Debt debases your money. AI debases your labor. Own what survives both.
A scored, cross-asset book for the debt and AI regime. Every call timestamped, sized, benchmarked and kept, winners and losers alike.
The hedge is weakening. What to do now: Swap long bonds for T-bills and TIPS; hold gold as part of the hedge, adding when its trend is up.
Stage One is the world most of us grew up in: steady careers, institutions you could count on, and a stock and bond portfolio that worked because bonds rose when stocks fell. Stage Two is what replaces it: governments living on debt, AI repricing work, a world splitting into blocs. Debt debases money, inflation drives markets, and stocks and bonds fall together.
What is driving it, and how far to trust it →The table · Oct 24, 2026
| Asset class | Stage | 3-yr run | Timing | Change | As of |
|---|---|---|---|---|---|
| Bitcoin BTC ~$80.5K, above 200d $72.9K · flows rising | Normal no sign of excess | +139% 35% below high | Hold +11% vs 200d -40% vs S&P 12m | new | Oct 23 |
| Large-cap crypto · descriptive ETH ~$2.4K, above 200d $2.2K | Recovering rebound after a crash | +34% 51% below high | Hold +11% vs 200d -12% vs BTC 12m | new | n/a |
| Small-cap crypto · descriptive Small-cap index 102.9, above 200d 83.1 | Recovering rebound after a crash | -36% 84% below high | Hold +24% vs 200d -7% vs BTC 12m | new | n/a |
| Large caps SPY $760, above 200d $727 | Normal no sign of excess | +86% 2% below high | Add +5% vs 200d | new | Oct 21 |
| Small caps · descriptive IWM $276, below 200d $278 | Normal no sign of excess | +71% 10% below high | Hold -1% vs 200d -2% vs S&P 12m Less reliable here: the 200-day line has not worked on the Russell 2000 since 1987. | new | n/a |
| Gold and silver GLD $367, below 200d $414 · gold COT 52nd pct, silver COT 10th pct | Normal no sign of excess | +100% 26% below high | Reduce -11% vs 200d -16% vs S&P 12m | new | Oct 13 |
| Cards · descriptive VCP index $443M Sep, +124% y/y | Normal no sign of excess | +28% at its high | Add +14% vs 40w +15% vs S&P 12m | new | n/a |
| NFTs and digital art · descriptive Punks floor $73.1K, +22% in 13w | Post-peak | +11% 68% below high | Reduce -22% vs 200d -63% vs S&P 12m | new | n/a |
| AI collectibles Hand read, set 2026-09-16 | Early manual read | n/a | n/a | n/a | Sep 16 |
Positioning and sentiment
| Asset class | Read | Positioning | Sentiment |
|---|---|---|---|
| Bitcoin | Mixed | ●●●○○ | n/a |
| Large-cap crypto | Descriptive | n/a | n/a |
| Small-cap crypto | Descriptive | n/a | n/a |
| Large caps | Uncrowded | n/a | ●●○○○ |
| Small caps | Descriptive | n/a | n/a |
| Gold and silver | No cap | n/a | n/a |
| Cards | Descriptive | n/a | n/a |
| NFTs and digital art | Descriptive | n/a | n/a |
| AI collectibles | Not scored | n/a | n/a |
Context only. Each input is ranked against its own last three years, 1 (washed out) to 5 (crowded), and an input with less than two years of history is left out. Neither sorted later returns in our tests, so neither feeds Stage or Timing.
The read: Uncrowded when both scores are 2 or under and the trend is up; Forgotten (printed as Uncrowded) when both are 2 or under and the trend is down; Crowded when both are 4 or over and the trend is up; Unwinding when both are 4 or over and the trend is down; Mixed anything in between. A sideways week keeps the previous read within its zone. No read means a score or the trend could not be had that week, rather than last week's number.
Only Large caps, Bitcoin, Gold & silver have inputs with tested forward information; the other rows describe conditions and make no forward claim. Every input declares its direction before it is tested, and one that points the wrong way after a year of live logging is dropped.
How to read this table
The asset classes we follow, read every week. AI collectibles carry a hand read only: no price series yet. Stage reads price alone: the three-year run, how far the price sits above its three-year median, and the 12-month return against its benchmark. Normal means no sign of excess, which is most of the time. Stretched is a big run (a mild warning), Late an extreme one, post-peak in red after a doubling and a 40% fall, then recovering once the price is back above its 200-day. The 3-yr run column shows the run and the distance below the three-year high; hover it for the terms. Positioning and sentiment run 1 (washed out) to 5 (crowded), each input ranked against its own last three years. they read together as uncrowded when both are 2 or under, crowded when both are 4 or over with the trend up, unwinding with it down, mixed otherwise. They sit under the table as context: neither sorted later returns in our tests. Timing is the next 1 to 6 months from price alone, with the distance from the 200-day and the 12-month return against its benchmark under the word: bitcoin for large-cap crypto, small-cap crypto and memes, the S&P 500 for the rest. Cards read the card screener's weekly vintage index on a 40-week line, the 200-day's span; it is built from trailing-year median sales, so it is smooth and lags card prices by months. Add: above a rising 200-day and ahead of the benchmark over 12 months (an uptrend that lags reads Hold). Small adds: more than 20% above it, still working but twice as likely to drop 15% in 3 months. Hold: flat or the line turning down; keep it, no new money. Wait: below the 200-day; no new money until it is back above. Reduce: below the 200-day after a 3-year run of 100% or more; cut the position back, not a full exit. In a test on 49 US industries since 1927, being below the 200-day doubled the chance of a 15% fall within 3 months, and Reduce lagged the market by about 2% over 3 months. Rows marked descriptive make no forward claim. The as-of date is a row's oldest input.
The pages
The ledger
every call →No calls yet. The first one goes on the record whether it works or not.
One email when a regime flips: Stage Two phase or a return-path tripwire, the dominant season, the Gold Regime band, or S&P Correction flags reaching two or three. Each links to the page that explains it. Free.
No alerts triggered currently.