01
ML models built on 70k+ samples. 17 hypotheses tested, validated, or killed with statistical scrutiny.
Applying for
AI-native capital allocation
Institutional research for the autonomous era
The Next Great Fund Will Be Built on AI.
Opal Research is developing autonomous trading systems that research, debate, and compete for capital — then learn from their own failures. We're not adding AI to a hedge fund. We're building a hedge fund out of AI.
02
Live trading on prediction and equity markets. Real capital, real results, publicly verifiable.
03
Production-grade infrastructure. Built by an ex-AWS distributed systems engineer.
Our Edge
AI-assisted is a faster horse. AI-native is the engine.
“I don't want to have to worry about the market every minute. I want models that will make money while I sleep.”
Jim Simons
Today's hedge funds use AI the way enterprises used the cloud in 2010 — as a faster tool for existing processes. Research summaries. Trade screening. But every decision still flows through a human bottleneck.
The firms that defined quantitative trading didn't add computers to human workflows. They rebuilt the workflow around computation. We're doing the same with AI: autonomous agents as the primitive, not the accessory. Every improvement in frontier models compounds directly into fund performance — no human bottleneck in the way.
Why now
Model gains compound
New frontier capabilities immediately improve research, debate, and risk assessment when the fund is built around agents.
Human bottlenecks fade
The operating system for the fund becomes computational rather than managerial.
Architecture first
Opal is building the architecture ahead of the model curve so future capability gains can be absorbed overnight.
The System
A Darwinian colony of autonomous fund managers.
Autonomous AI agents — each with a distinct trading philosophy, structured beliefs, and layered memory — share a common research layer but form independent theses and compete for capital allocation. New strategies prove themselves with pocket money — small live bets — before earning real allocation. Every thesis is stress-tested through adversarial debate before reaching a structurally privileged risk layer with veto power. Strategies that perform earn more capital. Strategies that fail get a post-mortem: flawed logic dies permanently and its lessons are absorbed. Sound strategies killed by hostile conditions enter dormancy — full state preserved, zero allocation. When dormant strategies show life again, that is the regime change signal. The colony doesn't just optimize. It evolves.
Capital allocation
Performance earns more capital. Failure gets quarantined, studied, and removed.
Memory and feedback
Each cycle improves the next generation of traders through structured post-mortem analysis.
Operating Principle
1
Autonomous research and thesis formation.
2
Adversarial debate before capital allocation.
3
Risk veto at a structurally privileged layer.
4
Capital flows to what works. Failed strategies are removed.
5
Post-mortems improve the next generation of agents.
Evidence
Evidence that the system is being built on real work.
Three proof points to show this thesis is backed by shipped systems, live risk, and institutional engineering discipline.
Scientific rigorProof point
Rigorous methodology
ML models built on 70k+ samples. 17 hypotheses tested, validated, or killed with statistical scrutiny. We don’t ship what doesn’t survive the data.
Live marketsProof point
Skin in the game
Live trading on prediction and equity markets. Real capital, real results, publicly verifiable.
InfrastructureProof point
Institutional-grade systems
Nautilus Trader execution engine. Walk-forward ML pipelines. AWS infrastructure. Built by an ex-AWS distributed systems engineer.
Where We Are
Phase 0 — Laying the foundation.
Human-driven research and live trading. Production-grade infrastructure. A structured knowledge base capturing every decision rationale and market observation. Preparing to distill professional trader cognition into AI agent seeds. Building the data pipeline that bootstraps the colony's initial intelligence.
Human-driven research and live trading
Production-grade infrastructure already in place
Structured knowledge base for agent seeding
Preparing professional trader cognition for initialization
Founder
Built by Shawn Wang — ex-AWS distributed systems engineer, serial founder, and full-time operator in algorithmic trading research since late 2025.
Opal Research is the convergence of production infrastructure, quantified experimentation, and a long-horizon conviction that autonomous agents will become the next computational primitive for capital allocation.