I recalled this "trick" I read in some magazine years ago and thought I'd try to explore it a bit more.
You have standard deck of 52 cards. The cards are on the table, facing up. Your friend secredly selects a number, $n_0$, from 1 to 13. Then your friend lifts $n_0$ cards one by one. Then he selects the value of the $n_0$th card, say $n_1$, as his new number and lifts again $n_1$ cards and takes the value of the $n_1$th as the new value. This continues until all the cards have been lifted. Let the value he has after the final card be $n_v$. All this is done silently.
The trick is, the magazine stated, that you can, with somewhat high probability, predict his final number. The idea is to choose your own number, $n_0^*$, and do the above process yourself as your friend does his counting. There is, apparently, high probability that your final value is the same as his. Not a very good trick, in my opinion.
I did some simulations. It seems that the probability that the values are the same is a bit lower than $0.7$.
How should one analyze this kind of a situation and what is the correct probability?
I remember the article mentioned Markov chains, but I cannot see how I could use them in this case. The process is finite and I don't think this has the Markov's property. I tried constructing some large transition matrix, but nothing came of it.
I tried some simplifications, like using only three cards with numbers 1,2 and 3, but the calculations become very messy rather quickly.
Edit: I did some more simulations, 200,000 runs for each pair $\{n_0,n_0^*\}$ and got the following matrix for probability of same result, rounded to save space.
$\left( \begin{array}{ccccccccccccc} 1. & 0.7 & 0.7 & 0.7 & 0.69 & 0.69 & 0.69 & 0.68 & 0.68 & 0.68 & 0.67 & 0.67 & 0.67 \\ 0.7 & 1. & 0.7 & 0.69 & 0.69 & 0.69 & 0.68 & 0.68 & 0.67 & 0.67 & 0.67 & 0.66 & 0.66 \\ 0.7 & 0.7 & 1. & 0.69 & 0.69 & 0.68 & 0.68 & 0.67 & 0.67 & 0.67 & 0.66 & 0.66 & 0.66 \\ 0.7 & 0.69 & 0.69 & 1. & 0.68 & 0.68 & 0.67 & 0.67 & 0.67 & 0.66 & 0.66 & 0.65 & 0.65 \\ 0.69 & 0.69 & 0.69 & 0.68 & 1. & 0.68 & 0.67 & 0.67 & 0.66 & 0.66 & 0.65 & 0.65 & 0.65 \\ 0.69 & 0.69 & 0.68 & 0.68 & 0.68 & 1. & 0.67 & 0.66 & 0.66 & 0.66 & 0.65 & 0.65 & 0.64 \\ 0.69 & 0.68 & 0.68 & 0.67 & 0.67 & 0.67 & 1. & 0.66 & 0.66 & 0.65 & 0.65 & 0.64 & 0.64 \\ 0.68 & 0.68 & 0.67 & 0.67 & 0.67 & 0.66 & 0.66 & 1. & 0.65 & 0.65 & 0.64 & 0.64 & 0.63 \\ 0.68 & 0.67 & 0.67 & 0.67 & 0.66 & 0.66 & 0.66 & 0.65 & 1. & 0.64 & 0.64 & 0.64 & 0.63 \\ 0.68 & 0.67 & 0.67 & 0.66 & 0.66 & 0.66 & 0.65 & 0.65 & 0.64 & 1. & 0.64 & 0.63 & 0.63 \\ 0.67 & 0.67 & 0.66 & 0.66 & 0.65 & 0.65 & 0.65 & 0.64 & 0.64 & 0.64 & 1. & 0.63 & 0.62 \\ 0.67 & 0.66 & 0.66 & 0.65 & 0.65 & 0.65 & 0.64 & 0.64 & 0.64 & 0.63 & 0.63 & 1. & 0.62 \\ 0.67 & 0.66 & 0.66 & 0.65 & 0.65 & 0.64 & 0.64 & 0.63 & 0.63 & 0.63 & 0.62 & 0.62 & 1. \\ \end{array} \right)$
Actually, as the order doesn't matter, I simulated only cases with $n_0^* > n_0$ and copied the results to the lower triangle. The diagonal elements are all one, for if both players select the same starting number, the results are identical.
If I calculated it correctly, for the elements outside the diagonal, a 95% two-sided confidence interval is $\text{value}\pm 0.00219131.$ In any case, the best strategy seems to be to select $n_0^* = 1$.