Stonkfly Project Overview

Coinbase software engineer Alex Wormuth has developed a digital experiment called Stonkfly that applies a detailed simulation of a fruit fly’s nervous system to Bitcoin trading. The project began with an initial allocation of $100 for trades involving BTC-USDC pairs sourced from Coinbase. After its first day of operation, the system recorded a $1 profit.

Stonkfly relies on the MaleCNS v1.0 connectome, a comprehensive map of an adult male fruit fly’s central nervous system that includes 166,700 neurons. This blueprint was released by an international team of scientists that included researchers from Google Research. The simulation does not involve any physical insect. Instead, it processes live market data computationally.

How the Simulation Processes Market Data

Real-time Bitcoin price information from Coinbase is converted into RGB visual representations. These images are then mapped onto thousands of simulated sensory neurons that correspond to the fruit fly’s visual system. Trading decisions emerge from the resulting patterns of neural activity within the model.

When the portfolio shows a gain, the system activates 15 simulated PAM11 dopamine neurons to represent a reward signal. In the event of a loss, two PPL101 neurons associated with aversive responses are triggered. This feedback mechanism is intended to mimic basic biological reactions to positive and negative outcomes.

Early Results and Limitations

Wormuth has stated that the $1 gain after one day is most likely the product of market fluctuations and the inherent randomness of the simulated neural responses rather than evidence of consistent learning or strategy formation. The system has not yet demonstrated reliable autonomous decision-making capabilities.

Digital connectome models capture neural wiring in detail but omit important biological features. These include chemical neurotransmitters, gene expression patterns, and broader modulatory processes that occur in living organisms. As a result, the simulation provides only a partial representation of actual brain function.

Prior Applications of the Connectome Model

The MaleCNS model has previously been adapted for interactive entertainment. Wormuth earlier linked it to the classic video game Doom in a project known as DOOMFLY. Other developers have connected the same simulation to titles including Beat Saber, Super Mario 64, Minecraft, and Pong. Stonkfly represents an extension of this work into the domain of financial market data processing.

The experiment remains in its early stages. Continued operation will determine whether the approach can develop more consistent trading patterns or whether outcomes continue to align primarily with random market movements and simulated neural variability.