Interactive Tool
AI Bubble Meter
An experiment in AI-assisted predictive modeling: two backtested dials measure whether the AI boom has the makings of a bubble, and what could make it burst.
Backtested 1965–2026 · Data as of October 17, 2026
Two dials: conditions and triggers
The meter separates bubble conditions (valuations, household equity exposure, price run-up, the investment share of GDP) from bubble triggers (credit growth and the rise in the 2-year Treasury yield). A quarterly backtest since 1965 shows why: conditions say how far there is to fall, but on their own they have had almost no power to time a fall. The triggers are what separated booms that bust within two years from booms that kept going. Read the two dials together.
A note on this experiment. The AI Bubble Meter is an experiment in using AI as a predictive-modeling tool. We built it with Anthropic’s Claude (Fable 5.1) to assess whether the stock market has the makings of a bubble and what might cause one to burst. This is the first iteration of an ongoing project: not every data point and rule has been independently audited, and the model is not perfect. Applying our own judgment and reading of the market, we consider its findings directionally accurate. Every calculation uses public data, and its limits are set out in the Notes. We welcome feedback.
Conditions
Average percentile rank vs. all prior history: CAPE 98th (Sep 2026), household equity share 100th (Q2 2026), 2-yr real run-up 75th (Sep 2026), investment share of GDP 60th (Q2 2026).
AI-specific scorecard (5 conditions): 90 High
Triggers
3-yr change in private credit/GDP 6th percentile (−13.1pp to Q4 2025, latest available); 2-yr Treasury yield rise from its trailing 8-quarter low 61st percentile (+1.15pp to 4.65% in Sep 2026, from 3.50% in Dec 2025).
For reference: fed funds is still 0.58pp below its level six quarters ago despite the Sep 16 hike (35th percentile), and the 2-yr sits 0.90pp above fed funds (83rd percentile), so the market is pricing further hikes.
AI-specific scorecard (5 triggers): 50 Moderate
Historical analog: High conditions, Low triggers
9% chance of a 20%+ decline within 2 years
In the 54 quarters since 1965 with this mix, the median real S&P 500 return over the next two years was +18.3% and the median worst drawdown was −2.8%. Base rate for all quarters: 15% and +9.2%.
If triggers turn High with conditions this stretched (19 quarters): 42% chance of a 20%+ decline within 2 years (68% within 3), median 2-yr real return −12.5%.
Backtest
A mechanical version of the meter, scored quarterly from 1965 with no hindsight, tested against every S&P 500 decline of 20% or more.
What the backtest found
Conditions alone do not time a bust.
Valuation, participation and run-up readings rank quarters that were followed by a 20%+ decline barely better than a coin flip (AUC 0.61; CAPE 0.55, 2-yr run-up 0.46). They were High for four years before the 2000 peak and for most of 2017–2021.Credit and the 2-year yield carry the signal.
The triggers dial (credit growth plus the rise in the 2-yr Treasury yield from its trailing two-year low) has an AUC of 0.79. When it read High, 42% of quarters with High conditions saw a 20%+ decline within two years, versus 9% when triggers were Low. The 2-yr yield is used rather than the fed funds rate because it moves first: on its own it scores 0.63 against 0.62 for fed funds, and it flagged the 1987 and 2000 peaks as High. This matches Jordà–Schularick–Taylor and Borio–Lowe: credit is what makes a boom dangerous.One number would hide the signal.
An equal-weight composite of the six indicators has an AUC of 0.70. Composite readings of 67–75 were followed by a 20%+ decline 15% of the time, the same as the base rate, because a stretched market with no credit boom scores the same as a credit boom with modest valuations. Only readings above 75 were informative (31%, median 2-yr real return +1.4%). Two dials keep the distinction visible.The sample is small.
Six qualifying peaks (2020 is shown for reference) and 238 overlapping quarters since 1965. The triggers dial was Low at the 1968 peak and only Moderate at the 1972 and 2021 peaks, none of which were credit booms, and it was High for three years before the 2007 peak but had eased to Moderate by the peak quarter, so early as much as wrong. Treat the probabilities as rough ratios, not forecasts.Outcomes by dial reading, 1965–2024
Each cell: share of quarters followed by a 20%+ S&P 500 decline within two years, and the median real S&P 500 return over the following two years. The highlighted cell is where the mechanical readings sit now.
| Conditions ↓ / Triggers → | Triggers Low | Triggers Moderate | Triggers High |
|---|---|---|---|
| Conditions Low | 0% 2-yr real +7.7% · n=45 | 19% 2-yr real +11.8% · n=26 | 14% 2-yr real +5% · n=14 |
| Conditions Moderate | 7% 2-yr real +14.2% · n=41 | 18% 2-yr real +3.5% · n=17 | 80% 2-yr real −1.8% · n=5 |
| Conditions High | 9% 2-yr real +18.3% · n=54 Now | 29% 2-yr real +7.8% · n=17 | 42% 2-yr real −12.5% · n=19 |
Which indicators carried the signal
| Indicator | Block | AUC | Read |
|---|---|---|---|
| 3-yr change in private credit/GDP | Trigger | 0.73 | Useful |
| Gross private investment, % of GDP | Condition | 0.67 | Weak |
| IPO count × first-day return (prior year, from 1981) | Reference | 0.64 | Weak |
| Household equity share of financial assets | Condition | 0.63 | Weak |
| 2-yr Treasury yield, rise from trailing 8-qtr low | Trigger | 0.63 | Weak |
| 6-qtr change in fed funds rate (not in the dial) | Reference | 0.62 | Weak |
| Shiller CAPE | Condition | 0.55 | No better than chance |
| 2-yr real S&P 500 return | Condition | 0.46 | No better than chance |
| 2-yr Treasury minus fed funds (not in the dial) | Reference | 0.29 | Inverted: busts followed negative spreads more often (0.71 with the sign flipped). A late-cycle signal, not a trigger. |
| Triggers dial (credit + 2-yr yield) | 0.79 | Useful | |
| Conditions dial (CAPE + household + run-up + investment) | 0.61 | Weak | |
| Equal-weight composite of all six | 0.70 | Useful |
Readings at past peaks
| Peak | Conditions | Triggers | Composite | Max decline, next 2 yrs | 2-yr real return |
|---|---|---|---|---|---|
| Go-go / conglomerate boom (S&P peak Dec 1968) | 78 | 41 | 65 | −29% | −24.4% |
| Nifty Fifty (S&P peak Jan 1973) | 69 | 53 | 64 | −42.9% | −53.3% |
| 1987 crash (peak Aug 1987) | 66 | 85 | 73 | −24.4% | +0.2% |
| Tech/telecom (peak Mar 2000) | 91 | 84 | 89 | −27.6% | −23.4% |
| Housing/credit (peak Oct 2007) | 79 | 59 | 73 | −49.4% | −32.6% |
| Pandemic (peak Feb 2020) (reference: no 20%+ decline on monthly averages) | 57 | 26 | 46 | None | +48.5% |
| Post-pandemic (peak Dec 2021) | 90 | 50 | 77 | −20.3% | −8.8% |
| Now (latest data, Sep 2026) | 83 | 34 | 67 | — | — |
Why not one number: outcomes by equal-weight composite reading
| Composite reading | Quarters | 20%+ decline within 2 yrs | Within 3 yrs* | Median 2-yr real return |
|---|---|---|---|---|
| 0–40 | 57 | 4% | 7% | +8.5% |
| 40–50 | 42 | 12% | 14% | +9.2% |
| 50–60 | 54 | 15% | 20% | +8.3% |
| 60–67 | 33 | 21% | 23% | +12% |
| 67–75 | 20 | 15% | 16% | +14.6% |
| 75+ | 32 | 31% | 47% | +1.4% |
| All quarters | 238 | 15% | 20% | +9.2% |
* Three-year columns cover quarters through Q3 2023 (234 of 238); the quarters excluded, Q4 2023 to Q3 2024, all fall in the High-conditions, Low-triggers cell, whose three-year figure is 5 of 50.
Method: six indicators with history to the 1950s are each converted to a percentile rank against all prior quarters (expanding window, minimum 10 years of history, so no hindsight). Conditions = mean of CAPE, household equity share, 2-yr real S&P 500 return and gross private investment/GDP percentiles; Triggers = mean of the 3-yr change in BIS private non-financial credit/GDP and the rise in the 2-yr Treasury yield from its lowest level in the trailing eight quarters (quarter-end months; FRED GS2 from 1976, GS1 before). Zones: Low below 50, Moderate 50–66, High 67+. Outcome: a 20%+ fall in the S&P 500 monthly-average price within the next 24 months. Quarters since Q3 2024 cannot yet be scored for outcomes. The IPO series (Ritter) starts in 1980 and is reported separately. Sources: Fed Z.1 and BEA via FRED, BIS via FRED, FRED GS2, GS1, FEDFUNDS and CPIAUCSL, multpl.com (Shiller CAPE and S&P 500 monthly averages), Jay Ritter. Economy-wide credit/GDP has been falling since 2021, which is why the mechanical Triggers dial reads Low even as the 2-yr yield rises and AI-specific financing accelerates: with credit at the 6th percentile, the rate trigger alone cannot lift the dial to Moderate unless the 2-yr rise reaches the 94th percentile of history, about +3.7pp from its low. The AI-specific scorecard is the overlay for that. A fully forward-looking rate measure, the 2-yr yield minus fed funds (hikes priced but not yet delivered), was tested and rejected as a trigger: AUC 0.29, with the sign inconsistent across peaks (+0.68pp at 2000Q1 and +1.12pp at 1987Q3, but -0.93pp at 2007Q3 and -0.2pp at 2020Q1). With the sign flipped it is a late-cycle signal (0.71), firing when tightening is over, so it is shown for reference only (table above and Fig. 3).
