Overview › Stock breadth
90% down days and 90% up days
Days when nearly all the money traded moved one way: panic selling, or a rush back in. Updated after every US market close: 90% down days and 90% up days (Lowry), with its exact rule and its record since 2017.
What it means now
90% up and down days is neutral today. Last 90% down day: Dec 18, 2024. Last 90% up day: Apr 22, 2025. It was also neutral a week ago (38.6%; now 71.2%). On days in this state since 2017, SPY returned +3.1% on average over the next three months, against +3.8% from any day. A description of the past, not a forecast.
- SPY
- Bearish
- Bullish
- % of dollar volume in falling stocks
What is a 90% down day?
Paul Desmond of Lowry Research found that major lows tend to come with one or more 90% down days, panic selling in nearly every stock, and are confirmed when a 90% up day follows.
How it is calculated
- A 90% down day: at least 90% of the day's dollar volume in falling stocks, and at least 90% of the stocks falling (US stocks over $5). A 90% up day is the same the other way.
- Bearish for 21 trading days after a down day, bullish for 21 after an up day; the more recent one wins. Neutral otherwise.
A standard setting, fixed before testing, not the best of the variants below. A state is decided from a day’s closing prices and published after that close.
How reliable has it been? Its record since Sep 18, 2017
Holding SPY, but T-bills while 90% up and down days was bearish, would have returned 15.3% a year with a worst fall of −26.5%, against 14.8% a year and −33.7% for SPY held throughout. Of its bearish signals with three months of history after them, 4 of 10 (40%) were followed by a fall of 5% or more in SPY within three months, against 33% of all 2,220 days (neighbouring days overlap, so days are not independent events). After its 6 bullish signals SPY returned +15.1% on average over three months, against +3.8% from any day. It turned bearish ahead of 2 of the 5 declines of 10% or more in this period.
- SPY
- Bearish
- Bullish
- % of dollar volume in falling stocks
The full record and research: every state, decline, signal, variant and setting tested
SPY after each state
| From | Days | SPY, next month (avg) | SPY, next 3 months (avg) |
|---|---|---|---|
| Bearish days | 209 | +2.0% | +4.3% |
| Neutral days | 1,944 | +0.9% | +3.1% |
| Bullish days | 130 | +5.9% | +12.7% |
| All days | 2,283 | +1.3% | +3.8% |
Averages over every day in that state; the periods overlap, so treat them as a description, not a forecast.
Every decline of 10% or more
| High | Low | Fall | Bearish by the high | Days bearish on the way down | A month after the low |
|---|---|---|---|---|---|
| Jan 26, 2018 | Feb 8, 2018 | −10.1% | No | 40% | Bearish |
| Sep 20, 2018 | Dec 24, 2018 | −19.3% | No | 55% | Bullish |
| Feb 19, 2020 | Mar 23, 2020 | −33.7% | Yes | 88% | Bullish |
| Jan 3, 2022 | Oct 12, 2022 | −24.5% | No | 11% | Neutral |
| Feb 19, 2025 | Apr 8, 2025 | −18.8% | Yes | 0% | Bullish |
“Bearish by the high”: bearish at some point from a month before SPY’s high to a week after it.
