How the overall assessment works

Four channels, one interpretation

The primary assessment combines valuation vulnerability, current financial stress, bank credit and employment. Its declared rules give observed stress and labor deterioration priority over valuation. Falling valuations cannot cancel an emerging crisis. Both SLOOS categories belong to one credit channel.

What changes the conclusion

• Valuation caution: the Board corporate equity-to-net-worth measure is in its top historical decile, using complete quarterly history since 1980.
• Market stress: the OFR global index is above its historical-average baseline.
• Credit warning: business lending shows net tightening above 20%, the Board significant-tightening boundary. Standards changes are not absolute tightness.
• Employment warning: payrolls fall over three months, or the exact current-vintage Sahm-style gap reaches 0.50 percentage points. This is not recession confirmation.
• Multiple warnings: financial stress coincides with credit or labor warnings. Broad stress requires financial stress, labor warning, and both same-quarter bank categories tightening above 20%.

The Treasury curve supplies a forward warning when inverted. A positive current curve does not erase earlier inversion risk. Each saved assessment exposes the exact policy, arithmetic, input dates and evidence fingerprints.

Freshness and missing evidence

ECO I checks sources on separate schedules. Daily, monthly and quarterly readings retain their actual periods and release lags. Missing or stale evidence reduces coverage; it cannot count as reassurance. The Sahm-style calculation needs all fifteen calendar months. The uncollected October 2025 CPS month is never interpolated or skipped.

A diagnosis, not a crash probability

The ECO I interpretation policy is explicit and reproducible, but has not been statistically calibrated to predict crashes. The 90th valuation percentile is a descriptive comparison, not a proven bubble threshold. Z.1 includes private corporations and inward FDI; OFR is global. Neither isolates AI or listed U.S. equities. Concentration, private leverage, bank runs and AI earnings expectations remain incompletely measured.

History needs its vintage

Charts and historical comparisons use observations selected for the displayed snapshot, including then-retained revisions. They are not a point-in-time predictive backtest. Every reading keeps its economic period, availability, retrieval and source identity.

Access and older results

The homepage and JSON API display the same saved assessment. Permitted raw source files, observations and processed readings remain accessible. The older 0–5 situation and mechanism scores are separate, limited models retained for historical compatibility; a score is not a percentage chance of a crash. Restricted inputs cannot enter the public primary assessment.