Crowded

Where is positioning extreme?

ProductionVerdict · PARTIALRecall · 0.30

Supporting · the glance (one-sided book) · anchor metric: net-spec positioning z

What it measures

The positioning counterpart to Stretched. Scans CFTC net-spec, ETF flows and survey sentiment (AAII, NAAIM) for one-sided books. A crowded trade is the fuel a surprise needs to unwind fast — it reads the lean, not the level.

z_i = \frac{\text{net\_pos}_{i,t} - \operatorname{mean}(\cdot \text{ over } W)}{\operatorname{std}(\cdot \text{ over } W)} \quad\text{\# positioning only}

\text{input\_types} \in \{\text{positioning, sentiment}\} \quad\text{\# inverse of Stretched}

\text{admit if } |z_i| \ge 1.0 \,;\ \text{rank by } |z_i| + \text{salience}

A net-long z of +2 means speculators are loaded long at a multi-year extreme. High |z| flags fragility, not direction — crowded longs can stay crowded, but they unwind violently when they turn.

Limitations

Small candidate pool — positioning inputs are sparse, so the lens sits at baseline (didn't lift in the refactor).
0.30 recall — descriptive of the lean; says nothing about timing.

In the product

Drives the glance's "Where's the crowd leaning?" block (CFTC FX positioning). In commentary it pairs with Stretched to build the one-sided calm thesis — cheap protection over a loaded book.

Grading & stability

—

Gate rate
loading…

—

Stability / churn
not emitted

—

Mean confidence
loading…

fetching /api/panel/about/lens_engine.crowded/W/all …

Graded on the Tier-1 regime-test harness — curated macro events × 18-yr backfill × recall@K. Verdict PARTIAL (19%).

Charts it reads

CFTC FX positioning

cftc_fx_positioning—

CFTC equity positioning

cftc_equity_positioning—

CFTC commodity positioning

cftc_commodity_positioning—

Put/call ratio

put_call_ratio—

Crowded · The Lens methodology← all engines