Hypothesis
Candidate strategies begin with a stated economic mechanism, observable proxy, expected behavior, and falsification path.
Diversified alpha discovery across markets.
Structured pipelines for discovering, validating, and combining statistical strategies across markets.
01 / Approach
A controlled search process for selected alpha candidates.
The objective is not to rely on a single market, signal family, or trading style. We explore a broad design space and apply disciplined validation to identify a smaller set of robust and diversifying strategies.
Each candidate is treated as a research artifact with explicit assumptions, required data, expected behavior, failure modes, validation criteria, and portfolio implications.
02 / Research Process
Systematic exploration, disciplined validation.
Specification
Research ideas are converted into structured candidate specifications before they are rendered into testable signals.
Validation
Data availability, leakage controls, duplicate screening, deterministic construction, and backtest evidence are enforced before promotion.
Portfolio Context
Candidates are evaluated not only by standalone performance, but also by their interaction with the broader alpha portfolio.
03 / Focus Areas
Statistical Arbitrage
Cross-sectional and relative-value strategies across liquid markets.
Market Structure
Liquidity, volatility, flow, positioning, and event-driven effects.
Prediction Markets
Information-driven inefficiencies and event probability dynamics.
Equity Markets
Anomaly research, factor behavior, and structural market edges.
Multi-Horizon
Intraday to multi-day horizons with consistent evaluation rules.
Portfolio Construction
Combining validated alphas into diversified risk-aware portfolios.
04 / Risk Discipline
Diversification is a research problem, not a slogan.
We seek to reduce strategy fragility by exploring multiple markets, mechanisms, horizons, and sources of return.
Correlation, redundancy, failure regimes, turnover, capacity, and regime sensitivity are treated as first-class research constraints.