Is anything warning of a market top?
Every warning signal we hold, scored one way: when this signal was at its most extreme readings, how often did the S&P 500 fall 10%, 15% or 20% within the next 3, 6 or 12 months, compared with how often that happens anyway. Then the part that matters if you would act on it: how early the warning came, how much higher the market went first, and whether a short opened on the warning day ever got 15% into the money.
Where are we now, and would these warnings have made money?
Pick a reading and a rule. Green grows $100 by following the rule; grey is $100 simply held in the S&P 500 over the same months. Along the bottom, red marks months the reading was in its top zone and green months in its bottom zone. Pink bands are stressed markets.
| Reading (click to chart) | Latest | Where it sits | Top readings: S&P fell 10%+ within a year | Bottom readings: S&P up 10%+ a year later | Stay invested, avoid tops | Short tops, buy bottoms |
|---|
Re-run 2026-09-14 without survivorship bias: the stock-market readings now include every stock in each year's 1,500 most traded, including companies that were later delisted, instead of only today's survivors.
Today
Bars: how many signals were extreme at each month-end. Line: the S&P 500's distance below its 1-year high. Shaded: the year before each 15% fall and the fall itself. The question the picture answers: do tall bars sit in the shaded stretches, or everywhere?
The combined reading
Count how many of the 10 core signals are at extreme readings. Then look at what followed months with that count.
| Signals extreme at once | Months | 10% fall within 6 mo | 10% fall within 12 mo | 15% fall within 12 mo | 20% fall within 12 mo | S&P lower a year later | Avg return, next 12 mo |
|---|---|---|---|---|---|---|---|
| Any month (the bar to beat) | 236 | 33% | 56% | 40% | 17% | 17% | 12.3% |
| 0 ◀ today | 21 | 19% | 24% | 0% | 0% | 0% | 18.5% |
| 1 | 37 | 27% | 59% | 30% | 3% | 5% | 17.5% |
| 2 | 44 | 32% | 45% | 39% | 16% | 7% | 17.3% |
| 3 | 43 | 26% | 42% | 28% | 16% | 7% | 17.8% |
| 4 or more | 55 | 42% | 80% | 56% | 15% | 18% | 10.2% |
Red = at least 15 points above the bar; green = at least 15 points below it. Straight answer: no count, and no single signal below, reliably predicted a 15% fall. The most extreme readings roughly double the odds over a bar that is already high, and they arrive early.
What happened at each of the 7 falls
| Fall | Peak | Depth | Warned (3+ extreme)? | First warning | Signals then | Months before the peak | Months before −15% | Market rose further by | Fell 15% below the warning-day price? |
|---|---|---|---|---|---|---|---|---|---|
| 2008-01 | 2007-10-09 | -51.5% | not scorable (needs 3 years of history first) | — | — | — | — | — | — |
| 2009-08 | 2009-08-03 | -22.4% | no (at most 1 extreme) | — | — | — | — | — | — |
| 2011-08 | 2011-04-29 | -18.6% | yes | 2009-12-31 | S&P 12-month return, M2 money growth, year over year, Share of stocks within 5% of their own high, Sahm rule: unemployment vs its 12-mo low | 15.9 | 19.2 | +25.4% | no |
| 2018-12 | 2018-09-20 | -19.3% | yes | 2017-01-31 | Shiller CAPE, Credit spread (tight = complacent), VIX (complacency = low) | 19.7 | 22.7 | +32.7% | no |
| 2020-03 | 2020-02-19 | -33.7% | yes | 2019-11-30 | Shiller CAPE, Yield curve 10y−3m (inverted = negative), VIX (complacency = low) | 2.6 | 3.2 | +8.2% | yes |
| 2022-05 | 2022-01-03 | -24.5% | yes | 2020-06-30 | Shiller CAPE, Yield curve 10y−3m (inverted = negative), Sahm rule: unemployment vs its 12-mo low | 18.1 | 22.3 | +58.3% | no |
| 2025-04 | 2025-02-19 | -18.8% | yes | 2023-06-30 | Credit spread (tight = complacent), Yield curve 10y−3m (inverted = negative), M2 money growth, year over year | 19.5 | 21 | +41.1% | no |
Each signal on its own
For each signal: in the months when it was at its most extreme 20% of readings, how often a fall followed, next to how often one follows any month. "Nothing happened" is the false-alarm rate: the share of extreme months with no 10% fall in the next 6 months.
