AI-Native Systematic Investment
Eigen Partner
The model proposes — the math disposes.
The Insight
The densest alpha was never for sale.
39%
Medallion — closed
≈8%
the open fund
01
The Renaissance paradox
Same team, same building. One book, closed to outsiders, compounded far above the market for thirty years; the firm's open fund tracks near the index. The gap is not a hidden secret — it is what a single research capability must produce under two different capacity limits. The densest alpha is structurally the part you cannot buy.
02
The empty fifth quadrant
Between the mega-funds that can fund a hundred-million-dollar research arms race and the boutiques that cannot, sits a position no one occupies: density-grade alpha that is still accessible to outside capital. It stayed empty because no incumbent's cost structure could reach it.
03
Why this is possible now
AI-native research resets that cost structure. A small senior team can reach the research throughput that once required dozens — generating, coding and testing hypotheses at machine scale, under human judgment. For a few years, the fifth quadrant is open.
Investment Approach
AI-native, end to end.
Eigen is an AI-native quant fund. The research throughput that once demanded a hundred-person desk now runs in a handful of people — so we reach density-grade alpha without the cost structure that forces incumbents to dilute it.
AI-native infrastructure
A proprietary stack does the work of a hundred researchers — the cost wall that caps everyone else, removed.
Capacity-aware
Capacity is a governance rule, not a marketing number. When density and scale conflict, we cap scale.
Methodology-portable
The alpha comes from how we research, not from where — each new market only re-defines the universe and the regimes.
Research Methodology
The model proposes. A deterministic core decides.
A proprietary architecture at the intersection of high-dimensional statistical learning and autonomous execution — engineered to extract persistent, uncorrelated alpha from global markets. An LLM layer breeds new factors from proven survivors, cross-pollinating temporal, cross-sectional and alternative data into structures invisible to human intuition; a deterministic core decides what ships.
The falsification gate
Most ideas die here. The few that survive ship.
Every arc and every backtest is logged to an experiment and hypothesis registry — so an industrialized search can neither silently overfit nor quietly forget. Breadth without the gate is just overfitting; the gate is what makes breadth honest.
LLM Orchestrationfactor mining · regime memory · self-improving loops
The Fund
Atelier No.1
China A-shares · Global commodities · Index futures
The first product spans China A-shares, global commodities and index futures — a multi-market systematic strategy, run small enough to stay dense. Capacity is defined and deliberately held; specifics are shared with qualified investors under diligence.
One mandate. Run by a senior team. Closed before it dilutes.
Track Record
Verified at diligence.
Performance is third-party audited and disclosed to qualified investors under NDA. No figures are published here.
Team
The Infrastructure Triangle, personified.
CEO · Eric
Research direction and the firm.
~70%
Annualized
<30%
Max drawdown
CIO · Frank
Investment strategy and the trading engine.
10+ yrs
Quant PM experience
2
Top quant firms
COO · Harry
Capital formation and operations.
7+ yrs
Institutional VC
CN·US·HK
Cross-border
The triangle's three vertices — infra innovation, quant first-principles and LLM orchestration — close across three operators. Combinatorial scarcity, not résumé stacking.
Contact
For qualified investors.
Eigen Partner works with qualified institutional investors. For a conversation or diligence materials, reach us directly.
[email protected]