Night BuildsNo. 003 · all builds
Night Foundry · Night One

The Panic Museum

Five rooms. Four are history; the fifth is a snapshot, frozen on 9 Oct 2026. Walk them in order — each panic has a type, a tape, and a lesson.

THE PRIOR  you'd guess the depth of the fall decides the recovery. The museum says it's the type of panic that decides — and one type never forgave.

Enter Wing I ↓
I
Wing I · The Rate Shock

Black Monday · 1987

What broke

Portfolio-insurance programs — computers told to sell futures mechanically as prices fell — poured gasoline on an overvalued market already smoldering from rising rates, a weak dollar, and trade-deficit headlines. When the cascade started, specialists couldn't keep up and the tape itself went blind: orders sat unfilled for over an hour.

The tape
−22.6%
Dow Jones, Monday Oct 19, 1987 — 508 points to 1,738.74. Largest one-day percentage fall in the index's history, before or since.
−20.5%
S&P 500 the same day. The Wilshire 5000 fell 18%. Ninety-five S&P stocks couldn't even open on time.
$1.71T
Estimated worldwide losses in a single session. NYSE volume: 604M shares — triple the daily average.
0
Recessions that followed in the US. Greenspan's Fed pledged liquidity the next morning — "ready to serve as a source of liquidity" — and the panic stayed a market event, not an economic one.

Figures: Dow Jones / NYSE archives; S&P 500 index history (FRED SP500). Settled history.

What worked

Cash into the close, then nerve. The Fed's liquidity backstop drew the line under the plumbing; buyers of the panic were made whole within roughly two years. Treasury bonds — the supposed kindling — became the shelter.

"A liquidity panic with the economy intact is a detour, not a destination — it forgives the patient faster than any other kind."

Next wing → II · 2000
II
Wing II · The Thesis Break

The Dot-Com Bust · 2000–2002

What broke

The story. "This time it's different" carried valuations with no earnings underneath — Pets.com had a Super Bowl ad and no path to profit. When the Fed tightened and the funding stopped, there was no there there: the thesis itself was the bubble, so the bust had nothing solid to land on.

The tape
−78%
Nasdaq Composite: 5,048 on Mar 10, 2000 → ~1,114 on Oct 9, 2002. The Nasdaq-100 fell harder: −83%.
~$5T
In market value destroyed. Fewer than half of the dot-com companies floated at the peak were still alive by 2004.
15 yrs
To reclaim the peak — the Nasdaq didn't see 5,000 again until March 2015. Cisco, a great company, was an −80% investment.
Mild
The recession that followed was one of the mildest in modern US history. No debt, no bank run — the crash never infected the plumbing, which is why the economy survived what investors didn't.

Figures: Nasdaq Composite (FRED NASDAQCOM); NBER. Settled history.

What worked

Owning the survivors — Amazon, eBay, Google — through the wreckage, or better, waiting for the base to prove itself before touching anything. Cash. And the discipline to distinguish a great company (Cisco) from a great investment (it wasn't, for a decade).

"When the thesis breaks — 逻辑被杀 — the dip is not a discount. This is the one wing where patience alone was not enough."

Next wing → III · 2008
III
Wing III · The Liquidity Crisis

The Global Financial Crisis · 2008–2009

What broke

Credit — the archetype. Subprime mortgages, sliced into securities nobody understood and rated like they did, froze the entire funding system when Lehman fell on Sep 15, 2008. Money-market funds broke the buck; corporations couldn't roll overnight paper to make payroll. This is the one where everything sold first.

The tape
−56.8%
S&P 500: 1,565.15 (Oct 9, 2007) → 676.53 (Mar 9, 2009). The Dow lost 54%. In the final 18 trading days the S&P fell another 25%.
80+
The VIX, at its peak — fear priced as no generation had seen. The Dow's 777-point drop on Sep 29, 2008 was then its largest point fall ever.
5.5 yrs
To reclaim the 2007 high — March 2013. TARP's $700B and the Fed's balance sheet bought the bridge, slowly.
−1.3%
IMF World Economic Outlook, April 2009: world output projected to contract 1.3% in 2009 — "the deepest post–World War II recession by far." The real economy did break this time.

Figures: S&P 500 (FRED SP500); VIX (FRED VIXCLS); IMF World Economic Outlook April 2009; Fed. Settled history.

