Normalization or Destruction: A Trend-Break Test for the Asset Drain
Builds-on: how-inflation-dies-the-empty-reservoir (the asset-side audit section) Related: failure-cascade-index, two-economy-gauge (this doc is the asset-side sibling — it gauges the top pool's funding valve), ai-circular-financing-and-banking-exposure-audit Informs: regime-check rotation
Purpose
The asset-side audit in how-inflation-dies-the-empty-reservoir found nearly every asset class already declining, but flagged the honest caveat: a fall from a 2021-22 speculative peak back to trend is normalization (demand curve intact, just repriced), not destruction (the buyer cohort is gone). The distinction matters because only destruction drains the top-tier reservoir in the way that transmits to consumption — and because the episodes that reorganized economies (US oil post-1979, EU industry post-2022) broke below trend and stayed there. This doc turns the distinction into three operational tests plus a standing set of forced-selling gauges.
Unlike the two gauge docs, this is primarily a test to run, with a small recurring component. The trend-break table gets computed once and refreshed quarterly; the forced-selling gauges run continuously.
The three tests
Test 1 — Trend-break, not peak-drawdown (state variable)
Fit a log-linear trend to each asset class over 2013-2019 (pre-distortion), extrapolate to present, and measure current price against trend, not against the 2021-22 peak. Peak-drawdown numbers (BTC -51%, watches -31%) are emotionally loud and analytically empty — the peaks were the anomaly.
- Above trend = still normalizing; more room to fall is the base case, and the decline so far is not evidence of a broken market.
- At trend = normalization complete; from here, further decline starts meaning something.
- Below trend and staying = the destruction signature.
Caveats: assets without a defensible pre-2020 trend (crypto — the 2013-2019 "trend" is itself a bubble sequence) get a qualitative treatment or an on-chain cost-basis proxy instead. Housing needs a rent-ratio or income-ratio version of the trend, not just price.
Test 2 — Volume recovery at the lower price (the discriminator)
This is the most powerful test and the least watched. When price falls to a level and transaction volume recovers, buyers exist at that price — demand curve intact, repriced: normalization. When price falls and volume stays dead, the buyer cohort is gone: destruction, with further price discovery still ahead (illiquid assets mark slowly; owners "operate out of working capital and hope," same option structure as the businesses in failure-cascade-index).
Timing note: in illiquid assets, volume leads price. Existing-home sales at multi-decade lows with sticky prices is not stability — it's destruction that hasn't been marked yet. The price series is the lagging indicator in every illiquid market; the volume series is the leading one. This is the single most common leading/lagging trap on the asset side.
Test 3 — The marginal buyer's funding (diagnosis)
When the bid returns, check what funds it. If buyers return only with cheaper leverage (rate cuts, financing incentives), the prior decline was rate-normalization and reverses with rates. If buyers return with cash at the lower price, demand was intact all along. If buyers don't return at any financing, the cohort is gone — destruction. This test can only be run after a bid reappears, so it's confirmatory, not predictive.
Forced-selling gauges (the recurring component)
Destruction becomes visible when holders stop being able to wait. These run continuously and are the leading edge of the whole asset-side watch — they also arm the top-pool alarm in two-economy-gauge:
| Gauge | Source | Freq | Timing | Reading (July 2026) |
|---|---|---|---|---|
| Secondaries discount × volume | Jefferies/Evercore secondary market reports | Semi-annual | Leading — widening discount on rising volume = distress (LPs must sell); narrowing discount on rising volume = market clearing (healthy) | Record $225B volume (2025) at ~-30% RE discounts after four years of below-trend distributions — the only clearly visible forced selling on the board |
| REIT NAV gap | S&P Global / Green Street | Monthly | Leading for private CRE marks (public price discovers first, private marks follow) | -16.2% median entering 2026 |
| Auction sell-through rate + guarantee share | Christie's/Sotheby's/Phillips season reports; Hagerty auction data | Seasonal | Coincident — sell-through falling with houses cutting guarantees = bid withdrawal in real time | "Return to quality" 2025 = tail bid already withdrawn |
| Margin debt YoY | FINRA monthly | Monthly | Coincident-to-lagging for equities, but the unwind speed gauge — high margin into a decline converts orderly repricing into forced selling | Pull on first run |
| Existing-home sales count vs. price | NAR monthly | Monthly | Volume leads, price lags (Test 2 applied to housing) | Sales near multi-decade lows, prices sticky, 77 of 300 metros negative — flashing the destruction pattern, not yet marked |
Initial qualitative pass (to be replaced by the computed table)
| Asset | vs. 2013-19 trend (est.) | Volume state | Provisional verdict |
|---|---|---|---|
| Sun Belt housing | Likely still above trend despite declines | Dead | Worst configuration: above trend and no volume — normalization has further to run, and the volume signature says the eventual mark is destruction-shaped |
| Supply-constrained housing (NE/MW/Seattle) | Above trend | Low but functional | Scarcity floor, not demand strength — a different regime, priced by inventory not by buyers' reservoirs |
| Watches (ex-holy-trinity) | Probably at/near trend after 13 down quarters | Grey-market volume exists at lower prices | Mostly normalization, approaching complete |
| Collector cars | Mixed by segment; rotation active | Volume rotating, not vanishing | Normalization + generational hand-off (buyer cohort changing, not disappearing) |
| Art (speculative tail) | Below its own 2015-19 trend | Lots withdrawn, sell-through weak in the tail | Closest to true destruction — the tail-buyer cohort (cheap-money speculation) is gone |
| Crypto | No defensible trend; -51% from peak | Volume present | Neither test applies cleanly; treat as sentiment gauge for the speculative tier, not a reservoir |
| CRE office | Below any pre-2020 trend, 3+ years | Transactions scarce; forced sales emerging | Destruction confirmed (the WFH cohort-loss is structural — the 1979-oil analog of this cycle) |
| PE/VC (cash price) | Marks above trend; cash price ~30% below marks | Secondaries volume at records | The gap is the measurement: the mark says normalization, the cash price says repricing well underway |
| AI mega-cap equities | Far above trend | Deep, liquid | The untested asset — this doc's tests apply the day the drawdown starts; until then it's the two-economy-gauge top-pool valve |
Reading protocol
Quarterly: refresh the forced-selling table; recompute the trend-break table for any asset whose price moved >10% or whose volume state changed. The verdict that matters for the macro thread is a single sentence per quarter: is the top pool's funding valve (equities + scarce housing) still the only thing above trend with healthy volume? When the answer stops being yes, how-inflation-dies-the-empty-reservoir scenario B is live and the two-economy-gauge alarm configuration should already be flashing.
First-run tasks
- Compute the actual trend-break table: Case-Shiller national + Seattle + a Sun Belt metro, Hagerty index, WatchCharts index, Artprice index, Green Street CPPI, S&P 500 — log-linear 2013-2019 fit, current deviation.
- Housing in ratio form: price-to-rent and price-to-income vs their 2013-19 means, national and by region.
- Pull FINRA margin debt and NAR volume series; baseline the forced-selling table.
- Decide the crypto treatment (exclude, or use realized-cap/cost-basis on-chain proxies).