The Midnight Mahjong Algorithm: How I Optimized Luck with Monte Carlo and Whisky in London’s Dark Net

I don’t play mahjong. I model it.

By day, I write Python scripts that map tile transitions like Markov chains—each chi-square distribution of player behavior parsed through 10,000 simulated hands. By night, I’m on the dark net forums writing cyberpunk fiction where the dragon tiles glow under neon lights and the bamboo table whispers in binary silence.

My father taught me that fortune isn’t random—it’s calibrated. The RNG doesn’t control outcomes; we do. With MBTI player typologies (ENTP: inventor archetype), I cluster gamblers into behavioral cohorts: the cautious classicist, the high-variance thrill-seeker, the whisky-fueled risk-taker who plays past midnight.

I’ve optimized win probabilities to 92% by removing superstition—not by changing rules, but by mapping cognitive biases. A “flat hup” (simple set) has higher expected value than “qingyise” (13-tile monster), which demands higher variance but collapses under pressure—like over-leveraged derivatives.

We track trends: when players chase streaks too long, they enter tilt cycles. We pause. We breathe. We recalibrate.

Last weekend at Babakan Art Centre, I presented this as live theatre: a blackjack of ancient ritual meets quantum probability. No deities here—just entropy, code, and cold brews at 2 AM.

You think you’re chasing luck?

You’re just missing the algorithm.