Coinbase Engineer Turns Fly Brain Into Bitcoin Trader

A Coinbase software engineer has connected a simulated fruit fly brain to a Bitcoin trading system, creating an unusual experiment that combines neuroscience, cryptocurrency and algorithmic trading.

The open-source project, called Stonkfly, was developed by Alex Wormuth and published on GitHub alongside a public dashboard. Rather than feeding the system Bitcoin price data directly, Stonkfly presents candlestick charts visually and rewards profitable trading decisions with simulated dopamine.

The digital brain is based on the connectome of an adult male fruit fly, a detailed map of its nervous system created by neuroscientists. The simulation contains about 166,700 neurons and approximately 25.6 million connections.

Stonkfly displays a small Bitcoin price chart across the fly's two simulated eyes. Light-sensitive cells process the chart's raw pixels before the neural network produces one of three possible decisions: buy, sell or hold. The virtual fly never receives numerical price information.

Profitable trades activate 15 reward neurons, mimicking a dopamine response, while losing positions activate two neurons associated with negative feedback. Trading fees are also treated as losses. The system cannot use leverage or short Bitcoin.

Wormuth has cautioned against interpreting the experiment as proof that a simulated brain can become a successful crypto trader. He noted that changes in synaptic connections do not demonstrate that Stonkfly has learned to trade profitably.

The software operates with paper money by default and begins with a virtual balance of $100. A live-trading option is available, although individual orders are capped at $10 and the system can make no more than 24 trading attempts per day. As of Saturday, the public Stonkfly dashboard showed no completed trades.

The experiment arrives as Bitcoin trades well below its previous peak. BTC was near $77,286 on Saturday, gaining about 0.6% on the day.

Whether Stonkfly ultimately makes money may be less important than the experiment itself. The project explores whether a digitally reconstructed biological brain can adapt its neural connections when exposed to an artificial reward system built around Bitcoin trading.