A software engineer at Coinbase has created Stonkfly, an open-source project that wires a digital copy of a fruit fly's brain into a Bitcoin trading account. The experiment presents simulated price charts to the artificial neural network and rewards profitable trades with bursts of virtual dopamine.
The simulated brain contains 166,700 neurons and approximately 25.6 million connections, based on a connectome—a published wiring map of an adult male fruit fly's nervous system developed by neuroscientists. Rather than displaying price numbers, the system converts candlestick charts into small images split across two simulated eyes, where light-sensing cells read pixel colors to inform buy, sell, or hold decisions.
Profitable trades trigger 15 reward neurons, while losses activate two neurons that register an unpleasant signal. Trading fees are counted as losses. The system prevents the fly from borrowing or betting on falling prices, constraints that distinguish it from many algorithmic trading bots.
According to engineer Alex Wormuth, who published the code on GitHub alongside a public dashboard, the project has significant limitations. Wormuth stated that synaptic changes do not establish that the system learns to trade profitably.
The program operates on paper money by default, beginning with $100. Live trading mode caps each order at $10 and limits attempts to 24 per day. The public dashboard showed zero completed trades as of Saturday.
The core question, according to Wormuth, is not whether the simulated fly generates profits, but whether a mapped brain can rewire itself for a reward that natural evolution never provided.
