Bids are balloons: trading rules made visible
Zero-intelligence experiments ask how much of a market's apparent order comes from the trading mechanism. Gode and Sunder's 1993 double-auction experiments made that question especially clear: traders drawing random bids and asks under no-loss constraints could achieve high allocative efficiency in the environments they studied.
That result is a reason to inspect the rules carefully. The permitted prices, individual values and costs, matching procedure and trading horizon all matter. A browser experiment gives those ingredients a visible form, so a visitor can watch offers become trades and ask what produced the outcome.
FROM A TRADER RULE TO A VISIBLE OFFER
I built zi-traders-game and zi-traders-sim in September 2025. The game uses Phaser to put buyers and sellers on screen. Its archived history ends with the commit “Bids are baloons now”: offers become speech bubbles above the traders. That small visual choice makes it easier to follow an individual offer through the market.
The companion simulator uses React and Vite with a parameter-oriented interface. It provides another way to explore the setup and watch trading activity. Each build has its own implementation, so their code and exposed options deserve to be inspected separately.
The game's trader source includes unconstrained and constrained zero-intelligence modes, alongside ZIP traders that adapt a profit margin after observing market activity. That distinction matters: an adaptive trader is doing more than drawing a random offer. The projects are experiments inspired by the literature; a claim to reproduce a paper's result would need a specified configuration and a separate validation.
The useful questions are concrete. Which mode is running? What prevents a loss-making trade? What information reaches an agent after a transaction? How sensitive is the result to the setup? Following those questions from the controls into the source is the part of the exercise I want the site to make easy.