Breadth is narrowing while the index isn't

The tape this week has the shape that always precedes a busy stretch for a reversion screen: the index looks fine and the median stock inside it does not.

When an index is carried by a handful of large weights, the arithmetic hides everything underneath. Cap-weighting means the top names can absorb the drag of a hundred smaller ones. The headline number stays flat, and the experience of holding the average name is nothing like flat.

Why this matters for a reversion book. Narrow breadth is the environment that fills the screen. More names fall two standard deviations below their own 50-day mean, so the candidate list gets long precisely when the index is telling you nothing is wrong. That is the setup and the trap in one sentence.

The setup: individual names are getting stretched for reasons that are about positioning and liquidity rather than about their earnings power. That is textbook — price moved, value did not.

The trap: narrow breadth is also what the beginning of a broad drawdown looks like. The distinction is not visible in the breadth number itself. It is visible in whether the anchors are moving — whether forward estimates across the screen’s candidates are drifting down together, or whether the selling is positioning-driven and estimates are stable.

What I’m watching this week, in order:

1. Whether the candidate list clusters in one or two sectors. Six oversold names in the same industry is one trade with six tickers on it, and it should be sized as one trade.
2. Whether the 50-day means themselves are still flat. A stretched z-score against a falling mean is not an opportunity, it is a slower way to lose.
3. Realized versus implied vol. If implied is cheap while realized is rising, defined-risk structures get relatively more attractive than shares.

What would change my read: breadth broadening back out with the index flat. That resolves the divergence in the boring direction and empties the screen, which is a fine outcome — no trade is a position.

Nothing here is a recommendation. This is my read on the tape, published so that when it turns out to be wrong there is a record of what I thought at the time.