QL Alpha Clusters

Every stock with live analyst coverage becomes a point on a Street-vs-Engine map: Wall Street's rating flow and price targets (via Polygon.io × Benzinga) fused with the QL signal engine and the technical tape, then clustered into six named regimes. Right of the line, the Street is bullish; above it, the engine is. The corners are where the money is made — and lost.

Loading…
drag to pan · scroll to zoom · double-click to reset
◉ bubble = one stock (click for detail)✦ = cluster centroidshaded hull = cluster territorysize = target upside or market cap 1.0×
Crowd × Engine — sentiment parity (the third judge: news-crowd sentiment vs the QL engine, same 0–100 scale)
diagonal = crowd and engine in agreementabove the band = engine ahead (quiet conviction)below = crowd hotter (hype watch)size = news volume
Alpha Score distribution — where the cohort stacks up
Sector × Regime — where conviction concentrates

The six regimes (fingerprint = this cluster's signature across all six factors)

Highest-conviction names (top 60 by QL Alpha Score — click a row for detail)

#TickerNameClusterQL AlphaStreetFirmsTarget upsideEngine20d momRSIPriceDay

The avoid list (bottom of the cohort by QL Alpha Score)

TickerNameClusterQL AlphaStreetTarget upsideEngine20d mom

Method

Street layer: every analyst action in the trailing window from the Polygon.io × Benzinga ratings feed — upgrades, downgrades, initiations and price-target changes across the whole market. The latest action per firm sets a name's consensus level (so one loud firm counts once); all actions set the revision tilt; the median price target sets the implied upside. Engine layer: the QL signal overlay — the same per-ticker composite score, RSI and 20-day momentum that power the screener and bubble map. Clustering: each name becomes a 6-feature vector (Street level, target upside, revision tilt, engine score, momentum, RSI position), z-normalized over the cohort and grouped by a deterministic, archetype-seeded k-means. Clusters are named by the archetype their final centroid lands nearest. Crowd layer: the same news-sentiment blend that powers the signals overlay, mapped to the engine's own 0–100 composite scale so the two are directly comparable. It is deliberately NOT a clustering feature or a score input — Street, Engine and Crowd are three independent judges, and the parity chart exists to show where they split. QL Alpha Score: an explainable 0–100 blend — engine score 24 pts, consensus level 11, revision tilt 7, target headroom up to 12, momentum up to 6, with an overbought haircut. This is a research view, not advice; analyst data via Polygon.io.