A digital simulation modeled on a fruit fly’s neural network was used for bitcoin trading and made $1 in profit after one day of operation. However, a single profit does not prove real investment learning ability or automated trading performance.

On Sept. 11 local time, blockchain outlet U.Today reported that Coinbase software engineer Alex Warmus funded a system called Stonkfly, built on the wiring structure of a fruit fly brain, with $100 and began a bitcoin trading experiment. Warmus released initial results one day after the experiment began, when profit stood at $1.

Stonkfly does not involve a live fruit fly directly trading bitcoin. It simulates in a digital environment MaleCNS v1.0, a connectivity map of an adult male fruit fly’s central nervous system, and links it to cryptocurrency price data.

The trading process is also unusual. Stonkfly takes Coinbase’s public BTC-USDC market data and converts it into RGB images. It then feeds the images into thousands of sensory inputs corresponding to neurons that make up an insect visual system to generate buy and sell signals.

Signals based on profit and loss also mimic the fruit fly nervous system. When the portfolio makes a profit, it stimulates 15 PAM11 dopamine neurons, and when a loss occurs, it activates 2 PPL101 neurons associated with avoidance responses.

But it is hard to interpret a $1 profit after one day as a learning result. The outlet also stressed that "profitable learning has not been proven." The possibility cannot be ruled out that profit resulted from market moves aligning with a chance trading decision. It has not yet been confirmed whether it produced the same result in repeated market environments or actually learned price patterns.

The fruit fly neural connectivity map that underpins the experiment is linked to research released earlier this month. Scientists at Google Research and other institutions used artificial intelligence to build a complete connectivity map of an adult male fruit fly’s brain and central nervous system. The map contains about 166,700 neurons and the structure of connections among them.

Fruit fly neural network simulations are also being used in games. Warmus earlier presented Doomfly, which linked the same MaleCNS model to the classic game Doom. There have also been continued attempts to connect fruit fly connectome simulations with Beat Saber, Super Mario 64, Minecraft and Pong.

Researchers also draw a line, saying the current simulation does not reproduce the entire brain of a real fruit fly exactly as it is. That is because complex biological processes such as neurotransmitters, chemical modulation and gene expression are not fully reflected.

Ultimately, Stonkfly at this stage is closer to an experiment combining neuroscience data and financial market data than an example proving automated bitcoin trading performance. The key question is whether, beyond short-term chance profit, a fruit fly neural network simulation can actually learn market patterns and deliver repeatable investment results.

I gave the fly brain $100 to trade bitcoin. Dopamine neurons are stimulated when the fly makes profit. Neuron activity controls buy/sell decisions and makes trades on coinbase. Will the fly get rich? https://t.co/XVWSFattGu