When Only 10% of the S&P 500 Is Above Its 20-Day Average, What Happens Next?
Chart accounts post the same line every time this indicator craters: "breadth this washed out means we're near a bottom." It's usually paired with a screenshot of the $NAA20R line pinned near zero and nothing else — no count of how many times that actually worked, no comparison to what stocks do in a random month, no mention of the times it kept falling.
We ran the numbers. Going back to 2000, we found 46 independent episodes where the percentage of S&P 500 stocks trading above their 20-day moving average bottomed at 12% or below (38 of those bottomed at 10% or below — the number circulating right now). Three months later, QQQ was up 6.07% on average versus a 2.71% unconditional baseline over the same stretch. Six months out, the gap widens: +11.72% versus +5.63%, with a 78.3% hit rate against a 72.4% baseline hit rate. The edge is real. It is also inconsistent, and it failed badly enough in 2008 and 2021 that you need to read the failure section before touching this.
As of September 10, 2026, the reading is 20.87% — down hard from 54.47% three weeks earlier, but not yet in the zone this analysis covers.
What we measured
Signal: percentage of current S&P 500 constituents with a daily close above their 20-day simple moving average, computed from daily closes, January 2000 through September 2026. A signal fires at a local trough where that percentage falls to 12% or below; the entry is the next trading day's close (the trough itself isn't tradable), and episodes within 20 trading days of each other are collapsed into one to avoid double-counting the same drawdown.
Universe caveat: this uses today's S&P 500 member list applied backward through the full history. Companies removed from the index over the years — bankruptcies, buyouts, demotions — are absent from the whole series, which means the reconstructed breadth troughs are probably a bit less extreme than what actually traded at the time. Treat every number below as a slightly conservative estimate of how oversold the market really got.
Sample: 46 signals, September 2001 through April 2025. That's not a huge sample, but it spans 2001, 2002, 2008 (nine separate times), 2010, 2011, 2018, 2020, 2022, and 2025 — a real mix of regimes, not one crash repeated.
What the data says
| Horizon | Signal mean | Baseline mean | Signal median | Signal hit rate | Baseline hit rate |
|---|---|---|---|---|---|
| 1 month | +1.28% | +0.91% | +2.82% | 63.0% | 62.5% |
| 3 months | +6.07% | +2.71% | +9.22% | 65.2% | 67.9% |
| 6 months | +11.72% | +5.63% | +13.80% | 78.3% | 72.4% |
Two things stand out. First, the edge builds with time — at one month the signal barely beats a random entry (63.0% vs 62.5% hit rate is noise), but by six months the mean return is roughly double the baseline. Second, the median beats the mean at every horizon, meaning the average result is dragged down by a handful of bad outcomes, not by a weak typical case — the typical case (the median) is actually stronger than the headline mean suggests.
We also checked whether 10-12% was cherry-picked. It wasn't: running the same test at thresholds of 8%, 10%, 12%, 15%, 18%, and 20% all produce 3-month means in the same 5.8% to 7.5% range. This isn't a signal that only exists at one magic number — it's a broad "breadth got this washed out" effect.
The actionable rule
When the percentage of S&P 500 stocks above their 20-day SMA bottoms at 12% or below and turns back up, that's historically been a better-than-random entry for QQQ on a 3-6 month horizon — lean toward adding exposure on the turn, not on the way down. Don't expect an edge inside the first month; the one-month numbers are indistinguishable from baseline, so this is a swing/position-trade signal, not a day-trade trigger. At 20.9% and falling as of today, we are not there yet — the setup this analysis covers hasn't triggered.
Where it fails
Ten of the 46 six-month outcomes were losses, and the worst ones were not close calls. The late-September 2008 signal (breadth trough of 6.02%, right as Lehman-driven forced selling was still accelerating) returned -20.7% over the next six months with a -34.3% drawdown along the way — the market undercut the "washed out" level by a wide margin before finally basing in March 2009. The June 26, 2008 signal fared even worse: -36.05% six months out with a 43.97% intra-trade drawdown. In both cases the breadth reading was correctly flagging distress, but distress was nowhere near finished.
More recently, the early-December 2021 signal (breadth trough 10.16%, the start of the 2022 rate-hike bear market) returned -21.3% over six months. This is the pattern to watch for: when the breadth washout is the first crack in a structural regime change rather than a sharp, event-driven flush, the "buy the extreme" version of this trade gets run over. All three of the worst failures shared the same tell — the trough wasn't an isolated spike, it kept making new lows for weeks afterward instead of reversing within the next 20 trading days.
How to watch it
Six7Alpha's Market Health dashboard tracks this breadth metric in real time. At 20.9% and dropping, we're getting closer to the zone this signal covers — worth watching for a turn, not worth acting on yet.