🌱 EtherSym Finance is the result of a long journey to translate biological principles into computational behavior.
Over the years, I have been exploring symbiotic interfaces — systems capable of learning, self-regulating, and evolving, inspired by living processes such as homeostasis, regeneration, and energetic balance.
This project was born from the desire to bring biology closer to artificial intelligence — to build networks that do not merely execute tasks, but live within their own parameters, adapting according to energy, error, and experience.
Technically, EtherSym Finance employs Deep Reinforcement Learning (DRL) with an extended Dueling DQN architecture, combining symbolic and biological mechanisms into a self-stabilizing learning system.
- Dueling DQN Network + Continuous Regression Head
- Discrete actions: buy / hold / sell
- Continuous output: predicted future return
- Symbiotic Mechanisms
- Adaptive homeostasis
- Synaptic pruning and regeneration
- Prioritized replay buffer with energy-based control
- Training
- Historical Binance candle data (1-hour interval)
- GPU support with PyTorch and
torch.compile - Persistent states and Sharpe-like performance metrics
python -m v1.main