Pick any three-flip pattern β HHH, say. I'll pick mine after you, we flip a coin until one pattern shows up, and that person wins. Sounds fifty-fifty, and it sounds like YOU have the edge going first. Peel it: whatever you pick, I can always pick one that beats it. There is no best choice β the order is a loop.
This is Penney's game, and it is non-transitive: for every length-3 pattern there is another that beats it more than half the time, so 'best pattern' does not exist β like rock-paper-scissors. MICHEALE always answers your call XYZ with (Β¬Y)XY, the known optimal counter, then the page runs 2000 live coin races and reports the win split β it lands between 2:1 and 7:1 in MICHEALE's favour, every time. Click a chip to change your call.
The 'bar bet', going-first-feels-safer patter, and MICHEALE as the hustler are the costume; the coin is a fair 50/50 Bernoulli underneath.
The empirical split is a simulation (2000 fair flips per race) β it converges to, but is not, the exact rational odds; and 'you always lose' is the follower's advantage with optimal play, not a guarantee for a careless second-picker.