| Signal | Since | 10% fall within 6 mo ? | 15% fall within 12 mo | 20% fall within 12 mo | Nothing happened (6 mo) | Falls warned | Warning came (months before −15%) | Market rose further | Short got 15% in the money | Today |
|---|---|---|---|---|---|---|---|---|---|---|
| FINRA short-volume share (2021→ only) FINRA | 2021-09 | 80% vs 33% | 80% vs 40% | 50% vs 17% | 20% | 2 of 7 | 7.2 | +11.7% | 2 of 2 | normal |
| Oil price, 6-month change FRED WTI | 2006-01 | 42% vs 33% | 69% vs 40% | 29% vs 17% | 58% | 6 of 7 | 8.3 | +11% | 2 of 6 | normal |
| Shiller CAPE Shiller | 2006-01 | 46% vs 33% | 65% vs 40% | 21% vs 17% | 54% | 3 of 7 | 15 | +30.6% | 0 of 3 | — |
| Median owner-earnings IRR of our universe (rich = low) SAVNG point-in-time table — SEC filings + FirstRate prices incl. later-delisted companies | 2012-02 | 39% vs 33% | 64% vs 40% | 24% vs 17% | 61% | 2 of 7 | 12.3 | +15.6% | 0 of 2 | — |
| Insider buy/sell ratio (3-mo, all companies) SAVNG Form 4 corpus | 2006-01 | 40% vs 33% | 60% vs 40% | 35% vs 17% | 60% | 3 of 7 | 16.2 | +26.6% | 2 of 3 | — |
| Sahm rule: unemployment vs its 12-mo low FRED UNRATE | 2006-01 | 46% vs 33% | 60% vs 40% | 24% vs 17% | 54% | 4 of 7 | 23.3 | +17.4% | 2 of 4 | normal |
| Credit spread (tight = complacent) FRED / Baa proxy | 2006-01 | 28% vs 33% | 58% vs 40% | 10% vs 17% | 72% | 4 of 7 | 15 | +26.6% | 2 of 4 | — |
| S&P 12-month return FirstRate SPY | 2006-01 | 44% vs 33% | 58% vs 40% | 17% vs 17% | 56% | 6 of 7 | 14.4 | +27.3% | 2 of 6 | normal |
| CPI rent, year over year FRED CUSR0000SEHA | 2006-01 | 33% vs 33% | 54% vs 40% | 10% vs 17% | 67% | 2 of 7 | 36 | +41.4% | 2 of 2 | normal |
| Insider buying vs its own 3-yr norm SAVNG Form 4 corpus | 2006-01 | 33% vs 33% | 48% vs 40% | 0% vs 17% | 67% | 4 of 7 | 14.7 | +22.2% | 0 of 4 | — |
| Share of stocks above their 200-day average FirstRate top-1,500 incl. later-delisted stocks | 2006-01 | 21% vs 33% | 48% vs 40% | 4% vs 17% | 79% | 6 of 7 | 11.6 | +10.3% | 2 of 6 | normal |
| M2 money growth, year over year FRED M2SL | 2006-01 | 31% vs 33% | 48% vs 40% | 0% vs 17% | 69% | 2 of 7 | 30 | +77.3% | 0 of 2 | normal |
| M2 growth, change over 6 months FRED M2SL | 2006-01 | 33% vs 33% | 48% vs 40% | 15% vs 17% | 67% | 4 of 7 | 12.3 | +15.6% | 1 of 4 | normal |
| Initial jobless claims, year over year FRED ICSA | 2006-01 | 33% vs 33% | 46% vs 40% | 29% vs 17% | 67% | 6 of 7 | 14.7 | +16.7% | 3 of 6 | normal |
| Yield curve 10y−3m (inverted = negative) FRED T10Y3M | 2006-01 | 19% vs 33% | 42% vs 40% | 15% vs 17% | 81% | 3 of 7 | 17.6 | +25.2% | 2 of 3 | normal |
| Share of stocks within 5% of their own high FirstRate top-1,500 incl. later-delisted stocks | 2006-01 | 31% vs 33% | 38% vs 40% | 2% vs 17% | 69% | 6 of 7 | 9.2 | +15.6% | 2 of 6 | normal |
| Yield curve 10y−2y FRED | 2006-01 | 17% vs 33% | 35% vs 40% | 6% vs 17% | 83% | 3 of 7 | 23.6 | +26.6% | 3 of 3 | — |
| Share of stocks 20%+ below their high FirstRate top-1,500 incl. later-delisted stocks | 2006-01 | 25% vs 33% | 35% vs 40% | 2% vs 17% | 75% | 4 of 7 | 14.4 | +18.5% | 1 of 4 | normal |
| Dollar index, 6-month change FRED DTWEXBGS | 2006-06 | 53% vs 33% | 32% vs 40% | 13% vs 17% | 47% | 5 of 7 | 9.1 | +5.2% | 3 of 5 | normal |
| Case-Shiller house prices, year over year FRED CSUSHPINSA | 2006-01 | 27% vs 33% | 31% vs 40% | 21% vs 17% | 73% | 1 of 7 | 17.2 | +34.3% | 0 of 1 | normal |
| 10-year Treasury yield, change over 6 months FRED DGS10 | 2006-01 | 21% vs 33% | 29% vs 40% | 4% vs 17% | 79% | 5 of 7 | 9.1 | +9.2% | 1 of 5 | extreme |
| VIX (complacency = low) CBOE | 2006-01 | 25% vs 33% | 23% vs 40% | 6% vs 17% | 75% | 4 of 7 | 7.7 | +12.1% | 3 of 4 | — |
Monthly, 2006-01-31 → 2026-08-31 (248 months). Insider data through 2026-06. Macro series from FRED with publication lags applied, so nothing is used before it was public. "Falls" are measured from the running high after each month-end. Educational; not a forecast and not advice.