Conditions vs. Triggers, 1965–2026
(mechanical dial readings, 0–100; shaded = S&P 500 declines of 20%+)
Busts came when both dials were High; conditions alone ran hot for years
- Conditions (Q3 2026 = 83)
- Triggers (Q3 2026 = 34)
- S&P 500 decline of 20%+
| Date | Conditions | Triggers |
|---|---|---|
| Q2 1965 | 80 | 39 |
| Q3 1965 | 87 | 36 |
| Q4 1965 | 85 | 28 |
| Q1 1966 | 78 | 30 |
| Q2 1966 | 66 | 30 |
| Q3 1966 | 47 | 40 |
| Q4 1966 | 53 | 30 |
| Q1 1967 | 60 | 16 |
| Q2 1967 | 54 | 18 |
| Q3 1967 | 65 | 31 |
| Q4 1967 | 64 | 37 |
| Q1 1968 | 50 | 31 |
| Q2 1968 | 72 | 34 |
| Q3 1968 | 71 | 27 |
| Q4 1968 | 78 | 41 |
| Q1 1969 | 65 | 43 |
| Q2 1969 | 54 | 50 |
| Q3 1969 | 44 | 55 |
| Q4 1969 | 30 | 56 |
| Q1 1970 | 15 | 39 |
| Q2 1970 | 7 | 50 |
| Q3 1970 | 9 | 38 |
| Q4 1970 | 11 | 2 |
| Q1 1971 | 36 | 11 |
| Q2 1971 | 35 | 50 |
| Q3 1971 | 35 | 48 |
| Q4 1971 | 35 | 34 |
| Q1 1972 | 55 | 37 |
| Q2 1972 | 63 | 45 |
| Q3 1972 | 60 | 53 |
| Q4 1972 | 69 | 53 |
| Q1 1973 | 51 | 68 |
| Q2 1973 | 35 | 66 |
| Q3 1973 | 35 | 70 |
| Q4 1973 | 27 | 63 |
| Q1 1974 | 23 | 66 |
| Q2 1974 | 24 | 73 |
| Q3 1974 | 19 | 72 |
| Q4 1974 | 20 | 54 |
| Q1 1975 | 4 | 18 |
| Q2 1975 | 5 | 19 |
| Q3 1975 | 5 | 34 |
| Q4 1975 | 9 | 12 |
| Q1 1976 | 22 | 5 |
| Q2 1976 | 32 | 19 |
| Q3 1976 | 46 | 11 |
| Q4 1976 | 46 | 3 |
| Q1 1977 | 39 | 17 |
| Q2 1977 | 35 | 17 |
| Q3 1977 | 38 | 28 |
| Q4 1977 | 33 | 36 |
| Q1 1978 | 28 | 45 |
| Q2 1978 | 30 | 51 |
| Q3 1978 | 32 | 58 |
| Q4 1978 | 29 | 63 |
| Q1 1979 | 31 | 66 |
| Q2 1979 | 31 | 69 |
| Q3 1979 | 35 | 74 |
| Q4 1979 | 34 | 76 |
| Q1 1980 | 31 | 78 |
| Q2 1980 | 31 | 32 |
| Q3 1980 | 31 | 64 |
| Q4 1980 | 44 | 69 |
| Q1 1981 | 44 | 64 |
| Q2 1981 | 41 | 58 |
| Q3 1981 | 29 | 54 |
| Q4 1981 | 32 | 52 |
| Q1 1982 | 24 | 57 |
| Q2 1982 | 21 | 55 |
| Q3 1982 | 21 | 14 |
| Q4 1982 | 17 | 8 |
| Q1 1983 | 29 | 6 |
| Q2 1983 | 41 | 25 |
| Q3 1983 | 49 | 33 |
| Q4 1983 | 50 | 33 |
| Q1 1984 | 52 | 47 |
| Q2 1984 | 50 | 65 |
| Q3 1984 | 51 | 64 |
| Q4 1984 | 44 | 51 |
| Q1 1985 | 45 | 64 |
| Q2 1985 | 45 | 49 |
| Q3 1985 | 41 | 58 |
| Q4 1985 | 54 | 52 |
| Q1 1986 | 60 | 52 |
| Q2 1986 | 61 | 52 |
| Q3 1986 | 56 | 53 |
| Q4 1986 | 59 | 53 |
| Q1 1987 | 65 | 58 |
| Q2 1987 | 64 | 76 |
| Q3 1987 | 66 | 85 |
| Q4 1987 | 50 | 76 |
| Q1 1988 | 47 | 68 |
| Q2 1988 | 47 | 78 |
| Q3 1988 | 46 | 84 |
| Q4 1988 | 46 | 87 |
| Q1 1989 | 45 | 89 |
| Q2 1989 | 48 | 68 |
| Q3 1989 | 51 | 58 |
| Q4 1989 | 60 | 38 |
| Q1 1990 | 50 | 54 |
| Q2 1990 | 55 | 34 |
| Q3 1990 | 39 | 27 |
| Q4 1990 | 36 | 15 |
| Q1 1991 | 47 | 9 |
| Q2 1991 | 43 | 6 |
| Q3 1991 | 43 | 6 |
| Q4 1991 | 44 | 4 |
| Q1 1992 | 49 | 17 |
| Q2 1992 | 48 | 7 |
| Q3 1992 | 55 | 3 |
| Q4 1992 | 58 | 18 |
| Q1 1993 | 56 | 8 |
| Q2 1993 | 55 | 13 |
| Q3 1993 | 55 | 4 |
| Q4 1993 | 59 | 15 |
| Q1 1994 | 58 | 27 |
| Q2 1994 | 60 | 39 |
| Q3 1994 | 57 | 42 |
| Q4 1994 | 56 | 49 |
| Q1 1995 | 63 | 46 |
| Q2 1995 | 68 | 40 |
| Q3 1995 | 71 | 43 |
| Q4 1995 | 77 | 39 |
| Q1 1996 | 79 | 39 |
| Q2 1996 | 84 | 43 |
| Q3 1996 | 87 | 43 |
| Q4 1996 | 88 | 35 |
| Q1 1997 | 91 | 40 |
| Q2 1997 | 94 | 40 |
| Q3 1997 | 94 | 42 |
| Q4 1997 | 94 | 39 |
| Q1 1998 | 97 | 33 |
| Q2 1998 | 95 | 38 |
| Q3 1998 | 93 | 43 |
| Q4 1998 | 96 | 45 |
| Q1 1999 | 97 | 59 |
| Q2 1999 | 95 | 70 |
| Q3 1999 | 93 | 73 |
| Q4 1999 | 95 | 77 |
| Q1 2000 | 91 | 84 |
| Q2 2000 | 92 | 84 |
| Q3 2000 | 93 | 77 |
| Q4 2000 | 82 | 67 |
| Q1 2001 | 69 | 49 |
| Q2 2001 | 69 | 49 |
| Q3 2001 | 63 | 50 |
| Q4 2001 | 59 | 50 |
| Q1 2002 | 61 | 60 |
| Q2 2002 | 58 | 50 |
| Q3 2002 | 48 | 48 |
| Q4 2002 | 51 | 50 |
| Q1 2003 | 47 | 48 |
| Q2 2003 | 54 | 52 |
| Q3 2003 | 62 | 64 |
| Q4 2003 | 67 | 66 |
| Q1 2004 | 68 | 62 |
| Q2 2004 | 76 | 76 |
| Q3 2004 | 81 | 71 |
| Q4 2004 | 85 | 78 |
| Q1 2005 | 87 | 83 |
| Q2 2005 | 81 | 84 |
| Q3 2005 | 80 | 84 |
| Q4 2005 | 82 | 87 |
| Q1 2006 | 81 | 92 |
| Q2 2006 | 77 | 87 |
| Q3 2006 | 80 | 86 |
| Q4 2006 | 81 | 80 |
| Q1 2007 | 78 | 71 |
| Q2 2007 | 82 | 79 |
| Q3 2007 | 79 | 59 |
| Q4 2007 | 73 | 54 |
| Q1 2008 | 58 | 54 |
| Q2 2008 | 56 | 75 |
| Q3 2008 | 42 | 63 |
| Q4 2008 | 20 | 50 |
| Q1 2009 | 15 | 50 |
| Q2 2009 | 21 | 53 |
| Q3 2009 | 31 | 45 |
| Q4 2009 | 34 | 31 |
| Q1 2010 | 39 | 23 |
| Q2 2010 | 33 | 7 |
| Q3 2010 | 40 | 5 |
| Q4 2010 | 61 | 11 |
| Q1 2011 | 66 | 14 |
| Q2 2011 | 61 | 5 |
| Q3 2011 | 43 | 5 |
| Q4 2011 | 48 | 10 |
| Q1 2012 | 58 | 11 |
| Q2 2012 | 56 | 11 |
| Q3 2012 | 61 | 10 |
| Q4 2012 | 52 | 10 |
| Q1 2013 | 62 | 11 |
| Q2 2013 | 67 | 14 |
| Q3 2013 | 74 | 16 |
| Q4 2013 | 75 | 14 |
| Q1 2014 | 73 | 15 |
| Q2 2014 | 79 | 18 |
| Q3 2014 | 79 | 21 |
| Q4 2014 | 83 | 23 |
| Q1 2015 | 83 | 24 |
| Q2 2015 | 81 | 24 |
| Q3 2015 | 72 | 26 |
| Q4 2015 | 73 | 30 |
| Q1 2016 | 68 | 29 |
| Q2 2016 | 66 | 26 |
| Q3 2016 | 67 | 27 |
| Q4 2016 | 72 | 32 |
| Q1 2017 | 73 | 35 |
| Q2 2017 | 77 | 38 |
| Q3 2017 | 81 | 42 |
| Q4 2017 | 82 | 49 |
| Q1 2018 | 83 | 54 |
| Q2 2018 | 82 | 56 |
| Q3 2018 | 86 | 56 |
| Q4 2018 | 76 | 50 |
| Q1 2019 | 80 | 42 |
| Q2 2019 | 80 | 32 |
| Q3 2019 | 80 | 25 |
| Q4 2019 | 77 | 14 |
| Q1 2020 | 57 | 26 |
| Q2 2020 | 65 | 42 |
| Q3 2020 | 78 | 43 |
| Q4 2020 | 89 | 49 |
| Q1 2021 | 85 | 51 |
| Q2 2021 | 81 | 48 |
| Q3 2021 | 86 | 46 |
| Q4 2021 | 90 | 51 |
| Q1 2022 | 92 | 61 |
| Q2 2022 | 77 | 65 |
| Q3 2022 | 71 | 60 |
| Q4 2022 | 70 | 57 |
| Q1 2023 | 65 | 53 |
| Q2 2023 | 68 | 50 |
| Q3 2023 | 68 | 50 |
| Q4 2023 | 70 | 48 |
| Q1 2024 | 76 | 43 |
| Q2 2024 | 87 | 35 |
| Q3 2024 | 88 | 6 |
| Q4 2024 | 87 | 23 |
| Q1 2025 | 89 | 18 |
| Q2 2025 | 83 | 18 |
| Q3 2025 | 86 | 7 |
| Q4 2025 | 85 | 7 |
| Q1 2026 | 81 | 19 |
| Q2 2026 | 84 | 26 |
| Q3 2026 | 83 | 34 |
Source: YRI calculations from Fed Z.1, BEA, BIS and FRED data; multpl.com; see method note.