Every signal (16), newest first
| Date | Turned | SPY 1 month later | 3 months | 6 months | Lowest close within 3 months |
|---|---|---|---|---|---|
| Apr 9, 2025 | Bullish | +2.9% | +14.0% | +23.0% | −6.3% |
| Dec 18, 2024 | Bearish | +3.8% | −1.5% | +4.5% | −5.6% |
| Aug 5, 2024 | Bearish | +6.5% | +10.7% | +17.5% | +0.2% |
| Nov 14, 2023 | Bullish | +5.2% | +12.3% | +18.7% | +0.2% |
| Oct 4, 2022 | Bullish | −0.8% | +2.0% | +8.8% | −5.7% |
| Jun 13, 2022 | Bearish | +1.2% | +5.3% | +7.3% | −2.2% |
| Jun 11, 2020 | Bearish | +5.2% | +11.6% | +23.1% | +0.3% |
| Apr 6, 2020 | Bullish | +7.3% | +19.0% | +29.4% | +0.1% |
| Apr 1, 2020 | Bearish | +14.9% | +26.7% | +37.2% | +0.8% |
| Mar 24, 2020 | Bullish | +14.8% | +28.9% | +37.0% | +1.2% |
| Feb 24, 2020 | Bearish | −24.1% | −7.8% | +6.4% | −30.4% |
| Aug 5, 2019 | Bearish | +3.6% | +8.4% | +17.0% | +0.0% |
| Dec 26, 2018 | Bullish | +7.1% | +14.5% | +19.5% | −0.8% |
| Dec 4, 2018 | Bearish | −5.3% | +2.2% | +7.5% | −12.8% |
| Oct 10, 2018 | Bearish | +0.8% | −6.4% | +5.3% | −15.3% |
| Feb 5, 2018 | Bearish | +3.4% | +1.5% | +8.8% | −2.4% |
Variants tested
The same measures for other common settings, so the result can be checked for luck: a rule that only works with one exact number is fragile. The highlighted row is the published rule.
| Variant | SPY or T-bills, a year | Worst fall | Days bearish | Bearish signals | Followed by a 5% fall | Bullish signals | SPY 3 months after |
|---|---|---|---|---|---|---|---|
| 80% days, 10 days | 9.7% | −22.9% | 22% | 60 | 40% (24 of 60) | 64 | +5.8% |
| 80% days, 21 days | 10.0% | −22.2% | 27% | 55 | 38% (21 of 55) | 57 | +6.3% |
| 80% days, 42 days | 7.8% | −22.9% | 33% | 54 | 37% (20 of 54) | 52 | +6.8% |
| 85% days, 10 days | 12.3% | −22.7% | 14% | 30 | 37% (11 of 30) | 24 | +7.8% |
| 85% days, 21 days | 9.4% | −22.3% | 22% | 30 | 37% (11 of 30) | 19 | +8.6% |
| 85% days, 42 days | 7.0% | −22.3% | 34% | 27 | 41% (11 of 27) | 19 | +8.6% |
| 90% days, 10 days | 16.7% | −25.6% | 6% | 10 | 40% (4 of 10) | 6 | +15.1% |
| 90% days, 21 days (this site's rule) | 15.3% | −26.5% | 9% | 10 | 40% (4 of 10) | 6 | +15.1% |
| 90% days, 42 days | 13.4% | −34.6% | 15% | 9 | 33% (3 of 9) | 6 | +15.1% |
| SPY held throughout | 14.8% | −33.7% | – | – | 33% (any day) | – | +3.8% (any day) |
“SPY or T-bills”: SPY, switching to T-bills (BIL) at the next close after the rule turns bearish and back after it stops, 0.10% per switch, interest and dividends included. Hypothetical: no costs beyond that, no taxes.
Overlap with the other indicators
It agreed most often with Zweig breadth thrust (82% of days in the same state). An indicator that almost always agrees with another adds little to the score.
Agreement with every indicator
| Indicator | Same state |
|---|---|
| Zweig breadth thrust | 82% |
| Up/down volume | 76% |
| Stocks above their 20-day average | 72% |
| Hindenburg Omen | 63% |
| Equal weight against SPY | 59% |
| Stocks beating SPY | 50% |
| McClellan Oscillator | 50% |
| Valuation | 43% |
| Stocks above their 200-day average | 34% |
| Sector breadth | 27% |
| Small against large companies | 13% |
| New highs minus new lows | 12% |
| Advance-decline line | 9% |
Research: what worked and what did not
The same test run many ways: the published rule year by year, every variant year by year, and grids over its settings. Return numbers are SPY-or-T-bills against SPY over the same days, in points; green means better than holding SPY, red worse.
