Home › Articles

RSI Oversold Crossovers Buy Low, but Rarely at the Bottom

2026-10-19 · Research note · Series: RSI oversold crossover backtest (part 1 of 3) · Not investment advice; figures are as of the sources and dates cited in the text · 中文
Contents11 sections
  1. Signal definition, data sources and calculation
    1. Signal rule: RSI6 crosses above RSI10 within 10 trading days of an oversold reading
    2. Data covers 6945 trading days for QQQ and 4051 for VOO
  2. VOO: 47 signals with an 85% 6-month win rate, slightly ahead of random entry
  3. QQQ: 77 signals trail random entry over the full history, turning positive only after 2010
  4. Only one-fifth to two-fifths of signals land within 3% of the relative low
    1. Waiting for the crossover costs 1.2% to 1.8% on average versus buying on the oversold day
  5. Signals fail in a row when the price is more than 20% below the 52-week high
  6. Public backtests and academic research do not support oversold signals outperforming over the long run
  7. QQQ complete event tables (main definition, 77 events)
    1. QQQ Table One: signal date, entry price and prices 1/3/6 months later
    2. QQQ Table Two: 6-month drawdown after entry and relative-low determination
  8. VOO complete event tables (main definition, 47 events)
    1. VOO Table One: signal date, entry price and prices 1/3/6 months later
    2. VOO Table Two: 6-month drawdown after entry and relative-low determination
  9. Limitations and the latest status as of 2026-10-16
    1. Overlapping holding periods, same-day signals and ignored costs all overstate the strength of the evidence
    2. QQQ has entered a new oversold stretch but has not yet produced a crossover
  10. Conclusion
  11. Frequently asked questions

Short answer: the signal "RSI6 drops below 20, then RSI6 crosses above RSI10" pushes your entry price toward the low end of the range, but most of the time it does not catch the "relative low," and in long bear markets it fails repeatedly. Backtesting the full daily history from Yahoo Finance (QQQ from 1999-03-10, VOO from 2010-09-09, both through the 2026-10-16 close), the main definition produces 77 signals for QQQ and 47 for VOO. Defining a relative low as "an entry price within 3% of the lowest close from 20 days before to 126 days after the signal date," only 14/76 (18.4%) of QQQ signals and 18/47 (38.3%) of VOO signals qualify. That beats random trading days (10.9% and 18.9%), but roughly 4/5 of QQQ signals and 3/5 of VOO signals are more than 3% above the low. On returns the two funds diverge: VOO's 6-month average is +8.5% with an 85.1% win rate, slightly ahead of random entry at +6.3% and 80.2%; QQQ's full-history 6-month average is +4.4% with a 69.7% win rate, actually trailing random entry at +6.5% and 72.9%, a gap that comes mainly from the 2000–2002 and 2008 bear markets. More notably, "waiting for the crossover" is itself paying for confirmation: the signal-day close is on average 1.75% (QQQ) and 1.19% (VOO) above the close on the first day it broke below 20 (D0), and buying the same events at D0 instead produces better 1/3/6-month returns across the board. The most reliable failure warning is depth of decline: QQQ had 20 signals that fired when the price was already more than 20% below its 252-day (about 52-week) high, with a 6-month win rate of only 20% and an average of −12.7%. As of 2026-10-16, QQQ's RSI6 = 15.5, inside a new oversold stretch, but it has not crossed yet, so it is not a signal. Below we first explain the definitions and data, then analyze returns, entry location and failure scenarios, and finally list the complete event-by-event tables for both funds.

This article is historical backtesting and methodology research. It is not investment advice and not a buy or sell signal. Figures are computed from Yahoo Finance daily data (price returns use split-adjusted closes without dividends; total returns use dividend-adjusted prices), with data through the 2026-10-16 close. The scope differs from this site's "Panic Oversold Diagnosis": this article uses QQQ/VOO themselves, while the panic diagnosis uses the three major indexes.

