Coinbase Engineer Creates Virtual Fly Brain for Bitcoin Trading

Coinbase engineer Alex Wormuth has developed an experiment called Stonkfly, in which a virtual copy of a fruit fly's brain analyzes the candlestick chart of Bitcoin and makes trading decisions based on a reward system. The brain model consists of 166,700 neurons and 25.6 million connections. The system does not use numerical quotes but works with the image of the chart. Successful trades activate 15 reward neurons, while losing trades activate 2 negative neurons. Stonkfly cannot open margin positions or short sales, which distinguishes it from many AI bots. Wormuth emphasizes that changes in synapses do not confirm that the brain has learned profitable trading. The program starts with a virtual balance of $100, with a trade limit of $10 and 24 trades per day. As of September 13, there were no completed trades on the public panel, when Bitcoin was trading around $77,150, having lost 1% over the day. Stonkfly is a research project at the intersection of neurobiology and the financial market, rather than a ready-made trading strategy.