← All projects
Finance & research
FlyTrade
A persistent neural trading experiment built around retained learning state and traceable execution.
Experimental trading researchThe idea behind the work
FlyTrade studies how a continuing learner can preserve its memories and decisions while the surrounding execution software changes. Neural state, deployment identity and later outcomes are treated as separate evidence.
Inside the project
- A resident runtime owns neural memories, decisions and optimiser state.
- Immutable releases distinguish the retained learner from replaceable execution and observation code.
- Status, recovery and reconciliation tooling preserve custody and delivery uncertainty.
Current scope
Experimental trading research. Retained state and passing software checks do not prove learning benefit or profitability.