Coinbase Engineer Deploys Simulated Fly Brain to Trade Bitcoin
A new open-source project called Stonkfly uses a model of an insect's nervous system to execute trades, though creators admit no profitable strategy has been proven.
Alex Wormuth, a software engineer at Coinbase, has introduced "Stonkfly," an open-source project that utilizes a simulated male fruit fly brain and ventral nerve cord to trade cryptocurrency, as The Register first reported. According to the report, Wormuth provided the simulated brain with $100 to trade bitcoin, using engineered interfaces to feed market data into the system and translate neural activity into buy and sell decisions on Coinbase.
The system is designed so that dopamine neurons are stimulated when the fly generates a profit. The network is substantial in scale, featuring 166,700 nodes, 124,177,617 synaptic contacts, and 25,582,938 directed connections. This differs from a previous experiment by Eon Systems researchers, who used a female fly brain scan to control a virtual body in a virtual world.
Despite the technical complexity, the project's GitHub documentation includes significant caveats. The documentation notes that synaptic changes within the model "do not establish that it learns to trade profitably." Furthermore, the project explicitly states that no live-funded performance, biological replication, or strategy improvement has been demonstrated through its tests. The documentation warns that simple exposure baselines must be used for comparison, as rising cryptocurrency prices can make any buyer appear skilled.