Year by year
| Year | SPY | SPY or T-bills | Difference, points | SPY’s worst fall | Days bearish | Days bullish | Bearish signals |
|---|---|---|---|---|---|---|---|
| 2017* | 7.4% | 7.4% | +0.0 difference in 2017 | −1.0% | 0% | 0% | 0 |
| 2018 | −4.6% | 0.6% | +5.1 difference in 2018 | −19.3% | 24% | 2% | 3 |
| 2019 | 31.2% | 26.7% | −4.5 difference in 2019 | −6.6% | 14% | 10% | 1 |
| 2020 | 18.4% | 32.7% | +14.3 difference in 2020 | −33.7% | 18% | 11% | 3 |
| 2021 | 28.7% | 28.7% | −0.0 difference in 2021 | −5.1% | 0% | 0% | 0 |
| 2022 | −18.2% | −20.4% | −2.2 difference in 2022 | −24.5% | 9% | 9% | 1 |
| 2023 | 26.2% | 26.2% | +0.0 difference in 2023 | −10.0% | 0% | 9% | 0 |
| 2024 | 24.9% | 20.7% | −4.2 difference in 2024 | −8.4% | 12% | 0% | 2 |
| 2025 | 17.7% | 13.7% | −4.0 difference in 2025 | −18.8% | 5% | 12% | 0 |
| 2026* | 12.8% | 12.8% | +0.0 difference in 2026 | −8.9% | 0% | 0% | 0 |
* Part of a year: Sep 18 to the end of 2017, and 2026 to Oct 16.
Every variant, year by year
Does a setting win steadily, or in one lucky year? Each square is that year’s difference from SPY.
| Variant | 2017* | 2018 | 2019 | 2020 | 2021 | 2022 | 2023 | 2024 | 2025 | 2026* | A year, whole period |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 80% days, 10 days | +0.0 80% days, 10 days, 2017 | −10.7 80% days, 10 days, 2018 | −16.0 80% days, 10 days, 2019 | −0.0 80% days, 10 days, 2020 | −10.2 80% days, 10 days, 2021 | +10.2 80% days, 10 days, 2022 | −11.2 80% days, 10 days, 2023 | −19.2 80% days, 10 days, 2024 | +7.0 80% days, 10 days, 2025 | −0.0 80% days, 10 days, 2026 | −5.1 80% days, 10 days, a year over the whole period |
| 80% days, 21 days | +0.0 80% days, 21 days, 2017 | −5.8 80% days, 21 days, 2018 | −21.4 80% days, 21 days, 2019 | +9.1 80% days, 21 days, 2020 | −19.9 80% days, 21 days, 2021 | +10.2 80% days, 21 days, 2022 | −9.3 80% days, 21 days, 2023 | −18.2 80% days, 21 days, 2024 | +7.8 80% days, 21 days, 2025 | −0.0 80% days, 21 days, 2026 | −4.8 80% days, 21 days, a year over the whole period |
| 80% days, 42 days | +0.0 80% days, 42 days, 2017 | −8.8 80% days, 42 days, 2018 | −21.5 80% days, 42 days, 2019 | −0.6 80% days, 42 days, 2020 | −24.2 80% days, 42 days, 2021 | +10.2 80% days, 42 days, 2022 | −9.3 80% days, 42 days, 2023 | −22.6 80% days, 42 days, 2024 | +8.6 80% days, 42 days, 2025 | +0.0 80% days, 42 days, 2026 | −7.1 80% days, 42 days, a year over the whole period |
| 85% days, 10 days | +0.0 85% days, 10 days, 2017 | −1.3 85% days, 10 days, 2018 | −9.5 85% days, 10 days, 2019 | +8.8 85% days, 10 days, 2020 | −5.5 85% days, 10 days, 2021 | +3.2 85% days, 10 days, 2022 | +2.2 85% days, 10 days, 2023 | −10.2 85% days, 10 days, 2024 | −12.4 85% days, 10 days, 2025 | +0.0 85% days, 10 days, 2026 | −2.5 85% days, 10 days, a year over the whole period |
| 85% days, 21 days | +0.0 85% days, 21 days, 2017 | −3.2 85% days, 21 days, 2018 | −14.6 85% days, 21 days, 2019 | +3.6 85% days, 21 days, 2020 | −5.3 85% days, 21 days, 2021 | +0.7 85% days, 21 days, 2022 | +0.7 85% days, 21 days, 2023 | −16.1 85% days, 21 days, 2024 | −16.2 85% days, 21 days, 2025 | +0.0 85% days, 21 days, 2026 | −5.4 85% days, 21 days, a year over the whole period |
| 85% days, 42 days | +0.0 85% days, 42 days, 2017 | −2.1 85% days, 42 days, 2018 | −16.9 85% days, 42 days, 2019 | −8.5 85% days, 42 days, 2020 | −12.8 85% days, 42 days, 2021 | +0.7 85% days, 42 days, 2022 | −2.1 85% days, 42 days, 2023 | −15.8 85% days, 42 days, 2024 | −17.5 85% days, 42 days, 2025 | −0.0 85% days, 42 days, 2026 | −7.9 85% days, 42 days, a year over the whole period |
