In vitro market

MAS — Multi Agent Society — is a model Andrei built during his MSc in Physics of Complex Systems (University of Turin, 2023) to simulate cryptocurrency markets. Agents live on a Barabási-Albert network; every link carries a level of trust between 0 and 1. When two linked agents' profits move together, their trust grows; when they diverge, it decays.

Each day every agent decides, then the exchange clears all orders at once: ln P′ = ln P + μ · (buy$ − sell$) / depth, and everybody trades at the new price. Depth grows with the price, a circuit breaker caps a day at −50% / +100%, and shorts that lose their collateral are liquidated. The exchange caps daily short selling — unless you press deregulate. A crash is −30% within two weeks; a bubble is ×2 within two months.

The society is humanity: each agent stands for 8 billion ÷ agents people (40 million at 200 agents). The two default whales are 80 million millionaires — about what the real world has — and the fishes are 6.3 billion people with $1,000 each. Every dollar in the ledger is what those people actually won or lost.

• Fish — retail. Holds $HOPIUM or cash, never shorts. Buys when its trusted friends buy (all in, during FOMO), trims its bag when they sell, panics with the timeline after a rug pull. Drops friends below its trust threshold, sleeps.
• Shark — never trades randomly. Copies the positions of the most popular agents, but only when they agree with its own belief on the market (bullish, bearish, lateral).
• Whale — rich and semi-periodic. Accumulates in silence; once its bag is big it goes loud, recruits followers and posts about its position every day — and keeps posting while it sells that position to them (distribution), until its profit streak (WPS) is cashed in.