What worked

Treasuries, gold, cash — then, after the deleveraging ended, the hunt: quality assets wrongly killed in the phase-1 indiscriminate selloff. March 2009 buyers never saw those prices again; the bottom was never retested.

"In a liquidity crisis, phase one sells everything — the money is made in the hunt after the deleveraging ends, among the wrongly killed."

Next wing → IV · 2020
IV
Wing IV · The Exogenous Shock

The COVID Crash · 2020

What broke

Nothing financial — the economy was switched off on purpose. A virus, lockdowns, and then the strangest sight in market history: oil futures printing minus $40 a barrel because there was nowhere left to put the crude. The panic was about the world, not the plumbing.

The tape
−34%
S&P 500 in 33 calendar days (Feb 19 → Mar 23, 2020). The fastest bear market on record — the median of the 20 bears since the 1920s took 302 days.
Record
The VIX hit its all-time high in March 2020. Ten of the largest single-day point drops and eight of the largest single-day gains in S&P history happened in the same year.
+$5T
Roughly, in Fed + fiscal stimulus. The S&P gained ~55% in the five months after the low; the Nasdaq reclaimed all-time highs by June.
−$40
WTI crude, April 2020. Demand didn't fall — it vanished, and storage became the scarcest commodity on earth for one afternoon.

Figures: S&P 500 (FRED SP500); VIX (FRED VIXCLS); WTI spot (EIA DCOILWTICO). Settled history.

What worked

Duration first (bonds rallied as the world hid), then the digital economy — Amazon, Netflix — then, once the rescue was credible, almost everything. The fastest falls forgive fastest if the plumbing holds: it did, because the cause was outside finance.

"Exogenous panics end when the cause ends. Speed of the fall tells you nothing about speed of the recovery — the plumbing does."

Next wing → V · 2026 — the live wing
V
Wing V · A dated snapshot — 9 Oct 2026 (this room does not update)
● You are here — 9 Oct 2026

October 2026 · frozen

What this room shows

On Friday 9 October 2026 the 10-year Treasury closed at 5.24%, four days after printing 5.31% — its highest close since April 2002. The S&P 500 closed at 7,811.54, a tenth of a percent below its 6 October closing high; the Nasdaq-100 1.1% below its own 6 October high. The VIX closed at 14.8. Duration at multi-decade highs, equities at all-time highs, no indiscriminate selling. The question this room asks — the museum's question, unanswered: what type is it?

The tape — frozen 9 Oct 2026
5.24%
10Y Treasury yield, Fri 9 Oct 2026 close. Printed 5.31% on Oct 5 — highest close since April 2002. (Yahoo Finance ^TNX; canonical series FRED DGS10)
−0.1%
S&P 500 vs its 6 Oct closing high (7,818.93 → 7,811.54). Nasdaq-100 −1.1% vs its 6 Oct high. Equities had not broken. (Yahoo Finance ^GSPC, QQQ; FRED SP500, NASDAQ100)
14.8
VIX at the 9 Oct close. Compare 80+ in 2008 and the March 2020 record: no phase-one everything-sells. (Yahoo Finance ^VIX; FRED VIXCLS)

All Wing V figures asof the Fri 9 Oct 2026 close. This room is frozen and does not update.

"A room with no ending yet. The museum can only ask its question here, not answer it — what type is it?"

Epilogue ↓
Epilogue

What the museum suggests

Three of the four finished wings forgave the patient. The one that didn't — the dot-com bust — was the one where the thesis broke. 1987 and 2020 were terrifying and fast, and both forgave quickly, because in neither did the economy or the credit system actually break. 2008 broke everything and took five and a half years; it also never retested the bottom, and the buyers of March 2009 were never given a second chance.

So the museum's question for the frozen wing is the same one on the brass plaque at the door: what type is it? The tape, as of 9 Oct 2026, showed duration at multi-decade highs against equities at all-time highs — the rate-type hypothesis, but only a hypothesis. Rate-types resolve one of two ways: duration heals, or weakness broadens into something the museum would have to reclassify.

The museum cannot tell you which. It is a museum, not an oracle. It can only tell you what to watch — and it can show you, room after room, that the worst trade in every wing was selling the panic and buying the complacency.

Walk out slowly. The fifth room is dated 9 Oct 2026 —
history hasn't decided what it is yet.