| 90% days, 10 days | +0.0 90% days, 10 days, 2017 | +11.2 90% days, 10 days, 2018 | −5.5 90% days, 10 days, 2019 | +21.9 90% days, 10 days, 2020 | −0.0 90% days, 10 days, 2021 | −1.2 90% days, 10 days, 2022 | −0.0 90% days, 10 days, 2023 | −8.9 90% days, 10 days, 2024 | −0.5 90% days, 10 days, 2025 | +0.0 90% days, 10 days, 2026 | +1.8 90% days, 10 days, a year over the whole period |
| 90% days, 21 days (this site's rule) | +0.0 90% days, 21 days (this site's rule), 2017, the site's rule | +5.1 90% days, 21 days (this site's rule), 2018, the site's rule | −4.5 90% days, 21 days (this site's rule), 2019, the site's rule | +14.3 90% days, 21 days (this site's rule), 2020, the site's rule | −0.0 90% days, 21 days (this site's rule), 2021, the site's rule | −2.2 90% days, 21 days (this site's rule), 2022, the site's rule | +0.0 90% days, 21 days (this site's rule), 2023, the site's rule | −4.2 90% days, 21 days (this site's rule), 2024, the site's rule | −4.0 90% days, 21 days (this site's rule), 2025, the site's rule | +0.0 90% days, 21 days (this site's rule), 2026, the site's rule | +0.4 90% days, 21 days (this site's rule), a year over the whole period, the site's rule |
| 90% days, 42 days | +0.0 90% days, 42 days, 2017 | +10.1 90% days, 42 days, 2018 | −6.3 90% days, 42 days, 2019 | +8.4 90% days, 42 days, 2020 | −0.0 90% days, 42 days, 2021 | −10.9 90% days, 42 days, 2022 | +0.0 90% days, 42 days, 2023 | −9.9 90% days, 42 days, 2024 | −1.0 90% days, 42 days, 2025 | +0.0 90% days, 42 days, 2026 | −1.4 90% days, 42 days, a year over the whole period |
The settings, tested on a grid
The 90% line and how long it lasts: return against SPY
Each square: how much more (green) or less (red) a year SPY-or-T-bills returned than SPY. Rows: the share of the day's dollar volume and of the stocks that must move one way; columns: how many trading days the state lasts.
| Lasts → | ||||||
|---|---|---|---|---|---|---|
| Share of volume and stocks ↓ | 5 days | 10 days | 21 days | 42 days | 63 days | Average |
| 80% | −4.9 Share of volume and stocks 80%, lasts 5 days | −5.1 Share of volume and stocks 80%, lasts 10 days | −4.8 Share of volume and stocks 80%, lasts 21 days | −7.1 Share of volume and stocks 80%, lasts 42 days | −6.2 Share of volume and stocks 80%, lasts 63 days | −5.6 |
| 85% | −2.9 Share of volume and stocks 85%, lasts 5 days | −2.5 Share of volume and stocks 85%, lasts 10 days | −5.4 Share of volume and stocks 85%, lasts 21 days | −7.9 Share of volume and stocks 85%, lasts 42 days | −7.5 Share of volume and stocks 85%, lasts 63 days | −5.2 |
| 90% | +0.4 Share of volume and stocks 90%, lasts 5 days | +1.8 Share of volume and stocks 90%, lasts 10 days | +0.4 Share of volume and stocks 90%, lasts 21 days, the site's rule | −1.4 Share of volume and stocks 90%, lasts 42 days | −0.7 Share of volume and stocks 90%, lasts 63 days | +0.1 |
| 95% | +1.5 Share of volume and stocks 95%, lasts 5 days | +1.8 Share of volume and stocks 95%, lasts 10 days | −0.8 Share of volume and stocks 95%, lasts 21 days | −1.7 Share of volume and stocks 95%, lasts 42 days | −2.2 Share of volume and stocks 95%, lasts 63 days | −0.3 |
| Average | −1.5 | −1.0 | −2.6 | −4.5 | −4.1 | |
The 90% line and how long it lasts: how much smaller the worst fall was
Each square: SPY's worst fall minus the strategy's, in points (green: it fell less). Rows: the share of the day's dollar volume and of the stocks that must move one way; columns: how many trading days the state lasts.