Signal definition, data sources and calculation

Signal rule: RSI6 crosses above RSI10 within 10 trading days of an oversold reading

RSI uses Wilder smoothing (RMA, α = 1/n, seeded with the average of the first n days), consistent with TradingView ta.rsi and most charting software (Wikipedia). The crossover day T requires RSI6 ≤ RSI10 on the prior day and RSI6 > RSI10 on the day itself; the oversold condition requires at least one day with RSI6 < 20 within the 10 trading days looking back from T (including T, and later than the previous signal). D0 is the earliest RSI6 < 20 day in that stretch, and the entry price is always the close on day T. A new oversold day must occur after the previous signal before another signal can fire, so the same oversold stretch is not counted twice. This combination of "RSI < 20 means oversold, and a short-period RSI crossing above a long-period RSI means buy" is exactly a rule of thumb common in Chinese-language investing education (CMoney), though some tutorials set the "low zone" at below 50 rather than 20 (Wantgoo); this study uses the stricter 20.

Each event is measured as shown in the table below. 1/3/6 months are counted as 21/63/126 trading days; price returns use split-adjusted closes, and total return (TR) uses the dividend-inclusive adjclose. Events with less than 6 months of data are marked N/A but are not dropped.

Measure Formula
+h-day return C[T+h] ÷ C[T] − 1, h = 21/63/126
6-month max drawdown (MDD) min(C[T+1..T+126]) ÷ C[T] − 1, recorded as 0 if positive
Relative low (Yes/No) C[T] ≤ min(C[T−20..T+126]) × 1.03
Entry-price percentile Share of days in C[T−63..T+126] with a close below C[T] (0 = window low)
From 252-day high C[T] ÷ max(C[T−251..T]) − 1, known on the signal day, not hindsight
Unconditional baseline Every trading day with complete forward data treated as an entry day, same measures

Data covers 6945 trading days for QQQ and 4051 for VOO

Data comes from the Yahoo Finance chart API daily bars (downloaded 2026-10-19) (Yahoo Finance QQQ; Yahoo Finance VOO). QQQ covers 1999-03-10 to 2026-10-16 and VOO covers 2010-09-09 (launch date) to 2026-10-16, with no missing values for either. The 2000-03 QQQ split and the 2013-10 VOO split are already adjusted by Yahoo, with no abnormal gaps around the split dates. RSI was computed with two independent implementations, which differ by less than 1e-5 after bar 200; an RSI6 started only from 2015-01-01 matches the full-history value on the last day to the second decimal place, so the seed does not affect signal detection. Spot-checking 2025-04-08, the lowest close of the tariff selloff, QQQ/VOO RSI6 readings were 13.2 and 11.1, consistent with extreme oversold. Under the main definition both funds average about 2.8 to 2.9 signals a year, so these are not rare events.

VOO: 47 signals with an 85% 6-month win rate, slightly ahead of random entry

VOO's results come closest to the intuition that "the signal works." Average 1/3/6-month returns of +1.9%/+4.5%/+8.5% are above the unconditional baseline's +1.1%/+3.2%/+6.3% in all three periods, with a 6-month win rate of 85.1% versus 80.2%. But the rough t-statistics are only 1.18 to 1.76, none above 2; with 47 samples and overlapping holding periods, the result cannot yet be statistically distinguished from random entry. Only 2 of VOO's 47 signals saw a drawdown of more than 20% within 6 months (the worst, an entry on 2020-03-02, at −28.0%), and the median 6-month MDD of −3.8% is almost the same as −3.7% for random trading days. In other words, the signal did not spare you drawdowns; it only delivered slightly higher returns.

VOO entry method 1M win rate 1M mean/median 3M win rate 3M mean/median 6M win rate 6M mean/median
Signal day T close (n=47) 72.3% +1.9%/+1.9% 74.5% +4.5%/+4.7% 85.1% +8.5%/+8.8%
Same events, D0 close instead 74.5% +2.7%/+3.3% 76.6% +5.5%/+5.5% 87.2% +9.6%/+9.7%
Unconditional baseline (all trading days) 67.9% +1.1%/+1.6% 76.0% +3.2%/+3.9% 80.2% +6.3%/+7.0%

VOO's sample falls entirely within the long bull market after 2010, which is the biggest caveat when reading it. It never went through a bear market like 2000–2002 or 2008, lasting more than a year with declines of more than half, so a more precise version of "it works for VOO" is "in the 2010–2026 bull market, buying sharp S&P 500 pullbacks did slightly better than buying on any given day." The 2022 rate-hike bear market is the only stress test: the 6-month returns of VOO #34 to #39 were −7.0%, −11.2%, +1.3%, −1.6%, +8.9% and +11.4% in order; signals in the middle of the bear market mostly failed, turning positive only near the end.

