A fast VIX spike marks forced, marketwide selling — shop it, don't fear it. The rate of change is the edge, not the absolute level: build a buy list of uptrend-intact names. Absolute fear (VIX 25-35+) is a scaling context, not a trigger; a vol spike doesn't fix a broken chart.

Objective signal

The alert fires for everyone: spike -> fire-sale window open; sustained VIX<15 -> complacency.

Action depends on intent

Buy-side (spike -> shop) = act. Sell-side (low-vol -> de-risk) = a trend-hold book NOTES it and does not sell; a harvest book trims. Same signal, action varies by declared play.

Don't add to a falling name; the trend is the reason, the story doesn't matter. Wait for a base: price stops making lows, range AND volume contract, nobody's talking about it. Then buy the base or its breakout. A base is a tag, not a buy — most bases still fail.

A name above its 200-day and beating the index over one and three years, but 20–55% off its high with no base, is a real setup that pays like a venture portfolio: the median 12-month entry loses to SPY (about −9 points), the hit rate falls the longer it is held (47% at one month → 43% at twelve), and the mean is carried by the top 5% of outcomes. Own it as a basket of small positions or not at all — never at single-name conviction size. The same screen within 10% of the high is the trend-hold instrument (positive median, 50% hit rate, half the downside tail). Structure and entry run on different clocks; the pullback does not make the proven name cheap, it makes it a different instrument.

Objective signal

The durability screen passes AND the name is 20–55% below its trailing 52-week high with no defined base.

Action depends on intent

Trend-hold book: not a core entry — basket of small positions across the jagged list, or wait for the clean cohort (within 10% of the high). Harvest/momentum book: a lottery ticket sized as one; expect the median draw to lose.

The scope rule that decides which entry tool applies where. An index correction that goes parabolic to the downside is forced selling (vol-target/CTA deleveraging, margin calls, dealer gamma) — when the forced flow exhausts there is no conviction-seller overhead supply, so price V's back in days, and waiting for an index 'base' misses the whole recovery. Buy the capitulation signature (the VIX spike trigger), in broad ETFs, in tranches, and hold through the gaps. A broken single name is the opposite: it must rebuild ownership, which takes weeks — there, wait-for-the-base applies. Credit is the discriminator: orderly spreads = positioning flush (buy it); spreads blowing out with earnings rolling = regime break (the first spike is early, not a bottom — stand aside).

Objective signal

The capitulation signature fires for everyone at the index level (>=30% 1d VIX spike; breadth washout as context); basing/breakout states fire per name.

Action depends on intent

A 1-2yr trend-hold book buys broad index ETFs in tranches on the index trigger and holds through overnight gaps; single-name entries still wait for the base. Short-horizon V-scalps are a different play this desk does not run.

Brittleness = a bigger move per unit of news and thinner shock absorbers (bonds no longer hedging stocks, leverage at records, one-factor concentration, a shortening policy fuse). It says nothing about direction or timing. A brittle tape stays 'violent inside the crowded lane, boring at the index' until the selling escapes the lane — and the one channel that converts brittle into breaking is credit (HYG vs its 200-day), not yields, auctions, or headlines. Until credit moves: hold trends on their 20-day, don't add into group-wide −8% days, don't sell on day one.

Objective signal

credit leg (HYG < 200-day) and equity leg (SPY < 200-day) fire for everyone on a settled close.

Action depends on intent

trend-hold: hold on the 20-day, no day-one action either way; harvest/active books: cut size into the lane's −8% days, re-enter on the lane's first higher low; index-V box takes over only when both legs fire.

When authorities react to a LEVEL (a 5.3% 30Y, a 164 yen) rather than to market dysfunction (a failed auction, a frozen repo), they are telling you the system can't absorb the next step — a brittleness tell. The first reaction to a blink is relief (long end bid, dollar offered, hard assets ripping, index flat); the trade is the relief's PULLBACK, not its first candle. Log each intervention with its trigger level; a rising count at falling trigger sizes is the series to watch. A Treasury buyback is NOT QE: Treasury swaps long bonds for bills (a duration punt, rollover risk moves forward); the Fed creating reserves is the escalation.

Objective signal

an intervention is announced (press release / joint statement) — fires as a world-event capture.

Action depends on intent

trend-hold: note it, buy the hard-asset/duration relief only on the pullback to the 20-day; active: fade the first-hour spike in the relief names; all books: Nov 4 refunding on the calendar.

An index that IS one trade (two memory names = the KOSPI) plus leverage stacked on that factor plus halts/limits that freeze it will gap ±4–12% day after day and export its selling to whatever still trades (ADRs, country ETFs, US cousins). Read its moves as the velocity template for the same trade elsewhere — the US AI-hardware lane trades Korea-style (three days of forced selling, one block sale, record up-day after) — and NOT as a forecast for a broad index with breadth, liquidity, and credit still confirming. The tell that a forced unwind is ending is the lane ceasing to make new lows on bad news — Korea's record up-day came the session after the seller finished.

When a name's daily tape breaks — a confirmed transition out of an uptrend, pullback, stretch, base or breakout and into a downtrend or a collapse — the verdict is sell-side and it speaks to the next few weeks: avoid a new entry, and a name already held goes to exit review rather than an automatic sale. The same discipline covers a held position whose own exit rule fires: a stop broken, a stop closing in, a winner's trail given back, an overbought trim on a tiring tape. Those are structure-first rules read off the tape — fixed percentage bands from the entry price were retired for contradicting that doctrine. What the verdict decides is narrow, and the narrowness is the point. It is NOT a call on the multi-year structure: in September 2026 $SHOP carried a broken daily tape beside a golden cross and three years of outperformance versus the index. Both were true. One answers a weeks question, the other a years question, and a reader shown one without the other reads the wrong horizon. Only a break FROM a healthy or neutral state counts — a name already in a downtrend grinding lower is not a fresh verdict, it is the old one still standing. How each book acts on it differs: a long-horizon trend book lets the 20-day and 200-day decide the hold, never a headline; short-horizon books exit at the stop with no thesis-holding; a base buyer stands aside because a broken tape is not a base; a scheduled-buy program pauses the next buy instead of improvising a sale.

Objective signal

Fires for everyone: a confirmed daily-tape transition out of a healthy or neutral state into a downtrend or collapse — or, on a held name, one of the structure-first exit rules firing.

Action depends on intent

Trend-hold / core-hold: no new entry; review the hold against its 20-day and 200-day, not the news. Active / parabolic books: stand aside, and exit the runner once its tape breaks. Base-buy: a broken tape is not a base — stand aside. Scheduled-buy program: pause the next buy and review the program; do not improvise a sale.

A proven leader in an uptrend dips below a level everyone watches (the 20- or 50-day, a prior high-volume close, a support shelf), runs the stops parked just under it, then closes back above it within one or two sessions on above-average volume. Buy the reclaim close, never the breakdown. The stop goes just under the undercut low; take it only with at least 3:1 to the prior high or next resistance; once it works, trail the 10-day or the last swing low. A 50-day or a prior high-volume close counts for more than the 20-day.