GLOBALCRASH · SCORECARD
The gauge, graded in public
For every verified crisis since 1990: what this index read six months before the peak of the episode, and at the peak itself — computed from the same reconstructed series shown on the chart. No cherry-picking; misses included.
| Episode | 6M before | At peak | S&P |
|---|---|---|---|
| Gulf War '90 | 50.8 | 57.3 | -19.9% |
| Asia '97 | 50.4 | 52.9 | -10.8% |
| LTCM '98 | 51.9 | 69.7 | -19.3% |
| Dot-com '02 | 64.8 | 67.5 | -49.1% |
| 9/11 '01 | 79.1 | 72.4 | -11.6% |
| Lehman '08 | 65.5 | 63.7 | -56.8% |
| Flash Crash '10 | 46.3 | 40.8 | -16.0% |
| Euro debt '11 | 38.0 | 44.3 | -19.4% |
| China '15 | 37.8 | 40.1 | -14.2% |
| Volmageddon '18 | 46.7 | 45.1 | -10.2% |
| Q4 2018 '18 | 43.3 | 45.6 | -19.8% |
| Repo '19 | 46.8 | 50.6 | -4.1% |
| COVID '20 | 50.8 | 51.5 | -33.9% |
| Terra '22 | 43.4 | 57.8 | -18.7% |
| 2022 bear '22 | 55.8 | 61.9 | -25.4% |
| FTX '22 | 58.9 | 63.3 | -22.4% |
| SVB '23 | 60.7 | 59.9 | -7.8% |
| Yen carry '24 | 60.1 | 60.3 | -8.5% |
| Tariffs '25 | 61.2 | 61.7 | -18.9% |
| AI scare '25 | 60.6 | 59.7 | -5.1% |
| Iran '26 | 58.4 | 59.0 | -9.1% |
7 of the 9 bear-market episodes since 1990 had this index at 50 or higher six months before peak stress — computed from the same series on the chart, re-graded at every print.
Base rates, against ourselves. 53% of ALL months since 1990 read 50 or higher — so "above 50" alone is weak evidence, and the table above should be read with that in mind. Of the 113 months at 60+, 46% were false alarms — no anchored episode within the following 12 months. That is the noise-to-signal trade-off (Kaminsky & Reinhart) every early-warning system pays; we publish ours instead of hiding it. Monthly granularity.
The honest miss: COVID-19. The index read low into January 2020 — a pandemic is invisible to financial data until it hits. What the gauge measures is how much fragility a shock lands on, not the shock itself.