QQQ: 77 signals trail random entry over the full history, turning positive only after 2010

QQQ's full-history numbers are the counterexample. The 6-month average of +4.4% and win rate of 69.7% are below the unconditional baseline's +6.5% and 72.9% (rough t = −1.06); the 1-month average is only +0.4%, also below the baseline's +1.0%. 14/76 entries saw a drawdown of more than 20% within 6 months, the worst being −58.1% after the 2000-10-13 entry.

QQQ entry method 1M win rate 1M mean/median 3M win rate 3M mean/median 6M win rate 6M mean/median
Signal day T close (n=77/76) 57.1% +0.4%/+1.7% 65.8% +2.8%/+5.9% 69.7% +4.4%/+7.9%
Same events, D0 close instead 64.9% +2.0%/+2.4% 71.1% +4.8%/+6.6% 71.1% +5.8%/+8.7%
Unconditional baseline (all trading days) 62.4% +1.0%/+1.6% 68.4% +3.1%/+4.3% 72.9% +6.5%/+8.7%

Split QQQ into two periods at VOO's launch date and the conclusion flips completely. The 37 signals from 1999-03 to 2010-09 averaged −2.7% over 6 months with a 51.4% win rate, against a same-period baseline of +2.7% and 60.1%; the 40 signals from 2010-09 onward averaged +11.1% over 6 months with an 87.2% win rate, against a same-period baseline of +9.2% and 82.4%. In the second half QQQ, like VOO, edges out the baseline; in the first half the signal failed outright. In other words, the signal's performance is determined almost entirely by "whether it was a bull market at the time," not by how accurate RSI itself is.

QQQ period Signals 6M win rate 6M mean/median Same-period baseline win rate Same-period baseline mean/median Relative lows 6M MDD median
1999-03 to 2010-09-08 37 51.4% −2.7%/+0.2% 60.1% +2.7%/+5.0% 7/37 −9.3%
2010-09-09 to 2026-10-16 40 87.2% +11.1%/+10.9% 82.4% +9.2%/+10.1% 7/39 −6.0%

Only one-fifth to two-fifths of signals land within 3% of the relative low

Back to the core question: does this buy the relative low? The answer is low-ish, but usually not the low. The median percentile of the entry price within the "prior 3 months plus following 6 months" window is 25.7 for QQQ and 16.9 for VOO, versus 38.1 and 37.0 for random trading days, so the signal does push the entry toward the bottom of the range. But under the strict definition (entry within 3% of the window's lowest close), only 14/76 (18.4%) for QQQ and 18/47 (38.3%) for VOO pass. Using a looser view that only asks "did the price fall more than 3% further within 6 months of entry," the share with no further decline is 20/76 (26%) for QQQ and 20/47 (43%) for VOO. Same conclusion: most signals are followed by lower prices.

Measure QQQ VOO
Relative-low rate: entry on signal day T 14/76 (18.4%) 18/47 (38.3%)
Relative-low rate: buy on the first day of each break below 20 (D0 stretch strategy) 28.7% (n=87) 47.2% (n=53)
Relative-low rate: unconditional, all trading days 10.9% 18.9%
Median entry-price percentile: signal / D0 stretch / all trading days 25.7/19.6/38.1 16.9/14.8/37.0
Entry-price percentile 25%–75% quantiles (signal) 7.9 to 49.9 8.7 to 38.9
T close above prior 20-day lowest close (mean/median) +3.61%/+2.80% +2.64%/+2.20%

The failure mechanism is clear: the crossover is a lagging confirmation. The signal-day close is on average 3.6% (QQQ) and 2.6% (VOO) above the lowest close of the prior 20 days, and as much as 12.2% above. The bounce you wait through for the crossover alone often pushes the entry outside the 3% band. The 2026-03-31 signal best shows how strict this definition is: QQQ and VOO both had an entry percentile of 1.6, nearly the very bottom of the whole window, and were up +28.2% and +17.3% 6 months later, yet QQQ is still judged "No" because its entry was 3.4% above the 03-30 low.

