Coinbase Engineer Tests Simulated Fruit Fly Brain for Cryptocurrency Trading Decisions

A Coinbase engineer has developed an unusual experiment using a simulated fruit fly brain to make cryptocurrency trading decisions. The project, called Stonkfly, is an open-source model that runs on real BTC-to-USDC data and uses dopamine signals to reinforce profitable trades.

The simulator contains about 166,700 neurons and 25.6 million connections, which process visual market information before sending it through the simulated nervous system. The model can choose to buy, sell or hold an asset, receiving rewards or penalties based on changes in its portfolio.

Stonkfly cycles through Bitcoin, Ethereum, and Solana, converting chart images into signals representing brightness and color. The project's developers warn that the experiment is not a proven trading system, but rather an exploration of biologically inspired computing and cryptocurrency trading.