A daily unsupervised machine-learning read of the US Equity market: 4,777 stocks grouped into 9 consensus clusters (KMeans + Gaussian-mixture + hierarchical, over robust-scaled PCA features) for 2026-10-14. Descriptive, not predictive — there is no buy or sell signal. Research, not investment advice.
large cap · favorable 1y edge · low volatility (3m) — 1,496 names; mostly Financial Services; drivers: log market cap (+0.43), pa edge 52w (+0.42), rv 63 (−0.41)
deep pullbacks (21d) · high volatility (1y) · high downside vol — 988 names; mostly Healthcare; drivers: pa mae 21d (−0.60), cfo growth (−0.58), vol of vol 63 (+0.56)
rising earnings · high growth · high leverage — 574 names; mostly Financial Services; drivers: cfo growth (+2.12), ni growth (+1.71), revenue growth (+1.40)
strong 1y momentum · wide upside excursions (21d) · rising (3m) — 441 names; mostly Technology; drivers: ret 126d (+1.86), ret 252d (+1.70), pa mfe 21d (+1.39)
expensive (low E/P) · low cash yield · cheap (high B/P) — 344 names; mostly Healthcare; drivers: val ep z (−3.80), val cfop z (−3.27), val bp z (+1.47)
attention spent · high turnover · wide upside excursions (21d) — 265 names; mostly Technology; drivers: pa sir recovered (+2.75), turnover to mcap (+2.65), pa mfe 21d (+1.90)
high leverage · high cash yield · expensive (low E/P) — 225 names; mostly Financial Services; drivers: leverage debt to mcap (+3.54), val cfop z (+3.01), val ep z (−2.55)
cheap (high E/P) · high cash yield · cheap (high B/P) — 224 names; mostly Financial Services; drivers: val ep z (+4.07), val cfop z (+4.03), val bp z (+3.81)
high downside vol · high volatility (3m) · attention spent — 220 names; mostly Healthcare; drivers: vol of vol 63 (+3.71), downside vol 63 (+3.40), rv 63 (+3.21)
Machine-readable data (free, read-only JSON)
The full map, per-ticker cluster assignments with confidence and anomaly scores, PCA structure, and within-cluster cointegrated relative-value pairs are published as open JSON for automated and AI-analyst consumption:
/api/ml/latest — clusters, names, sector mix, representative tickers, top features
/api/ml/pairs — within-cluster cointegrated pairs (Engle-Granger, OU half-life, Kalman z-score; US only)
Descriptive market-structure research only. Unsupervised clustering finds structure, not direction; a tight cluster or an anomaly is a starting point for research, never a trade signal.