Waiting for the crossover costs 1.2% to 1.8% on average versus buying on the oversold day

If the goal is to buy low, the "wait for the crossover" step is a negative contribution in this data. The median D0→T gap is only 3 trading days, but the T close is above the D0 close in 64/77 (83%) of QQQ events and 37/47 (79%) of VOO events, more expensive by 1.75% and 1.19% on average. Buying the same events at D0 instead gives better mean and median returns at 1/3/6 months for both funds; comparing 6-month returns event by event, D0 entry beats T entry in 55/76 QQQ events and 33/47 VOO events. Waiting for confirmation is commonly thought to buy a smaller drawdown, but the median 6-month MDD is −6.6% for QQQ and −3.5% for VOO under the D0 stretch strategy, versus −7.4% and −3.8% for signal entries, so waiting for the crossover was not safer. The most extreme example is 2025-04-09: QQQ and VOO both crossed only on that day's big rally, with entry prices 10.3% and 7.2% above D0. They were still up +31.1% and +23.6% 6 months later, but the first leg of the rebound went to someone else.

Sensitivity tests point the same way. Changing the crossover window to 5 or 20 days barely changes the results (QQQ 71/77 signals, VOO 46/48 signals, with 6-month win rates within 1 percentage point), because most crossovers already occur within 3 days. Making the oversold condition stricter (RSI10 < 20) shrinks the sample to 10 QQQ and 8 VOO signals, with 1-month win rates of only 40% and 50% and relative-low rates of 0/10 and 1/8. Deeper oversold readings often mean the decline is not over, and the signal does not become more accurate.

Signals fail in a row when the price is more than 20% below the 52-week high

The most practically useful finding is the failure mode, and it is visible on the signal day itself. Grouping signals by "distance from the 252-day high," QQQ has a 6-month win rate of 90.7% and an average of +9.8% when the decline is under 10%; 76.9% and +12.9% for a decline of 10% to 20%; once the decline exceeds 20%, the 20 signals have a 6-month win rate of only 20%, an average of −12.7% and a median MDD of −27.8%. All 20 fall within the 2000–2002, 2008 and 2022 bear markets; only four of them, 2001-09-24, 2002-09-25, 2008-11-24 and 2022-09-28, were positive after 6 months, and all were close to major bottoms.

Fund Group (signal day vs 252-day high) Events 6M win rate 6M mean 6M median 6M MDD median
QQQ Decline under 10% 43 90.7% +9.8% +10.7% −6.0%
QQQ Down 10% to 20% 13 76.9% +12.9% +15.1% −5.7%
QQQ Down more than 20% 20 20.0% −12.7% −12.6% −27.8%
VOO Decline under 10% 35 85.7% +8.0% +8.0% −3.3%
VOO Down 10% to 20% 8 75.0% +9.4% +11.3% −5.0%
VOO Down more than 20% 4 100.0% +11.4% +10.2% −4.6%

VOO's 4 signals with a decline of more than 20% (2020-03-13 and 3 in 2022) were all positive after 6 months, which looks like the opposite of QQQ. But 4 is too few, and all of them occurred in a V-shaped rebound or at the tail end of a bear market, so they are not enough to overturn the 20 QQQ data points. Clustered signals show the same dependence on the environment: QQQ's 14 signals within 20 trading days of the previous signal averaged −0.2% over 6 months versus +5.4% for non-clustered ones, while VOO's 7 clustered signals averaged +12.2%. Clustering itself is neither good nor bad; what matters is the broader environment it occurs in. During the dot-com bubble only 2 of QQQ's 12 signals #1 to #12 were positive after 6 months, and in the financial crisis only one of the 8 signals #26 to #33, 2008-11-24, was clearly positive. These two stretches are the main reason QQQ's full history trails random entry.

Published research has long treated this failure mode as standard. Chinese-language tutorials warn that RSI "desensitizes" below 20, and that buying in the low zone during a persistent decline easily leaves you trapped (Chiang Ming-cheng / Cnyes); Arthur Hill's 2016 RSI5 mean-reversion test on SPY and QQQ traded only when the S&P 500's 50-day moving average was above its 200-day moving average, precisely to avoid bear markets (StockCharts); the classic RSI(2) rule popularized by Connors likewise requires the close to be above the 200-day moving average before going long (LuxAlgo). The signal in this study has no trend filter at all, and QQQ's dismal first half is the price of going without one. "Down more than 20% from the 252-day high" can serve as a simple filter that can be judged on the day, but it was derived after the fact from the same data and still needs out-of-sample validation.

