Python strategy + definition
A strategy exposes its class, market requirements and validated parameters without requiring frontend changes for every new module.
STRATEGY FRAMEWORK / EVOLUTION
STRATEGIES ARE
SOFTWARE.
Metanyx treats strategy evolution as an evidence pipeline: record the market, replay the same conditions, let AI propose a draft, test it, review the diff and only then promote it toward deployment.
MODULAR STRATEGY RUNTIME
Strategies plug into a platform, not a pile of scripts.
The Monaco trading bot already supports a strategy library, typed Studio parameters, isolated instances, lifecycle controls, persistent datasets and hot discovery of trusted modules.
Run it in isolation
Each launched strategy receives its own identity, configuration, dataset path and runtime state.
Start, pause, resume, stop
Operator controls make state explicit, including open-order cleanup and conservative behaviour across restarts.
AI-ASSISTED EVOLUTION
Let AI propose. Let evidence compete.
The integrated assistant can read strategy code and measured results, run comparable backtests and create new settings or code drafts. It cannot silently change a running strategy.
1OBSERVESignals, fills, P&L, latency
→2HYPOTHESISEAI reads code + results
→3DRAFTNew settings or module
→4REPLAYSame periods, same sources
→5REVIEWHuman compares evidence
→6DEPLOYExplicit gated launch
Important design boundary
Metanyx's current implementation is AI-assisted, not self-authorising. The assistant can create drafts and run tests; installation and launch remain explicit human actions.
PINNED CODE + LOCKED SETTINGS
An evolving system still needs immutable history.
When a strategy launches, its settings lock and its strategy code is content-addressed and snapshotted. Later code updates do not rewrite the historical run.
Content-addressed source
Strategy modules and imported dependencies are copied into a code snapshot with SHA-256 hashes.
Bot commit + runtime
Each run records platform context so measured results can be tied to a specific implementation.
Export and recreate
Run exports can bundle settings, code snapshot, manifest and performance history for review or recreation.
EVOLUTION WITHOUT CHAOS
New logic earns promotion.
AI-generated strategies should advance through evidence gates rather than directly into capital.
Syntax and parameter checks
Draft modules can be loaded in isolation and validated before entering the installed strategy library.
Compare across data sources
Monaco, Binance and Hyperliquid histories can challenge changes across periods and market paths.
Escalate deliberately
Research can move from replay to paper or shadow operation and then to gated live execution.
METANYX EVOLUTION PRINCIPLEAI proposes → evidence competes → human approves → platform constrainsAdaptive software with a preserved audit trail, not uncontrolled self-modification.
BUILD STRATEGIES
THAT CAN IMPROVE.
Talk with Metanyx about AI-assisted quantitative research, strategy infrastructure and controlled deployment.
GET IN TOUCH ↗