Our consulting methodology is structured around three sequential, interdependent stages of systematic alpha production. Each stage is designed to function independently as a diagnostic or advisory engagement, or as part of an end-to-end build or rebuild of a mid-frequency equity strategy. The framework below reflects how we approach the problem of translating raw information into risk-adjusted, executable alpha.
STAGE 01
Alpha Signal Generation
The foundation of any systematic equity strategy is the quality and breadth of its raw information set. We construct signal libraries spanning thousands of weak, individually low-predictive-power features derived from a structured universe of data types:
- Cross-security relationships — co-movement, pairs, sector and factor relatives
- Fundamental data — earnings revision velocity, balance sheet dynamics, accruals
- Analyst intelligence — estimate dispersion, revision momentum, coverage drift
- Options market structure — implied volatility skew, put/call positioning, term structure
- Sentiment and alternative data — news flow, earnings call linguistics, short interest dynamic
Each signal is constructed against a clean, point-in-time security master to eliminate look-ahead bias and survivorship distortion. Feature engineering is applied systematically — normalization, decay weighting, neutralization — to ensure each signal contributes information that is statistically distinct and temporally stable. The objective at this stage is volume and diversity: a wide signal library forms the raw material for the fusion layer.
STAGE 02
Signal Fusion & Optimization
Individual weak signals carry insufficient predictive power to drive portfolio decisions directly. The objective at this stage is to combine them systematically into a composite output — a super signal — with materially higher Sharpe and information coefficients than any of its constituent inputs.
We apply a structured optimization layer using a combination of techniques:
- Hierarchical clustering — grouping signals by information content to reduce redundancy and implicit weighting toward correlated features
- Convex optimization — constrained combination weights that maximize predictive accuracy subject to stability and turnover penalties
- Mean-variance signal weighting — portfolio-theory framing applied at the signal level to manage signal-space variance and covariance
The fusion process is evaluated against out-of-sample information ratio, rank correlation stability, and decay half-life. Signal weights are reviewed periodically for drift and regime sensitivity. The output is a single, high-frequency composite score — ranked cross-sectionally across the tradeable universe — that drives portfolio construction.
STAGE 03
Trade Construction & Risk Management
A high-quality signal is a necessary but insufficient condition for institutional alpha. The final stage translates the composite signal into a legally, operationally, and market-impact-compliant trade program — one that survives contact with real market structure.
Constraint modeling at this stage includes:
- Transaction cost estimation — spread, market impact, and slippage modeled at the security level, calibrated to ADV and order size
- Position and concentration limits — hard and soft constraints by security, sector, and factor bucket
- Risk exposure management — factor neutrality targets (market, sector, style), tracking error budgets, and drawdown guardrails
- Crowding analysis — short interest, factor positioning, and institutional flow monitoring to flag adverse liquidity environments
The output is a daily trade list — long and short — sized for full-day VWAP execution and designed to absorb capacity without material signal decay. Portfolio turnover, holding period, and gross/net exposure are calibrated to the fund’s operational parameters and prime brokerage constraints.
Engagements may be structured as independent diagnostic reviews of any single stage, or as full end-to-end strategy builds. We do not offer pre-packaged products. All work is conducted under mutual NDA and is specific to the client’s data environment, capacity constraints, and investment mandate.
© 2026 Pragya Labs LLC. All Rights Reserved. Qualified Institutional Inquiries Only.