Public backtests and academic research do not support oversold signals outperforming over the long run

The research did not find any transparent public backtest of the RSI6/RSI10 crossover on SPY, VOO or QQQ, nor any forward statistics for fixed 1/3/6-month holding periods, so the event-by-event results in this study are themselves the main evidence. The literature on similar short-period RSI oversold strategies shows a common pattern: "high win rate, small gains per trade, overall lagging buy-and-hold." Backtrex's test of RSI(2) < 5 on the S&P 500 from 2016-10 to 2026-10 had a win rate of 71.23% after costs, but an annualized return of only 1.7%, versus 13.5% for buy-and-hold over the same period (Backtrex); CXO Advisory's test of RSI14 30/70 on SPY from 1993–2008 won 27 of 34 trades but still trailed buy-and-hold over the full period, lagging in bull markets and winning in bear markets (CXO Advisory); Sentimentrader's 2017 RSI2 breadth signal also acknowledged that out-of-sample performance was weaker than in-sample (Sentimentrader).

The academic background is even less favorable to the idea that "holding for 1 to 6 months after an oversold reading does especially well." Moskowitz, Ooi and Pedersen (2012) found across 58 assets, including equity index futures, that returns tend to persist rather than reverse over 1 to 12 months (SSRN), meaning an index that has just fallen sharply will not necessarily do better over the following months; Urquhart, Gebka and Hudson (2015) retested the technical rules of Brock et al. (1992) and found their predictive power had weakened markedly after 1986 (Southampton ePrints); Park and Irwin's literature review notes that most technical analysis studies suffer from data mining and ex-post rule selection (AgEcon Search). These studies are consistent with this backtest: the small excess returns of VOO and post-2010 QQQ look more like a dividend from "buying pullbacks in a bull market" than independent predictive power in the RSI crossover; when the medium-term trend is down, momentum overwhelms the oversold bounce.

QQQ complete event tables (main definition, 77 events)

The two tables below list all 77 events one by one from the CSV file events_QQQ_P_RSI6lt20_W10.csv, and the count matches the CSV. The first table covers prices and timing: D0 is the first day of the stretch with RSI6 < 20, T is the crossover signal day (entry day), and +1M/+3M/+6M are the date, close and price return on trading day 21/63/126 after T. The second table covers entry location: the lowest close and max drawdown within 6 months, the window low, whether it is a relative low, the percentile, and the 6-month return if the same event were entered at D0. #77 has less than 6 months since entry (only 55 trading days), so its 3M/6M are marked N/A. The context column is only a reading aid; events after 2025-05 are not labeled.