Note: Readings for 2026 carry the latest credit value (Q4 2025) forward; Q3 2026 also carries the Q2 household and investment values.
Note: Shaded areas are peak-to-trough periods of S&P 500 declines of 20% or more, measured on monthly averages.
Equal-Weight Composite, 1965–2026
(mean percentile of six indicators, 0–100; shaded = S&P 500 declines of 20%+)
An equal-weight composite crossed 75 before 2000, 2007 and 2021, but also sat above 67 for long stretches without a bust
- Composite (6 indicators) (Q3 2026 = 67)
- S&P 500 decline of 20%+
| Date | Composite (6 indicators) |
|---|---|
| Q2 1965 | 66 |
| Q3 1965 | 70 |
| Q4 1965 | 66 |
| Q1 1966 | 62 |
| Q2 1966 | 54 |
| Q3 1966 | 45 |
| Q4 1966 | 45 |
| Q1 1967 | 45 |
| Q2 1967 | 42 |
| Q3 1967 | 53 |
| Q4 1967 | 55 |
| Q1 1968 | 44 |
| Q2 1968 | 59 |
| Q3 1968 | 56 |
| Q4 1968 | 65 |
| Q1 1969 | 57 |
| Q2 1969 | 53 |
| Q3 1969 | 47 |
| Q4 1969 | 39 |
| Q1 1970 | 23 |
| Q2 1970 | 21 |
| Q3 1970 | 19 |
| Q4 1970 | 8 |
| Q1 1971 | 28 |
| Q2 1971 | 40 |
| Q3 1971 | 39 |
| Q4 1971 | 34 |
| Q1 1972 | 49 |
| Q2 1972 | 57 |
| Q3 1972 | 58 |
| Q4 1972 | 64 |
| Q1 1973 | 57 |
| Q2 1973 | 45 |
| Q3 1973 | 47 |
| Q4 1973 | 39 |
| Q1 1974 | 38 |
| Q2 1974 | 40 |
| Q3 1974 | 36 |
| Q4 1974 | 32 |
| Q1 1975 | 9 |
| Q2 1975 | 9 |
| Q3 1975 | 15 |
| Q4 1975 | 10 |
| Q1 1976 | 16 |
| Q2 1976 | 28 |
| Q3 1976 | 34 |
| Q4 1976 | 32 |
| Q1 1977 | 32 |
| Q2 1977 | 29 |
| Q3 1977 | 34 |
| Q4 1977 | 34 |
| Q1 1978 | 34 |
| Q2 1978 | 37 |
| Q3 1978 | 41 |
| Q4 1978 | 41 |
| Q1 1979 | 43 |
| Q2 1979 | 44 |
| Q3 1979 | 48 |
| Q4 1979 | 48 |
| Q1 1980 | 47 |
| Q2 1980 | 32 |
| Q3 1980 | 42 |
| Q4 1980 | 52 |
| Q1 1981 | 51 |
| Q2 1981 | 47 |
| Q3 1981 | 37 |
| Q4 1981 | 39 |
| Q1 1982 | 35 |
| Q2 1982 | 32 |
| Q3 1982 | 19 |
| Q4 1982 | 14 |
| Q1 1983 | 21 |
| Q2 1983 | 36 |
| Q3 1983 | 44 |
| Q4 1983 | 45 |
| Q1 1984 | 50 |
| Q2 1984 | 55 |
| Q3 1984 | 56 |
| Q4 1984 | 47 |
| Q1 1985 | 51 |
| Q2 1985 | 46 |
| Q3 1985 | 46 |
| Q4 1985 | 53 |
| Q1 1986 | 58 |
| Q2 1986 | 58 |
| Q3 1986 | 55 |
| Q4 1986 | 57 |
| Q1 1987 | 62 |
| Q2 1987 | 68 |
| Q3 1987 | 73 |
| Q4 1987 | 59 |
| Q1 1988 | 54 |
| Q2 1988 | 57 |
| Q3 1988 | 58 |
| Q4 1988 | 60 |
| Q1 1989 | 60 |
| Q2 1989 | 55 |
| Q3 1989 | 53 |
| Q4 1989 | 53 |
| Q1 1990 | 52 |
| Q2 1990 | 48 |
| Q3 1990 | 35 |
| Q4 1990 | 29 |
| Q1 1991 | 35 |
| Q2 1991 | 31 |
| Q3 1991 | 30 |
| Q4 1991 | 31 |
| Q1 1992 | 38 |
| Q2 1992 | 34 |
| Q3 1992 | 38 |
| Q4 1992 | 45 |
| Q1 1993 | 40 |
| Q2 1993 | 41 |
| Q3 1993 | 38 |
| Q4 1993 | 45 |
| Q1 1994 | 48 |
| Q2 1994 | 53 |
| Q3 1994 | 52 |
| Q4 1994 | 54 |
| Q1 1995 | 57 |
| Q2 1995 | 59 |
| Q3 1995 | 61 |
| Q4 1995 | 64 |
| Q1 1996 | 66 |
| Q2 1996 | 70 |
| Q3 1996 | 72 |
| Q4 1996 | 70 |
| Q1 1997 | 74 |
| Q2 1997 | 76 |
| Q3 1997 | 77 |
| Q4 1997 | 75 |
| Q1 1998 | 75 |
| Q2 1998 | 76 |
| Q3 1998 | 76 |
| Q4 1998 | 79 |
| Q1 1999 | 84 |
| Q2 1999 | 86 |
| Q3 1999 | 86 |
| Q4 1999 | 89 |
| Q1 2000 | 89 |
| Q2 2000 | 89 |
| Q3 2000 | 87 |
| Q4 2000 | 77 |
| Q1 2001 | 62 |
| Q2 2001 | 62 |
| Q3 2001 | 59 |
| Q4 2001 | 56 |
| Q1 2002 | 61 |
| Q2 2002 | 55 |
| Q3 2002 | 48 |
| Q4 2002 | 51 |
| Q1 2003 | 47 |
| Q2 2003 | 53 |
| Q3 2003 | 63 |
| Q4 2003 | 67 |
| Q1 2004 | 66 |
| Q2 2004 | 76 |
| Q3 2004 | 78 |
| Q4 2004 | 83 |
| Q1 2005 | 85 |
| Q2 2005 | 82 |
| Q3 2005 | 82 |
| Q4 2005 | 84 |
| Q1 2006 | 85 |
| Q2 2006 | 80 |
| Q3 2006 | 82 |
| Q4 2006 | 80 |
| Q1 2007 | 76 |
| Q2 2007 | 81 |
| Q3 2007 | 73 |
| Q4 2007 | 66 |
| Q1 2008 | 57 |
| Q2 2008 | 62 |
| Q3 2008 | 49 |
| Q4 2008 | 30 |
| Q1 2009 | 26 |
| Q2 2009 | 31 |
| Q3 2009 | 35 |
| Q4 2009 | 33 |
| Q1 2010 | 34 |
| Q2 2010 | 24 |
| Q3 2010 | 29 |
| Q4 2010 | 44 |
| Q1 2011 | 49 |
| Q2 2011 | 42 |
| Q3 2011 | 30 |
| Q4 2011 | 35 |
| Q1 2012 | 42 |
| Q2 2012 | 41 |
| Q3 2012 | 44 |
| Q4 2012 | 38 |
| Q1 2013 | 45 |
| Q2 2013 | 49 |
| Q3 2013 | 54 |
| Q4 2013 | 55 |
| Q1 2014 | 54 |
| Q2 2014 | 59 |
| Q3 2014 | 60 |
| Q4 2014 | 63 |
| Q1 2015 | 63 |
| Q2 2015 | 62 |
| Q3 2015 | 57 |
| Q4 2015 | 58 |
| Q1 2016 | 55 |
| Q2 2016 | 53 |
| Q3 2016 | 54 |
| Q4 2016 | 58 |
| Q1 2017 | 60 |
| Q2 2017 | 64 |
| Q3 2017 | 68 |
| Q4 2017 | 71 |
| Q1 2018 | 74 |
| Q2 2018 | 73 |
| Q3 2018 | 76 |
| Q4 2018 | 67 |
| Q1 2019 | 67 |
| Q2 2019 | 64 |
| Q3 2019 | 62 |
| Q4 2019 | 56 |
| Q1 2020 | 46 |
| Q2 2020 | 57 |
| Q3 2020 | 67 |
| Q4 2020 | 76 |
| Q1 2021 | 74 |
| Q2 2021 | 70 |
| Q3 2021 | 72 |
| Q4 2021 | 77 |
| Q1 2022 | 82 |
| Q2 2022 | 73 |
| Q3 2022 | 67 |
| Q4 2022 | 66 |
| Q1 2023 | 61 |
| Q2 2023 | 62 |
| Q3 2023 | 62 |
| Q4 2023 | 62 |
| Q1 2024 | 65 |
| Q2 2024 | 70 |
| Q3 2024 | 61 |
| Q4 2024 | 66 |
| Q1 2025 | 65 |
| Q2 2025 | 61 |
| Q3 2025 | 59 |
| Q4 2025 | 59 |
| Q1 2026 | 60 |
| Q2 2026 | 64 |
| Q3 2026 | 67 |
Source: YRI calculations; see method note.
Note: Shown for reference; the dials above are the meter’s reading. On this composite only readings above 75 have been informative.