| Lasts → | ||||||
|---|---|---|---|---|---|---|
| Share of volume and stocks ↓ | 5 days | 10 days | 21 days | 42 days | 63 days | Average |
| 80% | +12.1 Share of volume and stocks 80%, lasts 5 days | +10.8 Share of volume and stocks 80%, lasts 10 days | +11.5 Share of volume and stocks 80%, lasts 21 days | +10.8 Share of volume and stocks 80%, lasts 42 days | +10.8 Share of volume and stocks 80%, lasts 63 days | +11.2 |
| 85% | +8.7 Share of volume and stocks 85%, lasts 5 days | +11.0 Share of volume and stocks 85%, lasts 10 days | +11.4 Share of volume and stocks 85%, lasts 21 days | +11.4 Share of volume and stocks 85%, lasts 42 days | +11.4 Share of volume and stocks 85%, lasts 63 days | +10.8 |
| 90% | +7.9 Share of volume and stocks 90%, lasts 5 days | +8.1 Share of volume and stocks 90%, lasts 10 days | +7.2 Share of volume and stocks 90%, lasts 21 days, the site's rule | −0.9 Share of volume and stocks 90%, lasts 42 days | +5.9 Share of volume and stocks 90%, lasts 63 days | +5.6 |
| 95% | +9.2 Share of volume and stocks 95%, lasts 5 days | +9.2 Share of volume and stocks 95%, lasts 10 days | +9.2 Share of volume and stocks 95%, lasts 21 days | +9.2 Share of volume and stocks 95%, lasts 42 days | +9.2 Share of volume and stocks 95%, lasts 63 days | +9.2 |
| Average | +9.5 | +9.8 | +9.8 | +7.6 | +9.3 | |
The 90% line and how long it lasts: SPY after the bullish days
Each square: SPY's average return over the next 63 trading days from the bullish days, minus its average from every day, in points (green: bullish days were followed by better than usual returns). Rows: the share of the day's dollar volume and of the stocks that must move one way; columns: how many trading days the state lasts.
| Lasts → | ||||||
|---|---|---|---|---|---|---|
| Share of volume and stocks ↓ | 5 days | 10 days | 21 days | 42 days | 63 days | Average |
| 80% | +0.9 Share of volume and stocks 80%, lasts 5 days | +0.7 Share of volume and stocks 80%, lasts 10 days | +0.2 Share of volume and stocks 80%, lasts 21 days | −0.3 Share of volume and stocks 80%, lasts 42 days | −0.2 Share of volume and stocks 80%, lasts 63 days | +0.3 |
| 85% | +3.5 Share of volume and stocks 85%, lasts 5 days | +3.4 Share of volume and stocks 85%, lasts 10 days | +3.0 Share of volume and stocks 85%, lasts 21 days | +2.3 Share of volume and stocks 85%, lasts 42 days | +2.1 Share of volume and stocks 85%, lasts 63 days | +2.8 |
| 90% | +10.9 Share of volume and stocks 90%, lasts 5 days | +10.2 Share of volume and stocks 90%, lasts 10 days | +8.9 Share of volume and stocks 90%, lasts 21 days, the site's rule | +6.9 Share of volume and stocks 90%, lasts 42 days | +5.3 Share of volume and stocks 90%, lasts 63 days | +8.4 |
| 95% | – Share of volume and stocks 95%, lasts 5 days | – Share of volume and stocks 95%, lasts 10 days | – Share of volume and stocks 95%, lasts 21 days | – Share of volume and stocks 95%, lasts 42 days | – Share of volume and stocks 95%, lasts 63 days | – |
| Average | +5.1 | +4.8 | +4.0 | +2.9 | +2.4 | |
The two halves of the history
The best settings found over the whole period, and how each did in 2017–2021 and in 2022–2026 on its own. A setting that only works in one half is fitted to the past.
