A stale entry zone silently suppresses the name it is attached to

Claim

Watchlist.md carried INTU — Add $370-390 while the stock traded at $319.97. Three consecutive scans dutifully recorded "still below its add zone" and moved on. The phrase is true and the conclusion drawn from it was exactly backwards: the stock was not failing to reach a buy zone, it was sitting well inside the buy range a current fair value implies ($300-345 on FV $380-470) while the watchlist compared it against a zone derived from a fair value nobody had recomputed since before a 60% de-rating.

The name was also held in both accounts the entire time, and had never been analyzed.

The mechanism

A zone is not an independent fact. It is fair_value × margin_of_safety, and it inherits every assumption in the fair value. When the fair value moves, a zone that stays put stops describing the thesis and starts describing the previous thesis.

The failure is quiet because the stale zone produces a well-formed, plausible sentence. "Below its add zone" is what a scan says about a name that is genuinely too expensive. There is no error state, no warning, no parse failure — the watchlist reports normally and the name is skipped. Compare a missing zone, which is obvious on sight and gets fixed.

Direction matters and inverts the damage:

Zone drifted	Scan says	Actual failure
Above spot (FV fell, zone didn't)	"still below its add zone"	Suppression — a name in range never gets looked at
Below spot (FV rose, zone didn't)	"zone blown, waiting for a pullback"	Suppression — waiting for a price that no longer needs to come

Both directions produce inaction dressed as discipline. Neither ever produces a bad trade, which is precisely why the error survives — it has no visible cost.

The third case, which does have a cost: a stale TRIM zone

The rows above cover entry zones, where the damage is inaction. A stale trim zone is worse, because it fires an action. An entry zone that never triggers costs an opportunity; a trim zone that triggers on obsolete arithmetic sells a position.

Found 2026-08-05 across three names simultaneously:

Ticker	Watchlist instruction	Spot	Failure
DT	Add $31-35 · Trim $50+	$50.86	Trim line live now, on a beat-and-raise with FY27 EPS guided above consensus. The add zone's floor ($31) is essentially the 52-week low ($31.64) — both numbers were set against a materially smaller company
AMD	Trim $360+	$482.05	Above the trim line for months with no action taken. An instruction that is ignored in practice is not a zone, it is decoration
APP	"Let it run. Trim $650+"	$417.80	The instruction is written for a position that is $518 — the zone assumes a holding size that does not exist

The asymmetry to internalize: a stale add zone fails safe, a stale trim zone fails expensive. So trim levels need more refresh discipline than entries, not less — which is the opposite of how attention naturally allocates, since entries are what a scan is looking for.

The AMD variant deserves its own name. A trim level that has been breached for months without action is not stale in the ordinary sense — it is dead. It carries no information, and worse, its presence makes the line look reviewed. Either the level is wrong and should be re-derived, or it is right and the trim should happen. A standing instruction nobody follows is a decision that was never actually made.

This connects to the "Held — No Action" section specifically: names parked there accumulate stale zones fastest, because the whole point of the section is that nobody is deciding about them. That is fine for the thesis-break trigger, which is what the section says it tracks — but a trim level is not a break trigger, and it should not be sitting there unrefreshed.

The tell

A price that has moved far enough to need a new thesis has moved far enough to need a new zone. Concretely, treat these as automatic zone-invalidation events:

The name is >30% away from the zone in either direction — the zone is being asked to survive a move it was never derived across.
A drawdown or run larger than the zone's own width has happened since it was set.
The **A{date}** analysis stamp on the line predates a fiscal year's worth of results.
The scan output for the name has been the same sentence twice running. A watchlist line that does not change is either a stable thesis or an unexamined one, and the two look identical from outside.

Practical rule

Never carry a zone whose fair value you would not re-derive on request. If a scan is about to write "still below/above its add zone" for the second consecutive time, that is the trigger to re-derive the zone — not evidence the name is inactive.

The cheap version: whenever a scan touches a name, check spot vs zone for sign, not just membership. A zone that sits entirely on one side of the price and has for multiple scans is a flag, not a reading.

What would falsify this

A stale zone that was still approximately correct after a large move — i.e. the fair value genuinely did not change despite the price collapsing. That happens (a pure multiple de-rating with intact earnings is exactly that case) but it has to be demonstrated, not assumed by inertia. INTU is the counterexample: the fair value did move, from whatever supported a $370-390 zone down to $380-470, and the zone should have moved with it.

[[principle-down-a-lot-is-not-cheap]] — the mirror error. That principle guards against treating a drawdown as sufficient reason to buy; this one guards against a stale zone preventing you from looking at all. Both are failures to re-derive value after a price move, in opposite directions.

[[pattern-shortlist-runs-when-unexecuted]] — the same family: watchlist state that decays between scans and quietly changes what the list recommends.

History

2026-08-04 — established from the INTU case, found during that name's first ever /analyze. The zone had suppressed the name across three scans while it was held in both accounts. Zone corrected $370-390 → $300-345 in the same pass.
2026-08-05 — extended to trim zones, which fail expensively rather than safely. Found during the 8/05 movers screen: DT's Trim $50+ went live at $50.86 on a beat-and-raise, AMD had been above Trim $360+ for months untouched at $482.05, and APP's Trim $650+ was written for a $518 position. All three names are held and have never been analyzed — the same co-occurrence as INTU, suggesting the real predictor of a stale zone is not elapsed time but the absence of an analysis file for a name the user owns.