Field Equities — Multi-Strategy Quantitative System
Six scoring strategies — each with a different theory about what moves stock prices — plus one equal-weight baseline (Pandia), one random control (Tyche), and the S&P 500 as the benchmark. All eight methods rank the same 76-stock watchlist every session, and their picks are recorded on paper, no real money. As time passes, the actual forward returns fill in: did the picks perform in 1 day? In 3 months? In a year? The strategy that consistently picks the best-performing stocks earns its weight.
What $100k equally spread across each scorer’s top picks would have returned, using recorded forward returns. Shadow scorers only — not real trades or live P&L.
1d forward returns fill daily after market close. Equity curves will appear as data accumulates.
How each strategy scores a stock
Inputs and weights for each scoring method. Click a strategy name to read the full rationale in Methodology.
Four input layers combined into a 0–100 score per stock. All weights are placeholder — pending calibration on clean forward data.
| Input | Max pts |
|---|---|
| Sourcing (40% — Smart-money signals) | |
| Congress trades | 12.0 |
| Insider buys | 10.0 |
| 13F institutional flow | 10.0 |
| ARK fund trades | 8.0 |
| Factor (30% — Academic price factors) | |
| Price momentum (12m–1m) | 10.5 |
| Post-earnings drift | 9.0 |
| 52-week high proximity | 6.0 |
| Quality (ROIC/ROE/margins) | 4.5 |
| Fundamentals (20% — Sector-relative valuation/health/growth) | |
| Valuation (P/E, P/B, EV/EBITDA) | 6.8 |
| Financial health | 6.6 |
| Revenue + EPS growth | 6.6 |
| Qualitative (10% — AI theme overlay) | |
| AI theme membership | 10.0 |
Arachne and Asteriause the same inputs, then nudge the final ranking by up to ±5 positions based on each stock’s eigenvector centrality in the trailing correlation network. Metis dominates; topology is a bounded overlay.
Single input — trailing 12-month return
Each stock is ranked by its own trailing 252-day total return, highest to lowest. No multi-input weighting. Adjudicated against 21d–5y horizons where this signal is most relevant.
Single input — CFTC commercial positioning percentile
Ranks instruments by where commercial (hedger) net positions sit in their own 3-year history. Covers 4 ETFs only: SPY, QQQ, IWM, GLD. Runs weekly (Fridays). Insufficient-history instruments are skipped.
No input — every stock equal
Every name in the watchlist gets identical rank. The primary race control: any strategy that cannot beat Pandia has no per-name ranking edge.
No input — random order (seeded)
Universe shuffled by a deterministic seed each scan. The secondary control and the statistical noise floor.