Note: Shaded areas are peak-to-trough periods of S&P 500 declines of 20% or more, measured on monthly averages.
2-Year Treasury Minus Fed Funds, 1965–2026
(percentage points, quarter-end months; positive = market prices hikes; shaded = S&P 500 declines of 20%+)
Reference, not a trigger: at +0.90pp the market is pricing further hikes, the widest gap since 2022, but past busts followed both signs of this spread
- 2-yr yield minus effective fed funds (Q3 2026 = +0.90pp)
- S&P 500 decline of 20%+
| Date | 2-yr yield minus effective fed funds |
|---|---|
| Q1 1965 | +0.01pp |
| Q2 1965 | -0.06pp |
| Q3 1965 | +0.18pp |
| Q4 1965 | +0.40pp |
| Q1 1966 | +0.31pp |
| Q2 1966 | -0.20pp |
| Q3 1966 | +0.42pp |
| Q4 1966 | -0.20pp |
| Q1 1967 | -0.18pp |
| Q2 1967 | +0.50pp |
| Q3 1967 | +1.25pp |
| Q4 1967 | +1.20pp |
| Q1 1968 | +0.53pp |
| Q2 1968 | -0.09pp |
| Q3 1968 | -0.33pp |
| Q4 1968 | +0.17pp |
| Q1 1969 | -0.45pp |
| Q2 1969 | -1.86pp |
| Q3 1969 | -1.33pp |
| Q4 1969 | -0.80pp |
| Q1 1970 | -0.79pp |
| Q2 1970 | -0.06pp |
| Q3 1970 | +0.44pp |
| Q4 1970 | +0.10pp |
| Q1 1971 | -0.02pp |
| Q2 1971 | +0.73pp |
| Q3 1971 | -0.14pp |
| Q4 1971 | +0.46pp |
| Q1 1972 | +0.84pp |
| Q2 1972 | +0.47pp |
| Q3 1972 | +0.65pp |
| Q4 1972 | +0.19pp |
| Q1 1973 | -0.24pp |
| Q2 1973 | -1.18pp |
| Q3 1973 | -2.47pp |
| Q4 1973 | -2.68pp |
| Q1 1974 | -1.59pp |
| Q2 1974 | -3.26pp |
| Q3 1974 | -2.47pp |
| Q4 1974 | -1.22pp |
| Q1 1975 | +0.57pp |
| Q2 1975 | +0.74pp |
| Q3 1975 | +1.51pp |
| Q4 1975 | +1.40pp |
| Q1 1976 | +1.37pp |
| Q2 1976 | +1.58pp |
| Q3 1976 | +1.17pp |
| Q4 1976 | +0.73pp |
| Q1 1977 | +1.40pp |
| Q2 1977 | +0.74pp |
| Q3 1977 | +0.57pp |
| Q4 1977 | +0.62pp |
| Q1 1978 | +0.79pp |
| Q2 1978 | +0.64pp |
| Q3 1978 | +0.12pp |
| Q4 1978 | -0.31pp |
| Q1 1979 | -0.30pp |
| Q2 1979 | -1.07pp |
| Q3 1979 | -1.37pp |
| Q4 1979 | -2.39pp |
| Q1 1980 | -2.31pp |
| Q2 1980 | -0.74pp |
| Q3 1980 | +0.70pp |
| Q4 1980 | -4.82pp |
| Q1 1981 | -1.13pp |
| Q2 1981 | -4.59pp |
| Q3 1981 | +0.59pp |
| Q4 1981 | +0.92pp |
| Q1 1982 | -0.49pp |
| Q2 1982 | +0.32pp |
| Q3 1982 | +1.47pp |
| Q4 1982 | +0.71pp |
| Q1 1983 | +0.89pp |
| Q2 1983 | +1.20pp |
| Q3 1983 | +1.34pp |
| Q4 1983 | +1.37pp |
| Q1 1984 | +1.40pp |
| Q2 1984 | +1.85pp |
| Q3 1984 | +0.90pp |
| Q4 1984 | +1.80pp |
| Q1 1985 | +2.13pp |
| Q2 1985 | +1.16pp |
| Q3 1985 | +1.06pp |
| Q4 1985 | -0.12pp |
| Q1 1986 | -0.27pp |
| Q2 1986 | +0.26pp |
| Q3 1986 | +0.46pp |
| Q4 1986 | -0.64pp |
| Q1 1987 | +0.29pp |
| Q2 1987 | +0.84pp |
| Q3 1987 | +1.12pp |
| Q4 1987 | +1.09pp |
| Q1 1988 | +0.69pp |
| Q2 1988 | +0.52pp |
| Q3 1988 | +0.27pp |
| Q4 1988 | +0.33pp |
| Q1 1989 | -0.17pp |
| Q2 1989 | -1.12pp |
| Q3 1989 | -0.74pp |
| Q4 1989 | -0.67pp |
| Q1 1990 | +0.35pp |
| Q2 1990 | +0.06pp |
| Q3 1990 | -0.12pp |
| Q4 1990 | 0.00pp |
| Q1 1991 | +0.98pp |
| Q2 1991 | +1.06pp |
| Q3 1991 | +0.73pp |
| Q4 1991 | +0.60pp |
| Q1 1992 | +1.71pp |
| Q2 1992 | +1.29pp |
| Q3 1992 | +0.67pp |
| Q4 1992 | +1.75pp |
| Q1 1993 | +0.88pp |
| Q2 1993 | +1.12pp |
| Q3 1993 | +0.76pp |
| Q4 1993 | +1.25pp |
| Q1 1994 | +1.66pp |
| Q2 1994 | +1.68pp |
| Q3 1994 | +1.66pp |
| Q4 1994 | +2.14pp |
| Q1 1995 | +0.80pp |
| Q2 1995 | -0.28pp |
| Q3 1995 | +0.01pp |
| Q4 1995 | -0.28pp |
| Q1 1996 | +0.35pp |
| Q2 1996 | +1.03pp |
| Q3 1996 | +0.93pp |
| Q4 1996 | +0.49pp |
| Q1 1997 | +0.83pp |
| Q2 1997 | +0.53pp |
| Q3 1997 | +0.34pp |
| Q4 1997 | +0.22pp |
| Q1 1998 | +0.07pp |
| Q2 1998 | -0.04pp |
| Q3 1998 | -0.84pp |
| Q4 1998 | -0.17pp |
| Q1 1999 | +0.24pp |
| Q2 1999 | +0.86pp |
| Q3 1999 | +0.44pp |
| Q4 1999 | +0.80pp |
| Q1 2000 | +0.68pp |
| Q2 2000 | -0.05pp |
| Q3 2000 | -0.44pp |
| Q4 2000 | -1.05pp |
| Q1 2001 | -0.97pp |
| Q2 2001 | +0.11pp |
| Q3 2001 | +0.05pp |
| Q4 2001 | +1.29pp |
| Q1 2002 | +1.83pp |
| Q2 2002 | +1.24pp |
| Q3 2002 | +0.25pp |
| Q4 2002 | +0.60pp |
| Q1 2003 | +0.32pp |
| Q2 2003 | +0.01pp |
| Q3 2003 | +0.70pp |
| Q4 2003 | +0.93pp |
| Q1 2004 | +0.58pp |
| Q2 2004 | +1.73pp |
| Q3 2004 | +0.92pp |
| Q4 2004 | +0.85pp |
| Q1 2005 | +1.10pp |
| Q2 2005 | +0.60pp |
| Q3 2005 | +0.33pp |
| Q4 2005 | +0.24pp |
| Q1 2006 | +0.14pp |
| Q2 2006 | +0.13pp |
| Q3 2006 | -0.48pp |
| Q4 2006 | -0.57pp |
| Q1 2007 | -0.69pp |
| Q2 2007 | -0.27pp |
| Q3 2007 | -0.93pp |
| Q4 2007 | -1.12pp |
| Q1 2008 | -0.99pp |
| Q2 2008 | +0.77pp |
| Q3 2008 | +0.27pp |
| Q4 2008 | +0.66pp |
| Q1 2009 | +0.75pp |
| Q2 2009 | +0.97pp |
| Q3 2009 | +0.81pp |
| Q4 2009 | +0.75pp |
| Q1 2010 | +0.80pp |
| Q2 2010 | +0.54pp |
| Q3 2010 | +0.29pp |
| Q4 2010 | +0.44pp |
| Q1 2011 | +0.56pp |
| Q2 2011 | +0.32pp |
| Q3 2011 | +0.13pp |
| Q4 2011 | +0.19pp |
| Q1 2012 | +0.21pp |
| Q2 2012 | +0.13pp |
| Q3 2012 | +0.12pp |
| Q4 2012 | +0.10pp |
| Q1 2013 | +0.12pp |
| Q2 2013 | +0.24pp |
| Q3 2013 | +0.32pp |
| Q4 2013 | +0.25pp |
| Q1 2014 | +0.32pp |
| Q2 2014 | +0.35pp |
| Q3 2014 | +0.48pp |
| Q4 2014 | +0.52pp |
| Q1 2015 | +0.53pp |
| Q2 2015 | +0.56pp |
| Q3 2015 | +0.57pp |
| Q4 2015 | +0.74pp |
| Q1 2016 | +0.52pp |
| Q2 2016 | +0.35pp |
| Q3 2016 | +0.37pp |
| Q4 2016 | +0.66pp |
| Q1 2017 | +0.52pp |
| Q2 2017 | +0.30pp |
| Q3 2017 | +0.23pp |
| Q4 2017 | +0.54pp |
| Q1 2018 | +0.77pp |
| Q2 2018 | +0.71pp |
| Q3 2018 | +0.82pp |
| Q4 2018 | +0.41pp |
| Q1 2019 | 0.00pp |
| Q2 2019 | -0.57pp |
| Q3 2019 | -0.39pp |
| Q4 2019 | +0.06pp |
| Q1 2020 | -0.20pp |
| Q2 2020 | +0.11pp |
| Q3 2020 | +0.04pp |
| Q4 2020 | +0.05pp |
| Q1 2021 | +0.08pp |
| Q2 2021 | +0.12pp |
| Q3 2021 | +0.16pp |
| Q4 2021 | +0.60pp |
| Q1 2022 | +1.71pp |
| Q2 2022 | +1.79pp |
| Q3 2022 | +1.30pp |
| Q4 2022 | +0.19pp |
| Q1 2023 | -0.35pp |
| Q2 2023 | -0.44pp |
| Q3 2023 | -0.31pp |
| Q4 2023 | -0.87pp |
| Q1 2024 | -0.74pp |
| Q2 2024 | -0.59pp |
| Q3 2024 | -1.51pp |
| Q4 2024 | -0.25pp |
| Q1 2025 | -0.36pp |
| Q2 2025 | -0.44pp |
| Q3 2025 | -0.65pp |
| Q4 2025 | -0.22pp |
| Q1 2026 | +0.07pp |
| Q2 2026 | +0.48pp |
| Q3 2026 | +0.90pp |
Source: FRED GS2 (GS1 before 1976) and FEDFUNDS; YRI calculations.