| Setting | Whole period, a year | 2017–2021 | 2022–2026 | Worst fall smaller by | Ahead in both halves |
|---|---|---|---|---|---|
| The site's rule, but lasting 10 days | +1.8 whole period | +6.8 2017 to 2021 | −2.1 2022 to 2026 | +8.1 worst fall | No |
| The site's rule, but 95% days, lasting 10 days | +1.8 whole period | +4.0 2017 to 2021 | +0.0 2022 to 2026 | +9.2 worst fall | Yes |
| The site's rule, but 95% days, lasting 5 days | +1.5 whole period | +3.4 2017 to 2021 | +0.0 2022 to 2026 | +9.2 worst fall | Yes |
| The site's rule, but lasting 5 days | +0.4 whole period | +2.6 2017 to 2021 | −1.5 2022 to 2026 | +7.9 worst fall | No |
| The site's rule, but lasting 63 days | −0.7 whole period | +1.4 2017 to 2021 | −2.4 2022 to 2026 | +5.9 worst fall | No |
| The site's rule, but 95% days | −0.8 whole period | −1.8 2017 to 2021 | −0.0 2022 to 2026 | +9.2 worst fall | No |
| The site’s rule | +0.4 whole period, the site's rule | +3.7 2017 to 2021, the site's rule | −2.2 2022 to 2026, the site's rule | +7.2 worst fall, the site's rule | No |
Summary: the settings that helped the most
- Best single setting tested: the site's rule, but lasting 10 days: +1.8 points a year over SPY (+6.8 in 2017–2021, −2.1 in 2022–2026), against +0.4 for the site’s rule. Found by looking at the results, so it is the most flattering number here, not a forecast.
- Share of volume and stocks: 90% did best on average (+0.1), 80% worst (−5.6).
- Lasts: 10 days did best on average (−1.0), 42 days worst (−4.5).
- Share of volume and stocks (for the bullish days): 90% did best on average (+8.4), 80% worst (+0.3).
- Lasts (for the bullish days): 5 days did best on average (+5.1), 63 days worst (+2.4).
- Choosing the best setting in advance: the setting with the best result in 2017–2021 (the site's rule, but lasting 10 days, +6.8) scored −2.1 in 2022–2026, against −2.2 for the site’s rule; 11 of the 20 settings tested beat the rule in those later years.
The site keeps the rule it published, so its live record stays honest. The heat maps show which ingredients mattered across many settings; any one best square is partly luck, because the history holds only a handful of real declines.
Live record
The rule went live after the close of Oct 16, 2026. Every reading since was made by the daily run after that day’s close and is kept exactly as made; the newest few are corrected only if late price data changes them.
No live readings yet: the first comes after the next trading day.
Free · every Friday after the close
Know when the market’s underlying health changes
Every Friday, a two-minute report on market health, rising risk and the indicators that changed, without having to watch every chart yourself.
- The health score and the top-risk score, and how far each moved
- Every indicator that turned bullish or bearish, and why
- Whether breadth confirmed the S&P 500’s new highs
- The charts that matter, in under two minutes
From the same author
Bull Rhythm: momentum strategies, tracked live every trading day
Eight model stock portfolios built on 12-1 momentum, one of the most researched return factors in finance, run by fixed rules and measured against the S&P 500.
- Every holding and every trade published, with the reason in plain words
- The live record since go-live, next to a backtest since 2017
- The Bull Rhythm Weekly by email every Friday