Critical Context: This framework worked in 2 cases (2000, 2008) and failed in others (2022, 2020, 1987). Understanding BOTH successes AND failures is essential for intellectual honesty.
1. Infrastructure showed stress 6-12 months before peak
2. Equipment (Cisco) triggered April 2000 when SPY was -11% from peak
3. Narrative leaders broke mid-2000
✓ Signal triggered at SPY -11%
✓ Avoided additional -37% decline
✓ 6-12 month early warning validated
✗ Didn't catch exact top
✗ 3-5 false positives during 1998-1999 melt-up
Infrastructure → Equipment → Narrative cascade confirmed. False positives during melt-up phase required patience. Composite approach (if used) would have provided more stable signal than single-stock tracking.
1. Mortgage originators (infrastructure) failed early 2007
2. Banks (capital providers) peaked May 2007
3. Broader financial sector followed
✓✓ Exceptional early warning (5 months before peak)
✓✓ Signal at SPY -7%, avoided -51% decline
✓✓ ZERO false positives - cleanest example
✓ Very clear cascade sequence
Best historical validation. Clear infrastructure layer (mortgage originators), obvious equipment layer (banks), strong narrative (housing never goes down). Sector-specific nature made framework highly applicable.
✗ Tier 1: No clear infrastructure leader, mixed signals
✗ Tier 2: SOXX fell WITH market (no relative underperformance)
✗ Tier 3: Mag 7 fell WITH market
✗ Result: NO CASCADE - everything fell together
Macro-driven correction (Fed rate hikes raising discount rates). All sectors compressed simultaneously. No speculative infrastructure buildout to fail first. System designed for sector bubbles, not policy-driven corrections.
Framework does NOT work for macro corrections. This is a known LIMITATION, not a flaw to hide. Acknowledging failures builds credibility. Clear boundaries: works for sector bubbles, not macro events.
External shock (pandemic). Everything fell in 3 weeks. Too fast for any warning system. No cascade pattern - simultaneous collapse across all sectors. System would not have helped.
Sudden panic (-20% in one day). Program trading/portfolio insurance feedback loop. No cascade, no warning. System irrelevant for sudden shocks.
Oil crisis and stagflation. Macro-driven (like 2022). Different dynamics than sector bubbles. System likely would not have applied.
| Period | Worked? | Signal Timing | DD Avoided | False Pos |
|---|---|---|---|---|
| 2000-2002 Dot-Com | ✓ Yes | SPY -11% | -37% | 3-5 |
| 2007-2009 Financial | ✓✓ Yes | SPY -7% | -51% | 0 |
| 2022 Bear | ✗ No | No signal | N/A | N/A |
| 2020 COVID | ✗ No | Too fast | N/A | N/A |
| 1987 Black Monday | ✗ No | No cascade | N/A | N/A |
Success Rate: 2/5 major downturns (40%)
But: 2/2 sector-specific bubbles (100% when applicable)
The current AI situation appears to be:
This resembles 2000 more than 2022, suggesting framework MAY be relevant.
However: Only 2 confirmed examples (small sample). 2025 may differ from 2000/2008. Could prove macro-driven like 2022. Past patterns do not guarantee future results.