Note: Tested as a forward-looking rate trigger and rejected (see method note): busts have followed a negative spread (2007, 2020) as often as a positive one (1987, 2000). Four-quarter change to Q3 2026: +1.55pp, 94th percentile.
Note: Shaded areas are peak-to-trough periods of S&P 500 declines of 20% or more, measured on monthly averages.
AI-Specific Scorecard
Ten conditions that preceded the tech/telecom and housing busts, scored by judgment against today’s AI-cycle data. Each is tagged as a Condition or a Trigger. This layer captures what the economy-wide backtest cannot see, such as vendor-backed chip financing. It is not itself backtestable.
| Condition | Block | Score | Mid-July 2026 | Before the 2000 and 2007 busts | Now |
|---|---|---|---|---|---|
| Valuations | BlockCondition | Score nowHigh | Mid-July 2026High CAPE ~40 (Jan 39.7); S&P 500 forward P/E 20.3 on Jul 17 (FactSet), slightly higher than today. | ThenCAPE peaked at 44.2 in Dec 1999 (multpl; FRBSF gives 43 for 2000, FRBSF 2017); IT sector forward P/E hit a record 48.3 in Mar 2000 (Yardeni QuickTakes). Housing: CAPE was 27.2 in Jan 2007. | NowCAPE 41.0 on Oct 15, 2026 (multpl), within 8% of the 2000 peak. S&P 500 forward P/E 19.0, just below its 10-year average of 19.1 (FactSet, Oct 16), but only because CY2026 earnings growth is forecast at 32.8%. Fed FSR (May 2026): “Asset valuation pressures were elevated” (Fed). IMF GFSR (Oct 2026) names “an artificial intelligence-driven market correction” as one of three rising risks to global financial stability (IMF). CAPE is at a level seen only in 1999–2000. The forward P/E looks ordinary only if the earnings forecasts hold. |
| Market concentration | BlockCondition | Score nowHigh | Mid-July 2026High Top-10 weight 37.9% at the June filing, below the Dec 2025 peak; IT weight for July not verified. | ThenIT reached a record 32.9% of the S&P 500 in Mar 2000 (Yardeni QuickTakes); the 20 largest Nasdaq firms were over one-third of Nasdaq’s value (FRBSF 2001). | NowIT is 39.9% of the S&P 500 as of Sep 30, 2026 (S&P DJI), above the 2000 record. Top-10 stocks 37.9% (Jun 2026), after a 40.7% peak in Dec 2025 (ChartRow, from fund filings; press). IMF: “a small set of firms increasingly drive equity markets.” Concentration already exceeds the 2000 peak on the sector measure. |
| Investment boom | BlockCondition | Score nowHigh | Mid-July 2026High Latest data were Q1 2026: IPE & software investment 4.97% of GDP; Big-5 capex $149bn in Q1. | ThenReal IT investment grew 24% a year 1995–2000, then fell ~11% in 2001; telecom carriers’ capex went from $47bn (1995) to $121bn (2000) to $49bn (2002) (Doms, FRBSF). Information-processing equipment & software investment peaked at 4.3–4.4% of GDP in 2000 (YRI calc from BEA). Housing: residential investment hit 6.7% of GDP in 2005 vs a 4.3% 1990s average (BEA/FRED). | NowIPE & software investment is 5.05% of GDP in Q2 2026, a record, up 19.7% y/y in nominal terms (YRI calc from BEA and GDP). Big-5 hyperscaler capex is 5x its Q1 2023 level. The investment share of GDP is already above the 2000 peak and still accelerating. |
| Revenue validation of the build-out | BlockTrigger | Score nowModerate | Mid-July 2026Moderate Q1 2026 results: Google Cloud +63%, AWS +28%; backlog ~$2.0tn. Anthropic’s loss figures were not yet public (Reuters, Sep 28). | ThenOnly 2–5% of US fiber was lit by 2002 (FDIC); 76% of 1999 IPOs and 81% of 2000 IPOs had negative earnings (Ritter). Housing: Bernanke, Oct 2005: price gains “largely reflect strong economic fundamentals” (testimony). | NowCloud revenue growth of 37–82% y/y and ~$2.35tn of contracted backlog across Microsoft, Oracle, Google Cloud and Amazon. But the labs signing the contracts are loss-making: Anthropic’s draft prospectus shows a 2025 operating loss above $8bn on $4.6bn of revenue (Reuters via InvestmentNews; press). IMF GFSR (Oct 2026) again flags “circular financing arrangements involving equity and debt flows between AI firms” (IMF). Unlike dark fiber, this capacity is being used and paid for. The payers are not yet self-funding. |
| Internal funding capacity | BlockTrigger | Score nowModerate | Mid-July 2026Moderate T4Q to Q1 2026: Amazon capex 102% of OCF, Alphabet 63%, Meta 63%, Microsoft 57%; Big-4 FCF $181bn. | ThenUS telecom companies borrowed more than $320bn in 1999–2001 to fund the build-out (Noam, Columbia); telecom was 25.8% of corporate bond default volume in 2001 (FDIC). | NowTrailing-4Q capex is 107% of operating cash flow at Amazon, 71% at Alphabet and Meta, 63% at Microsoft; combined Big-4 free cash flow fell from $232bn to $147bn. Amazon’s own TTM free cash flow is an outflow of $7.6bn (Q2 2026 release). Bank of England, Jul 2026: “FCF declining” (BoE). Still mostly cash-funded, but the margin of safety is shrinking every quarter. |
| Debt and financing structures | BlockTrigger | Score nowHigh ▲ raised | Mid-July 2026Moderate Debt was mostly plain investment-grade bonds from the strongest balance sheets (Alphabet, Meta, Oracle, Amazon); Fed FSR counted ~$100bn of cloud IG issuance in Q1; IMF had flagged “circular financing” in April. The vendor-backed chip-lease structures came later: Broadcom’s ~$29bn lease backstop and $42bn note facility (10-Q, Sep 10), the $60bn Anthropic syndication (Oct 5), >$50bn for OpenAI and $40bn for SpaceX (Oct 7). | Then1990s: equipment makers’ vendor financing and $320bn of telecom borrowing. 2000s: originate-to-distribute; shadow-bank liabilities of ~$22tn vs ~$14tn for banks by Jun 2007 (NY Fed); Bernanke (2010) on “excessive leverage” and AIG’s unhedged correlated risks (Fed). | NowCloud-related investment-grade issuance neared $100bn in Q1 2026 alone; private credit ~$1.4tn (Fed FSR). BoE: hyperscaler issuance in H1 2026 exceeds all of 2025, with “increasing dependence on accommodative future refinancing conditions” (BoE). Software is ~1/5 of direct lending (IMF). Vendor-backed and off-balance-sheet deals: $60bn for Anthropic’s chip leases, $42bn Broadcom-backed (FT via Dealroom); Broadcom arranging >$50bn for OpenAI (WSJ via Yahoo); SpaceX seeking $40bn for Nvidia chips (FT via Yahoo); Meta–Blue Owl Hyperion JV ~$27bn (Meta). IMF GFSR (Oct 2026): a review of 14 AI transactions since 2023 found about $300bn of equity and $650bn of debt exchanged for $1.4tn of future revenue commitments; “hyperscalers are borrowing more” (IMF). Press (Oct 16–17): the $60bn Anthropic package includes an $18bn subordinated tranche, with Blackstone taking ~$9bn at a 10% coupon, about 3pp above BB-rated borrowers (TokenPost; press). The structures that turned past booms into busts — vendor financing, SPVs, private credit, circular deals — are all present and growing fast. Hyperscaler balance sheets themselves remain strong. |
| Interest-rate tightening | BlockTrigger | Score nowModerate ▲ raised | Mid-July 2026Low 2-yr Treasury yield 4.11% in June, up 61bp from its Dec 2025 low of 3.50% (a 45th-percentile rise on the mechanical measure); fed funds 3.63%, on hold after a year of cuts, with no hike until Sep 16. | ThenThe 2-yr Treasury yield rose from 4.51% (Dec 1998) to 6.53% (Mar 2000), +202bp, as fed funds went from 4.6% to 6.5%; the Nasdaq peaked in Mar 2000. It rose from 1.58% (Mar 2004) to 5.12% (Jun 2006), +354bp, as fed funds went from 1.0% to 5.25%; house prices peaked Jul 2006 (FRED GS2, FEDFUNDS). | NowThe 2-yr yield has risen from 3.50% (Dec 2025) to 4.65% (Sep 2026), +115bp, and is up 108bp y/y, a 79th-percentile one-year move; it was 4.77% on Oct 15 (FRED GS2, DGS2). The FOMC followed the market, raising the target range 25bp to 3.75–4.0% on Sep 16, 2026, 12–0, after cutting from 5.33% to 3.63% (FOMC). Tightening is underway and the market is leading the Fed. So far the 2-yr move is a third to a half of the 1999–2000 and 2004–06 campaigns. |
| Household participation and leverage | BlockCondition | Score nowHigh | Mid-July 2026High Household equity share 44.7% (Q1 2026, then a record); margin debt $1.50tn in June, the high of the period. | ThenEquities were a record 38.7% of household financial assets in Q1 2000, up from 21.2% in 1995; 31.5% in Q4 2007 (Fed Z.1 via FRED). Margin debt peaked in Mar 2000. | NowEquities are 48.2% of household financial assets in Q2 2026, a record (Fed Z.1 via FRED). FINRA margin debt $1.48tn in Sep 2026, up 31% y/y and just below the $1.50tn June high (FINRA); ~4.5% of GDP (YRI calc). Households are more exposed to equities than at either prior peak, and margin borrowing is rising fast. |
| Speculative issuance | BlockCondition | Score nowModerate | Mid-July 2026Moderate OpenAI and Anthropic had filed confidentially in June; 2025 Ritter data unchanged. | Then476 IPOs in 1999 with a 71% average first-day return; 380 in 2000 at 56% (Ritter). Only 14% of 1999 tech IPOs were profitable. | Now90 IPOs in 2025 with a 29% average first-day return, up from 73 and 15% in 2024; 26% of tech IPOs profitable (Ritter). OpenAI (last valued $852bn) and Anthropic filed confidentially for IPOs in Jun 2026 (TechCrunch; press). Greenwood–Shleifer–You find issuance is a predictive feature of crashes (NBER). Volumes are a fraction of 1999, but first-day pops are rising and the largest IPOs in history are queued. |
| Idle-capacity signals | BlockTrigger | Score nowLow | Mid-July 2026Low Sightline/Axios delay report (Feb) already published; backlogs contracted. | ThenCommunications-equipment orders fell from $13.3bn (Jun 2000) to $3.6bn (Sep 2001); capacity utilization 55% vs 87% (FDIC). Housing starts fell from 2.27mn (Jan 2006) to 0.48mn (Apr 2009). | NowData-center construction +73% y/y to a record $85bn saar. Up to half of 2026 projects may face delays and 9 cancellations were tracked (Sightline via Axios; press), but the binding constraint is power and permitting, not demand; backlogs are contracted. No evidence yet of capacity sitting idle. This is the indicator that turned first in 2000–01. |
Then and Now
Matched readings at the 2000 and 2007 peaks versus the latest.
