Research note · provenance-first

The Put-Skew Regime Archetype: how a model reads the price of fear, and why it mostly stands aside

It reads the options market's cost of downside protection rather than the price path, and takes a directional stance only when a second options read corroborates the first; the rest of the time it stays flat.

Published Jun 16, 2026

Method archetype

A single asymmetric curve, heavy and warm on the left and light and cool on the right, suggesting the lopsided cost of downside protection. — The model reads how richly fear is priced in the options market, not where the underlying traded. Read it as a positioning signal, not a price chart.

Protection is not free. In any market with options, someone is always quoting a price to insure against a fall and a price to bet on a rise, and those two prices are rarely the same. When insuring against a fall costs far more, that gap is the crowd putting a price on its own worry. This archetype reads that priced worry, not the path of the underlying.

What separates a put-skew-regime model from a simple fear gauge is what it does with that reading. It does not buy or sell the moment protection looks dear. It treats the price of fear as one of two options-market reads, standardizes each against its own history, and takes a position only when the two agree. Most of the time they do not, and the model stays out. The sections below describe that machinery, read for read and gate to gate, rather than telling a story about why the market moved.

What the model reads

The inputs come from the options market, not the price chart. The first is the cost of downside protection: how richly the market is pricing insurance against a fall, relative to the cost of betting on a rise. The second is a confirming options read that either points the same way as the first or does not. A put-skew-regime model is built on these positioning reads, which is structure the spot price alone cannot show.

Two panels: the cost of downside protection sloping from dear to cheap, beside a confirming options read sloping the other way; each measured against its own range. — Two options reads, each measured against its own normal. Neither a raw level nor the price path is the signal.

The first read is the price of downside protection: dear protection means the crowd is paying up to be insured, and the lean is defensive.

The second read is a separate options signal whose only job is to agree or disagree with the first. Agreement is what the model waits for.

How it decides: the regime gate

Here is the mechanism the older framing skips. Neither read opens a position on its own. The two reads must point the same way at the same time, and the combined reading must be stretched far enough versus its own history to clear a conviction bar. When both conditions hold, the gate opens and a directional stance is taken. When they do not, the model returns to flat and waits. This is decision plumbing, a read feeding a gate feeding a stance, not a forecast of where the underlying is headed.

A flow: two options reads feed a regime gate that opens only when both agree and clear a conviction bar, leading to either a directional stance or a flat stand-aside. — Read this as decision plumbing: a stance opens only when both reads agree and clear the bar. The gate sits closed most of the time, by design.

Measured against its own baseline

The model does not act on a raw level of fear. It standardizes each read against its own recent range, so what counts is how stretched a reading is versus its own normal rather than how high it is in absolute terms. A market whose protection is always dear looks ordinary when measured against itself and produces no read. A reading that pulls sharply away from its own band is the one that registers.

Two panels: a reading pulling sharply outside its own shaded baseline band, beside a permanently high reading sitting flat inside its own band. — A permanently dear level carries no signal on its own; what registers is a reading pulling away from its own band.

Why the read is relative, not absolute

Two markets can both look fearful in absolute terms while only one is sending a signal. Judging each read against its own history is what stops a structurally nervous market from looking like a permanent trade. It also means a level that has been high for a long time is treated as quiet, because against its own baseline it is no longer unusual.

A dear reading alone is not a stance

An elevated price of protection is where most fear reads stop and act. This one does not. A dear reading on its own may be overdone, the crowd paying for a fall that will not come, so the model holds. Only when the second read corroborates the first does it treat the stress as earned and lean with it. That corroboration step is the difference between a disciplined regime read and a reflexive bet against fear.

Two panels under a rising protection cost: on the left the second read does not corroborate and the model stands aside, on the right it corroborates and the model leans with it. — The discipline in one image: an elevated price of protection is acted on only when the second read corroborates it, never on its own.

The reading	What it means	The stance
Protection dear, not corroborated	The fear may be overpriced; the second read does not back it	Stand aside
Protection dear, corroborated	The stress is treated as earned; both reads point the same way	Lean with it
The reads diverge or go quiet	No regime to act on	Flat

Three states, and bounded risk

The output is one of three states. When the gate opens, the model takes a directional lean, long or short, in the direction the aligned reads imply. When the gate is closed, it holds nothing. A lean is never open-ended: every position is bounded in advance, closing at whichever comes first of a profit-target, a stop-loss, or a maximum-hold timeout. After a position opens the management is mechanical, not a matter of judgement.

Three cards labelled long, short, and flat, with the directional ones bounded by a profit-target, a stop-loss, and a timeout exit. — Three terminal states; every directional entry is bounded in advance by a profit-target, a stop-loss, or a timeout exit, so nothing is open-ended.

Selective by design

Because both reads must agree, the model participates in only a minority of sessions and stays flat the rest. That selectivity is the point, not a flaw, and it has a consequence worth stating plainly. A model that is in the market only occasionally can trail a plain always-in benchmark on total return, simply because the benchmark is exposed the whole time. The selective model carries far less exposure and, with it, a more contained worst loss.

Two strips: the archetype active in only a few blocks and flat in the rest, beside an always-in benchmark active in every block. — Long flat stretches are the design. Trailing an always-in benchmark on total return is the cost of standing aside, not a fault.

How to read it on the dashboard

A list pairing each dashboard element with how to read it: regime label, review state, open or flat, benchmark, walk-forward, and a single live reading. — A live reading is a monitoring snapshot; proof lives in the validation. Read each element for what it actually claims.

How it earns its place

A four-stage track: hypothesis, backtest after costs, walk-forward reproduction on unseen time, and acceptance gates with staged human review. — The walk-forward proves the signal reproduces, not that it earns money; a gate it fails routes the model to review rather than deployment.

A model of this archetype moves along a fixed track before it is trusted. It begins as a hypothesis that the price of protection carries a tradeable bias, is measured on history after costs, and is then replayed on time it was never fit to. That walk-forward stage checks whether the signal reproduces the same moments on unseen data, which is a test of the signal, not a claim about profit. The acceptance gates are shown openly as a scorecard, and a reading that falls short of a gate routes the model to human review rather than to deployment.

What this archetype is not

Not a simple contrarian fear-fade: it does not sell every spike in protection, and it requires a second read to corroborate before any stance.
Not a raw options-skew chart: it is a regime read built on the price of protection, standardized against its own history and gated, rather than a plot of skew.
Not a volatility forecast: it does not predict how volatile the market will be; it classifies a positioning regime and reads an implied directional bias.
Not a recommendation or a signal service: it is a method, and an exemplar model built this way can sit under review and carry no deployed capital.
Not a promise of returns: a model built this way can still be wrong, and a stress the options market did not anticipate can produce an outcome outside what was seen.

Specific models built on this archetype, with their benchmarks and review state, are listed in the model catalog. The archetype itself is a method, not any one of those models.

This explainer describes a method, not a recommendation, and a model built this way is not a guarantee of future results. It is a disciplined way of reading what the options market is paying to be afraid, acting only when a second read agrees, and standing aside the rest of the time.

Sources

Stonewell One model archetype library: the Put-Skew Regime method family, a regime read built on the options market's price of downside protection and a confirming options read, judged against their own history.
Method basis: the joint-condition regime gate, where a directional stance is taken only when both options reads agree and the combined reading clears a conviction bar; otherwise the model stands aside.
Stonewell One model validation and governance: signal-reproduction (walk-forward) testing, a published acceptance-gate scorecard, and staged human review before any deployment.