SignalForge
Autonomous multi-agent trading engine for Hyperliquid perpetuals.
SignalForge fuses TradingView signals, social sentiment, on-chain whale activity, and market microstructure into a Temporal-orchestrated trading pipeline with institutional-grade risk controls and walk-forward backtesting. Designed for quant teams and crypto desks who want autonomous execution without giving up the safety rails.
8-stage agent pipeline
Signal ingestion, sentiment, smart-money mirroring, microstructure, arbitrator, risk, execution, post-trade ML — each stage independently testable.
Temporal durable execution
Crash-safe replay semantics; every workflow is recoverable, never loses a trade in flight.
Sovereign risk engine
Portfolio margin guard, L2 microstructure veto, asymmetric fault tolerance — fail-soft on sentiment, hard-fail on risk.
Walk-forward backtesting
Survivorship-bias-resistant historical evaluation; tune strategies on out-of-sample windows only.
On-chain enrichment
Track Hyperliquid whale wallets, mirror smart-money allocations, factor flows into signal weighting.
Schmitt-trigger arbitration
Hysteresis between competing models prevents signal churn and over-trading.
Tech stack
- Python 3.11
- Temporal
- FastAPI
- React 19
- PostgreSQL
- Prometheus
- Grafana
- Docker
Who it's for
Built for systematic traders.
Signal tooling for teams that make their own calls.
Quant & trading teams
Multi-agent analysis over market data, surfaced as signals you evaluate.
Fintech builders
A signal layer to build on — observable, not a black box.
Systematic traders
Tooling that supports your strategy; you stay in control of every decision.
FAQ
Frequently asked questions
Will this make me money — what kind of returns can I expect?
It's signal and execution tooling, not investment advice, so there are no return or profit guarantees. It improves how you execute and control risk; whether a strategy is profitable stays with you and your decisions.
Is it battle-tested?
It's early — alpha-mainnet — and we're explicit about that maturity. Two advanced agents are deliberately feature-flagged off pending validation, rather than shipped before they're ready.
Is it crypto-only?
Today it targets Hyperliquid perpetuals, but the underlying agent and risk architecture generalises. The durable execution and risk-first design aren't tied to one venue.
Have a project in mind?
Tell us what you want to build. We respond within one business day.