| Measure | Tech/telecom peak, 2000 | Housing peak, 2006–07 | Now, 2026 |
|---|---|---|---|
| Shiller CAPE | 44.2 (Dec 1999) | 27.2 (Jan 2007) | 41.0 (Oct 15, 2026) |
| Largest sector, % of S&P 500 | IT 32.9% (Mar 2000) | Financials ~20–22% (press, conflicting) | IT 39.9% (Sep 30, 2026) |
| Boom-sector investment, % of GDP | IPE & software 4.3–4.4% (2000) | Residential 6.7% (2005) | IPE & software 5.05% (Q2 2026) |
| Equities, % of household financial assets | 38.7% (Q1 2000) | 31.5% (Q4 2007) | 48.2% (Q2 2026) |
| 2-yr Treasury yield at the peak | 6.53% (Mar 2000) after +202bp from Dec 1998 | 5.12% (Jun 2006) after +354bp from Mar 2004 | 4.65% (Sep 2026) after +115bp from Dec 2025; 4.77% on Oct 15 |
| Fed funds at the peak | 6.5% after +175bp | 5.25% after +425bp | 3.75–4.0% after +25bp (Sep 2026) |
| IPO market | 476 IPOs, +71% first day (1999) | — | 90 IPOs, +29% (2025); OpenAI/Anthropic filed |
| Debt build-up | Telecom borrowed >$320bn (1999–2001) | Mortgage debt 2.2x (2000–07); shadow banking ~$22tn | Cloud IG issuance ~$100bn in Q1 2026; H1 > all of 2025 |
| Capacity use | 2–5% of fiber lit (2002) | Starts 2.27mn → 0.48mn | Backlog ~$2.35tn; power-constrained |
| What followed | Nasdaq −78% (Mar 2000–Oct 2002); S&P 500 −15% by May 2001 and further after | S&P 500 −57% (Oct 2007–Mar 2009); home prices −27% | — |
Charts
The series behind the scores.
Shiller CAPE at Market Peaks
(cyclically adjusted P/E, S&P 500)
Valuations are within 8% of the 2000 peak
- Shiller CAPE
| Period | Shiller CAPE |
|---|---|
| Dec 1999 | 44.19 |
| Jan 2000 | 43.77 |
| Jan 2007 | 27.21 |
| Jan 2025 | 37.14 |
| Jan 2026 | 39.65 |
| Oct 15, 2026 | 41.01 |
Source: multpl.com (Shiller data) (secondary); FRBSF Economic Letter 2017-33 gives 43 for 2000.
Note: Long-run mean since 1871: 17.4 (multpl).
Fed Funds Rate: Three Cycles
(effective federal funds rate, percent; years from start of each window)
Both prior busts followed a sustained Fed tightening campaign; this cycle has just turned (reference: the dial uses the 2-yr yield, next chart)
- 1998–2002 (peak 6.5%, May 2000)
- 2004–2008 (peak 5.25%, Jun 2006)
- 2022–2026 (hiked to 3.75–4% Sep 2026)
| Date | 1998–2002 (peak 6.5%, May 2000) | 2004–2008 (peak 5.25%, Jun 2006) | 2022–2026 (hiked to 3.75–4% Sep 2026) |
|---|---|---|---|
| Jan 1998 | 5.56% | — | — |
| Feb 1998 | 5.51% | — | — |
| Mar 1998 | 5.49% | — | — |
| Apr 1998 | 5.45% | — | — |
| May 1998 | 5.49% | — | — |
| Jun 1998 | 5.56% | — | — |
| Jul 1998 | 5.54% | — | — |
| Aug 1998 | 5.55% | — | — |
| Sep 1998 | 5.51% | — | — |
| Oct 1998 | 5.07% | — | — |
| Nov 1998 | 4.83% | — | — |
| Dec 1998 | 4.68% | — | — |
| Jan 1999 | 4.63% | — | — |
| Feb 1999 | 4.76% | — | — |
| Mar 1999 | 4.81% | — | — |
| Apr 1999 | 4.74% | — | — |
| May 1999 | 4.74% | — | — |
| Jun 1999 | 4.76% | — | — |
| Jul 1999 | 4.99% | — | — |
| Aug 1999 | 5.07% | — | — |
| Sep 1999 | 5.22% | — | — |
| Oct 1999 | 5.20% | — | — |
| Nov 1999 | 5.42% | — | — |
| Dec 1999 | 5.30% | — | — |
| Jan 2000 | 5.45% | — | — |
| Feb 2000 | 5.73% | — | — |
| Mar 2000 | 5.85% | — | — |
| Apr 2000 | 6.02% | — | — |
| May 2000 | 6.27% | — | — |
| Jun 2000 | 6.53% | — | — |
| Jul 2000 | 6.54% | — | — |
| Aug 2000 | 6.50% | — | — |
| Sep 2000 | 6.52% | — | — |
| Oct 2000 | 6.51% | — | — |
| Nov 2000 | 6.51% | — | — |
| Dec 2000 | 6.40% | — | — |
| Jan 2001 | 5.98% | — | — |
| Feb 2001 | 5.49% | — | — |
| Mar 2001 | 5.31% | — | — |
| Apr 2001 | 4.80% | — | — |
| May 2001 | 4.21% | — | — |
| Jun 2001 | 3.97% | — | — |
| Jul 2001 | 3.77% | — | — |
| Aug 2001 | 3.65% | — | — |
| Sep 2001 | 3.07% | — | — |
| Oct 2001 | 2.49% | — | — |
| Nov 2001 | 2.09% | — | — |
| Dec 2001 | 1.82% | — | — |
| Jan 2002 | 1.73% | — | — |
| Feb 2002 | 1.74% | — | — |
| Mar 2002 | 1.73% | — | — |
| Apr 2002 | 1.75% | — | — |
| May 2002 | 1.75% | — | — |
| Jun 2002 | 1.75% | — | — |
| Jul 2002 | 1.73% | — | — |
| Aug 2002 | 1.74% | — | — |
| Sep 2002 | 1.75% | — | — |
| Oct 2002 | 1.75% | — | — |
| Nov 2002 | 1.34% | — | — |
| Dec 2002 | 1.24% | — | — |
| Jan 2004 | — | 1.00% | — |
| Feb 2004 | — | 1.01% | — |
| Mar 2004 | — | 1.00% | — |
| Apr 2004 | — | 1.00% | — |
| May 2004 | — | 1.00% | — |
| Jun 2004 | — | 1.03% | — |
| Jul 2004 | — | 1.26% | — |
| Aug 2004 | — | 1.43% | — |
| Sep 2004 | — | 1.61% | — |
| Oct 2004 | — | 1.76% | — |
| Nov 2004 | — | 1.93% | — |
| Dec 2004 | — | 2.16% | — |
| Jan 2005 | — | 2.28% | — |
| Feb 2005 | — | 2.50% | — |
| Mar 2005 | — | 2.63% | — |
| Apr 2005 | — | 2.79% | — |
| May 2005 | — | 3.00% | — |
| Jun 2005 | — | 3.04% | — |
| Jul 2005 | — | 3.26% | — |
| Aug 2005 | — | 3.50% | — |
| Sep 2005 | — | 3.62% | — |
| Oct 2005 | — | 3.78% | — |
| Nov 2005 | — | 4.00% | — |
| Dec 2005 | — | 4.16% | — |
| Jan 2006 | — | 4.29% | — |
| Feb 2006 | — | 4.49% | — |
| Mar 2006 | — | 4.59% | — |
| Apr 2006 | — | 4.79% | — |
| May 2006 | — | 4.94% | — |
| Jun 2006 | — | 4.99% | — |
| Jul 2006 | — | 5.24% | — |
| Aug 2006 | — | 5.25% | — |
| Sep 2006 | — | 5.25% | — |
| Oct 2006 | — | 5.25% | — |
| Nov 2006 | — | 5.25% | — |
| Dec 2006 | — | 5.24% | — |
| Jan 2007 | — | 5.25% | — |
| Feb 2007 | — | 5.26% | — |
| Mar 2007 | — | 5.26% | — |
| Apr 2007 | — | 5.25% | — |
| May 2007 | — | 5.25% | — |
| Jun 2007 | — | 5.25% | — |
| Jul 2007 | — | 5.26% | — |
| Aug 2007 | — | 5.02% | — |
| Sep 2007 | — | 4.94% | — |
| Oct 2007 | — | 4.76% | — |
| Nov 2007 | — | 4.49% | — |
| Dec 2007 | — | 4.24% | — |
| Jan 2008 | — | 3.94% | — |
| Feb 2008 | — | 2.98% | — |
| Mar 2008 | — | 2.61% | — |
| Apr 2008 | — | 2.28% | — |
| May 2008 | — | 1.98% | — |
| Jun 2008 | — | 2.00% | — |
| Jul 2008 | — | 2.01% | — |
| Aug 2008 | — | 2.00% | — |
| Sep 2008 | — | 1.81% | — |
| Oct 2008 | — | 0.97% | — |
