Build multi-factor ranking systems: choose factors, weights, direction and ranking scope (universe / sector / industry), get composite percentile scores and validate them with quintile performance.
Functions: rank(x), zscore(x), log(x), abs(x), min(a,b), max(a,b), pow(x,y). Fields: cap, pe, pb, ps, roe, gross, opm(net), div, eps5y, revg, beta, debtEq, curRatio, payout, shortFloat, target, perf3y, perf5y.
Train a ridge-regression model: stock characteristics predict next-month returns, validated walk-forward out-of-sample.
λ 1 · 60m
Each factor node ranks stocks into percentile scores from 100 (best) to 0 (worst) by rank position; nodes are combined by weighted average (weights normalized). Sector/industry scope ranks stocks within their own group — the sector-neutral technique used by institutional ranking systems. Stocks missing a factor are not penalized (NA-neutral).
Top 0 of 0 stocks by composite percentile score (0–100)
| # | Ticker | Name | Sector | Score | Distribution | Market cap | 1Y perf | Analyst upside |
|---|
240 months · excess vs equal-weight universe
Latest cross-section rank against the 240-month average rank — big gaps flag names drifting between snapshots.
Histogram of composite scores across the universe