after being paid 1 dollar. In other words, if I threw 11 darts in a row inside a radius of $R/2$, I would make 2 dollars.

With our assumptions, the probability of hitting the board inside a radius of $R/2$ is $1/4$.
Let $X_i$ be the random variable that gives the payout of the game on the $i$th throw, where $10 \leq i \leq 101$ ($i$ ends at 101 since we are 1-indexing). So $X_i = 1-0.1 = 0.9$ with probability $(1/4)^{10}$ and $X_i = -0.1$ with probability $1-(1/4)^{10}$. Let $\tilde{X}$ be the random variable that describes the expected payout of the game. Then $$E[\tilde{X}] = E[X_{10} + \dots + X_{101}] = 91E[X_i].$$ Furthermore, we have $E[X_i] = (0.9)((1/4)^{10}) + (-0.1)(1-(1/4)^{10})$ so $E[\tilde{X}] \approx -9.099$. Now, we just have to add the cost of the first 9 throws where we deterministically lost 0.1 dollars, i.e. the total expectation approximately $(-9.099 + 9(-.1)) \approx -10$.

Is this analysis correct?

  • $\begingroup$ Seems plausible $\endgroup$
    – Henry
    Commented Jul 12, 2023 at 22:51
  • $\begingroup$ When you doubt your analysis, my advice (besides carefully justifying all your steps) is to try to simulate it. It's often a useful check, and with a good environment, can be done in a few moments. If you use R, replicate is a really handy tool for this $\endgroup$
    – Glen_b
    Commented Jul 13, 2023 at 4:13


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.