| Nov 2008 | — | 0.39% | — |
| Dec 2008 | — | 0.16% | — |
| Jan 2022 | — | — | 0.08% |
| Feb 2022 | — | — | 0.08% |
| Mar 2022 | — | — | 0.20% |
| Apr 2022 | — | — | 0.33% |
| May 2022 | — | — | 0.77% |
| Jun 2022 | — | — | 1.21% |
| Jul 2022 | — | — | 1.68% |
| Aug 2022 | — | — | 2.33% |
| Sep 2022 | — | — | 2.56% |
| Oct 2022 | — | — | 3.08% |
| Nov 2022 | — | — | 3.78% |
| Dec 2022 | — | — | 4.10% |
| Jan 2023 | — | — | 4.33% |
| Feb 2023 | — | — | 4.57% |
| Mar 2023 | — | — | 4.65% |
| Apr 2023 | — | — | 4.83% |
| May 2023 | — | — | 5.06% |
| Jun 2023 | — | — | 5.08% |
| Jul 2023 | — | — | 5.12% |
| Aug 2023 | — | — | 5.33% |
| Sep 2023 | — | — | 5.33% |
| Oct 2023 | — | — | 5.33% |
| Nov 2023 | — | — | 5.33% |
| Dec 2023 | — | — | 5.33% |
| Jan 2024 | — | — | 5.33% |
| Feb 2024 | — | — | 5.33% |
| Mar 2024 | — | — | 5.33% |
| Apr 2024 | — | — | 5.33% |
| May 2024 | — | — | 5.33% |
| Jun 2024 | — | — | 5.33% |
| Jul 2024 | — | — | 5.33% |
| Aug 2024 | — | — | 5.33% |
| Sep 2024 | — | — | 5.13% |
| Oct 2024 | — | — | 4.83% |
| Nov 2024 | — | — | 4.64% |
| Dec 2024 | — | — | 4.48% |
| Jan 2025 | — | — | 4.33% |
| Feb 2025 | — | — | 4.33% |
| Mar 2025 | — | — | 4.33% |
| Apr 2025 | — | — | 4.33% |
| May 2025 | — | — | 4.33% |
| Jun 2025 | — | — | 4.33% |
| Jul 2025 | — | — | 4.33% |
| Aug 2025 | — | — | 4.33% |
| Sep 2025 | — | — | 4.22% |
| Oct 2025 | — | — | 4.09% |
| Nov 2025 | — | — | 3.88% |
| Dec 2025 | — | — | 3.72% |
| Jan 2026 | — | — | 3.64% |
| Feb 2026 | — | — | 3.64% |
| Mar 2026 | — | — | 3.64% |
| Apr 2026 | — | — | 3.64% |
| May 2026 | — | — | 3.63% |
| Jun 2026 | — | — | 3.63% |
| Jul 2026 | — | — | 3.63% |
| Aug 2026 | — | — | 3.63% |
| Sep 2026 | — | — | 3.75% |
Source: FRED FEDFUNDS; FOMC statement, Sep 16 2026.
2-Year Treasury Yield: Three Tightening Cycles
(percent, quarter-end months; years from the cycle low in the 2-yr yield)
The market rate trigger: the 2-yr yield is up 115bp from its low in three quarters, against +202bp and +354bp before the 2000 and 2006 peaks
- Low Dec 1998 → 6.53% Mar 2000 (+202bp)
- Low Mar 2004 → 5.12% Jun 2006 (+354bp)
- Low Dec 2025 → 4.65% Sep 2026 (+115bp)
| Date | Low Dec 1998 → 6.53% Mar 2000 (+202bp) | Low Mar 2004 → 5.12% Jun 2006 (+354bp) | Low Dec 2025 → 4.65% Sep 2026 (+115bp) |
|---|---|---|---|
| Dec 1996 | 5.78% | — | — |
| Mar 1997 | 6.22% | — | — |
| Jun 1997 | 6.09% | — | — |
| Sep 1997 | 5.88% | — | — |
| Dec 1997 | 5.72% | — | — |
| Mar 1998 | 5.56% | — | — |
| Jun 1998 | 5.52% | — | — |
| Sep 1998 | 4.67% | — | — |
| Dec 1998 | 4.51% | — | — |
| Mar 1999 | 5.05% | — | — |
| Jun 1999 | 5.62% | — | — |
| Sep 1999 | 5.66% | — | — |
| Dec 1999 | 6.10% | — | — |
| Mar 2000 | 6.53% | — | — |
| Jun 2000 | 6.48% | — | — |
| Sep 2000 | 6.08% | — | — |
| Dec 2000 | 5.35% | — | — |
| Mar 2001 | 4.34% | — | — |
| Jun 2001 | 4.08% | — | — |
| Sep 2001 | 3.12% | — | — |
| Dec 2001 | 3.11% | — | — |
| Mar 2002 | 3.56% | 3.56% | — |
| Jun 2002 | 2.99% | 2.99% | — |
| Sep 2002 | 2.00% | 2.00% | — |
| Dec 2002 | 1.84% | 1.84% | — |
| Mar 2003 | — | 1.57% | — |
| Jun 2003 | — | 1.23% | — |
| Sep 2003 | — | 1.71% | — |
| Dec 2003 | — | 1.91% | — |
| Mar 2004 | — | 1.58% | — |
| Jun 2004 | — | 2.76% | — |
| Sep 2004 | — | 2.53% | — |
| Dec 2004 | — | 3.01% | — |
| Mar 2005 | — | 3.73% | — |
| Jun 2005 | — | 3.64% | — |
| Sep 2005 | — | 3.95% | — |
| Dec 2005 | — | 4.40% | — |
| Mar 2006 | — | 4.73% | — |
| Jun 2006 | — | 5.12% | — |
| Sep 2006 | — | 4.77% | — |
| Dec 2006 | — | 4.67% | — |
| Mar 2007 | — | 4.57% | — |
| Jun 2007 | — | 4.98% | — |
| Sep 2007 | — | 4.01% | — |
| Dec 2007 | — | 3.12% | — |
| Mar 2008 | — | 1.62% | — |
| Dec 2023 | — | — | 4.46% |
| Mar 2024 | — | — | 4.59% |
| Jun 2024 | — | — | 4.74% |
| Sep 2024 | — | — | 3.62% |
| Dec 2024 | — | — | 4.23% |
| Mar 2025 | — | — | 3.97% |
| Jun 2025 | — | — | 3.89% |
| Sep 2025 | — | — | 3.57% |
| Dec 2025 | — | — | 3.50% |
| Mar 2026 | — | — | 3.71% |
| Jun 2026 | — | — | 4.11% |
| Sep 2026 | — | — | 4.65% |
Source: FRED GS2; YRI alignment. The mechanical Triggers dial scores this rise against its trailing 8-quarter low.
Note: Windows run from eight quarters before each low to four years after. Lows are the lowest quarter-end month: Dec 1998, Mar 2004, Dec 2025.
FINRA Margin Debt
(debit balances in customers’ securities margin accounts, billion dollars)
Margin debt is up 31% y/y, just below June’s high
- FINRA margin debt
| Period | FINRA margin debt |
|---|---|
| Sep 2025 | $1,126bn |
| Oct 2025 | $1,184bn |
| Nov 2025 | $1,214bn |
| Dec 2025 | $1,226bn |
| Jan 2026 | $1,279bn |
| Feb 2026 | $1,253bn |
| Mar 2026 | $1,221bn |
| Apr 2026 | $1,304bn |
| May 2026 | $1,416bn |
| Jun 2026 | $1,502bn |
| Jul 2026 | $1,417bn |
| Aug 2026 | $1,454bn |
| Sep 2026 | $1,480bn |
Source: FINRA margin statistics.
Note: FINRA’s public table covers the latest 13 months; the long-run record was not verified.
Hyperscaler Capex as % of Operating Cash Flow
(percent, trailing four quarters)
Capex is absorbing a rising share of cash flow
- Microsoft (Q2 2026 = 63.4%)
- Alphabet (Q2 2026 = 71.3%)
- Amazon (Q2 2026 = 107.2%)
- Meta (Q2 2026 = 70.9%)
| Date | Microsoft | Alphabet | Amazon | Meta |
|---|---|---|---|---|
| Q4 2023 | 34.3% | 31.7% | 62.1% | 39.5% |
| Q1 2024 | 35.9% | 35.5% | 53.9% | 36.3% |
| Q2 2024 | 37.5% | 42.1% | 55.2% | 38.1% |
| Q3 2024 | 40.5% | 46.9% | 61.9% | 39.0% |
| Q4 2024 | 44.2% | 41.9% | 71.6% | 42.9% |
| Q1 2025 | 46.9% | 43.5% | 81.7% | 48.1% |
| Q2 2025 | 47.4% | 50.1% | 88.9% | 53.5% |
| Q3 2025 | 46.9% | 51.4% | 91.9% | 60.3% |
| Q4 2025 | 51.8% | 55.5% | 94.5% | 62.4% |
| Q1 2026 | 57.1% | 63.0% | 101.7% | 63.2% |
| Q2 2026 | 63.4% | 71.3% | 107.2% | 70.9% |
Source: Company 8-K filings; YRI calculations.
Big Four: Free Cash Flow
(operating cash flow less capex, MSFT+GOOGL+AMZN+META, billion dollars, T4Q)
Combined free cash flow is falling as capex accelerates
- Combined FCF (T4Q)
| Date | Combined FCF (T4Q) |
|---|---|
| Q4 2023 | $212.2bn |
| Q1 2024 | $234.0bn |
| Q2 2024 | $231.8bn |
| Q3 2024 | $221.9bn |
| Q4 2024 | $227.8bn |
| Q1 2025 | $215.0bn |
| Q2 2025 | $199.4bn |
| Q3 2025 | $204.8bn |
| Q4 2025 | $202.0bn |
| Q1 2026 | $180.5bn |
| Q2 2026 | $146.5bn |
Source: Company 8-K filings; YRI calculations.
Cloud Revenue Growth
(yearly percent change)
Demand is real: cloud growth accelerated through 2026
- AWS (Q2 2026 = 36.8%)
- Azure (reported) (Q2 2026 = 43%)
- Google Cloud (Q2 2026 = 81.8%)
- Oracle Cloud (Q2 2026 = 47.1%)
| Date | AWS | Azure (reported) | Google Cloud | Oracle Cloud |
|---|---|---|---|---|
| Q1 2024 | 17.2% | 31% | 28.4% | 24.7% |
| Q2 2024 | 18.7% | 29% | 28.8% | 19.7% |
| Q3 2024 | 19.1% | 33% | 35% | 21.3% |
| Q4 2024 | 18.9% | 31% | 30.1% | 24.3% |
| Q1 2025 | 16.9% | 33% | 28.1% | 22.9% |
| Q2 2025 | 17.5% | 39% | 31.7% | 26.8% |
| Q3 2025 | 20.2% | 40% | 33.5% | 27.8% |
| Q4 2025 | 23.6% | 39% | 47.8% | 34.4% |
| Q1 2026 | 28.4% | 40% | 63.4% | 43.5% |
| Q2 2026 | 36.8% | 43% | 81.8% | 47.1% |
Source: Company 8-K filings; YRI calculations. Azure growth as reported by Microsoft.
Residential Investment Share of GDP
(percent of nominal GDP)
The housing analog: investment share peaked at 6.7% in 2005, then halved
- Residential investment, % of GDP
- Recession
| Date | Residential investment, % of GDP |
|---|---|
| Q1 1990 | 4.4% |
| Q2 1990 | 4.1% |
| Q3 1990 | 3.9% |
| Q4 1990 | 3.7% |
| Q1 1991 | 3.5% |
| Q2 1991 | 3.6% |
| Q3 1991 | 3.6% |
| Q4 1991 | 3.7% |
| Q1 1992 | 3.8% |
| Q2 1992 | 3.9% |
| Q3 1992 | 3.9% |
| Q4 1992 | 4.0% |
| Q1 1993 | 4.0% |
| Q2 1993 | 4.1% |
| Q3 1993 | 4.2% |
| Q4 1993 | 4.4% |
| Q1 1994 | 4.4% |
| Q2 1994 | 4.5% |
| Q3 1994 | 4.5% |
| Q4 1994 | 4.4% |
| Q1 1995 | 4.3% |
| Q2 1995 | 4.1% |
| Q3 1995 | 4.2% |
| Q4 1995 | 4.3% |
| Q1 1996 | 4.4% |
| Q2 1996 | 4.5% |
| Q3 1996 | 4.5% |
| Q4 1996 | 4.4% |
| Q1 1997 | 4.4% |
| Q2 1997 | 4.4% |
| Q3 1997 | 4.4% |
| Q4 1997 | 4.4% |
| Q1 1998 | 4.5% |
| Q2 1998 | 4.6% |
| Q3 1998 | 4.7% |
| Q4 1998 | 4.8% |
| Q1 1999 | 4.8% |
| Q2 1999 | 4.8% |
| Q3 1999 | 4.8% |
| Q4 1999 | 4.8% |
| Q1 2000 | 4.8% |
| Q2 2000 | 4.7% |
| Q3 2000 | 4.7% |
| Q4 2000 | 4.7% |
| Q1 2001 | 4.7% |
| Q2 2001 | 4.8% |
| Q3 2001 | 4.9% |
| Q4 2001 | 4.9% |
| Q1 2002 | 5.0% |
| Q2 2002 | 5.1% |
| Q3 2002 | 5.1% |
| Q4 2002 | 5.2% |
| Q1 2003 | 5.4% |
| Q2 2003 | 5.4% |
| Q3 2003 | 5.6% |
| Q4 2003 | 5.8% |
| Q1 2004 | 5.9% |
| Q2 2004 | 6.1% |
| Q3 2004 | 6.2% |
| Q4 2004 | 6.3% |
| Q1 2005 | 6.4% |
| Q2 2005 | 6.5% |
| Q3 2005 | 6.7% |
| Q4 2005 | 6.7% |
| Q1 2006 | 6.6% |
| Q2 2006 | 6.2% |
| Q3 2006 | 5.9% |
| Q4 2006 | 5.6% |
| Q1 2007 | 5.3% |
| Q2 2007 | 5.0% |
| Q3 2007 | 4.6% |
| Q4 2007 | 4.2% |
| Q1 2008 | 3.9% |
| Q2 2008 | 3.6% |
| Q3 2008 | 3.4% |
| Q4 2008 | 3.1% |
| Q1 2009 | 2.8% |
| Q2 2009 | 2.6% |
| Q3 2009 | 2.7% |
| Q4 2009 | 2.7% |
| Q1 2010 | 2.6% |
| Q2 2010 | 2.6% |
| Q3 2010 | 2.4% |
| Q4 2010 | 2.4% |
| Q1 2011 | 2.4% |
| Q2 2011 | 2.4% |
| Q3 2011 | 2.4% |
| Q4 2011 | 2.5% |
| Q1 2012 | 2.6% |
| Q2 2012 | 2.6% |
| Q3 2012 | 2.7% |
| Q4 2012 | 2.8% |
Source: BEA via FRED A011RE1Q156NBEA.
Note: Shaded areas denote recessions according to the National Bureau of Economic Research.
US IPO Volume
(number of IPOs per year; first-day return in note)
Issuance is a fraction of 1999, though the pops are back
- US IPOs (count)
| Period | US IPOs (count) |
|---|---|
| 1999 | 476 |
| 2000 | 380 |
| 2024 | 73 |
| 2025 | 90 |
Source: Jay Ritter, IPO Statistics (Oct 2026).
Note: Mean first-day returns: 1999 71.2%; 2000 56.3%; 2024 15.3%; 2025 29.3%. Tech IPOs profitable: 14% (1999), 26% (2025).
What History Says
Findings from the academic and central-bank literature on which conditions matter.
Jordà, Schularick & Taylor (NBER 2015)
“What makes some bubbles more dangerous than others is credit.” Credit-financed bubbles are followed by deeper recessions and slower recoveries (17 countries, 140 years). Source
Greenwood, Shleifer & You (JFE 2019)
A 100% two-year industry price run-up raises the probability of a crash from a 14% baseline to 53%; issuance, volatility and acceleration are predictive. (Current run-up not scored: no verified series.) Source
Borio & Lowe (BIS 2002)
Sustained rapid credit growth combined with large asset-price increases raises the probability of financial instability; low inflation can mask the build-up. Source
Bernanke (2010)
The housing boom was “a feedback loop between optimism regarding house prices and developments in the mortgage market”; excessive leverage, rating conflicts and originate-to-distribute were the vulnerabilities. Source
Notes
Method
The page has two layers. The mechanical layer (dials and backtest) uses six economy-wide indicators with history to the 1950s and is scored by rule. The AI-specific layer is the ten-condition scorecard: ten conditions that preceded the 1999–2000 tech/telecom bust and the 2006–07 housing bust are each scored Low (0), Moderate (1) or High (2) against the latest verified data. Its items are tagged Condition or Trigger and averaged within each block rather than into one number, because the backtest shows an equal-weight composite in the 67–75 range was no more likely to precede a bust than any other quarter. Each score links to its evidence; the reader can disagree with any line and re-score it.
Change since mid-July 2026
The “Mid-July 2026” column re-scores each condition using only what was published by July 14, 2026: Q1 2026 company results, the Q1 Z.1 release, the Fed’s May Financial Stability Report, the June IPO filings, June margin debt, a 2-yr Treasury yield of 4.11% and a Fed funds rate of 3.63%. Two conditions were raised since then: Debt and financing structures (Moderate → High), on the vendor-backed chip-lease financings disclosed from September 10 onward, and Interest-rate tightening (Low → Moderate), on the further 54bp rise in the 2-yr yield to 4.65% and the September 16 rate hike. Within unchanged scores, the readings also worsened: Big-4 free cash flow fell from $181bn to $147bn, Amazon’s capex rose from 102% to 107% of operating cash flow, and the household equity share jumped from 44.7% to 48.2%. One condition improved without changing score: cloud growth accelerated and backlog rose from about $2.0tn to $2.35tn. The Debt line is a boundary call.
What this meter does not say
A high reading means the preconditions of past bubbles are present, not that a bust is imminent. In both prior episodes the trigger was a sustained rise in the cost of capital followed by evidence that capacity was outrunning demand. Those two indicators (Interest-rate tightening, Idle-capacity signals) are the ones to watch.
Credit spreads are not factored in. If bond investors demand more compensation for risk, wider spreads would raise borrowing costs and could moderate the credit risk the triggers dial tracks.
Not scored (could not be verified)
- Two-year net-of-market price run-up for the AI sector (Greenwood–Shleifer–You test).
- S&P 500 IT sector forward P/E (FactSet publishes it as a chart image; below 22.8 as of Oct 9).
- Dollar value of 2000 margin debt; vendor-financing amounts at Lucent/Nortel/Cisco; subprime share of 2005–06 originations (FCIC figures not retrievable).
Sources
Federal Reserve Financial Stability Report (May 2026); IMF Global Financial Stability Report (Apr and Oct 2026); Bank of England FSR remarks (Jul 2026); Fed Z.1 and BEA via FRED; FINRA; S&P Dow Jones Indices; FactSet; Jay Ritter IPO statistics; FRBSF Economic Letters (2001, 2017) and Doms (2004); FDIC Bank Trends (2002); NY Fed Staff Report 458; NBER working papers; company filings; press reports where flagged. Prepared with AI assistance; figures were checked against the cited documents. Press-sourced items are marked.