QQQ Table One: signal date, entry price and prices 1/3/6 months later

# Context D0 (first RSI6<20) D0 close Signal date T D0→T days T close T vs D0 From 252-day high +1M date +1M close +1M return +3M date +3M close +3M return +6M date +6M close +6M return
1 Dot-com crash 2000-04-13 88.00 2000-04-17 2 89.62 +1.8% −23.9% 2000-05-17 88.94 −0.8% 2000-07-18 99.50 +11.0% 2000-10-16 82.00 −8.5%
2 Dot-com crash 2000-07-28 86.88 2000-08-03 4 90.44 +4.1% −23.2% 2000-09-01 102.62 +13.5% 2000-11-01 81.00 −10.4% 2001-02-02 61.55 −31.9%
3 Dot-com crash 2000-10-10 78.31 2000-10-13 3 81.25 +3.8% −31.0% 2000-11-13 71.86 −11.6% 2001-01-16 61.98 −23.7% 2001-04-17 41.25 −49.2%
4 Dot-com crash 2001-02-09 56.40 2001-02-15 4 58.40 +3.5% −50.4% 2001-03-19 43.26 −25.9% 2001-05-17 47.90 −18.0% 2001-08-16 39.56 −32.3%
5 Dot-com crash 2001-02-22 51.00 2001-03-05 7 47.55 −6.8% −59.6% 2001-04-03 34.66 −27.1% 2001-06-04 45.83 −3.6% 2001-08-31 36.63 −23.0%
6 Dot-com crash 2001-03-12 42.30 2001-03-13 1 44.45 +5.1% −62.3% 2001-04-11 41.00 −7.8% 2001-06-12 45.80 +3.0% 2001-09-17 31.20 −29.8%
7 Dot-com crash 2001-04-03 34.66 2001-04-05 2 37.31 +7.6% −65.0% 2001-05-07 47.20 +26.5% 2001-07-06 41.65 +11.6% 2001-10-10 32.50 −12.9%
8 Dot-com crash 2001-09-06 33.90 2001-09-24 8 29.61 −12.7% −68.7% 2001-10-23 34.63 +17.0% 2001-12-21 39.48 +33.3% 2002-03-26 35.89 +21.2%
9 Dot-com crash 2002-02-07 35.19 2002-02-08 1 36.17 +2.8% −41.3% 2002-03-12 38.00 +5.1% 2002-05-10 29.56 −18.3% 2002-08-09 23.33 −35.5%
10 Dot-com crash 2002-04-26 31.04 2002-05-08 8 31.77 +2.4% −37.8% 2002-06-07 28.30 −10.9% 2002-08-07 22.80 −28.2% 2002-11-05 26.10 −17.8%
11 Dot-com crash 2002-06-11 27.46 2002-06-17 4 28.56 +4.0% −37.6% 2002-07-17 25.65 −10.2% 2002-09-16 22.66 −20.7% 2002-12-13 25.04 −12.3%
12 Dot-com crash 2002-09-23 21.02 2002-09-25 2 21.87 +4.0% −49.0% 2002-10-24 23.98 +9.6% 2002-12-24 25.35 +15.9% 2003-03-27 26.37 +20.6%
13 — 2004-03-11 34.87 2004-03-12 1 35.51 +1.8% −8.0% 2004-04-13 36.63 +3.2% 2004-06-14 36.32 +2.3% 2004-09-13 35.58 +0.2%
14 — 2004-07-14 35.19 2004-07-20 4 35.43 +0.7% −8.2% 2004-08-18 33.85 −4.5% 2004-10-18 36.14 +2.0% 2005-01-18 38.72 +9.3%
15 — 2004-08-06 32.72 2004-08-16 6 32.89 +0.5% −14.8% 2004-09-15 35.25 +7.2% 2004-11-12 38.66 +17.5% 2005-02-14 37.87 +15.1%
16 — 2005-01-05 38.54 2005-01-12 5 38.58 +0.1% −4.7% 2005-02-11 37.70 −2.3% 2005-04-14 35.55 −7.9% 2005-07-14 38.77 +0.5%
17 — 2005-01-24 36.53 2005-01-26 2 37.15 +1.7% −8.2% 2005-02-25 37.62 +1.3% 2005-04-27 35.10 −5.5% 2005-07-27 39.72 +6.9%
18 — 2005-04-15 34.74 2005-04-21 4 35.62 +2.5% −12.0% 2005-05-20 37.66 +5.7% 2005-07-21 39.47 +10.8% 2005-10-19 38.67 +8.6%
19 — 2005-06-24 36.97 2005-07-05 6 37.11 +0.4% −8.3% 2005-08-03 40.06 +7.9% 2005-10-03 39.56 +6.6% 2006-01-03 41.31 +11.3%
20 — 2005-10-12 37.49 2005-10-14 2 38.06 +1.5% −6.0% 2005-11-14 40.71 +7.0% 2006-01-17 42.70 +12.2% 2006-04-18 42.46 +11.6%
21 — 2006-05-12 40.19 2006-05-25 9 39.36 −2.1% −8.9% 2006-06-26 38.25 −2.8% 2006-08-24 38.25 −2.8% 2006-11-22 44.73 +13.6%
22 — 2006-06-12 37.40 2006-06-15 3 38.65 +3.3% −10.6% 2006-07-17 36.03 −6.8% 2006-09-14 39.99 +3.5% 2006-12-13 43.88 +13.5%
23 — 2006-07-14 35.94 2006-07-19 3 36.62 +1.9% −15.3% 2006-08-17 38.73 +5.8% 2006-10-17 42.00 +14.7% 2007-01-19 44.17 +20.6%
24 — 2007-03-02 42.48 2007-03-08 4 42.97 +1.2% −5.4% 2007-04-09 44.45 +3.4% 2007-06-07 46.34 +7.8% 2007-09-06 49.14 +14.4%
25 — 2007-07-27 47.99 2007-08-07 7 48.31 +0.7% −4.0% 2007-09-06 49.14 +1.7% 2007-11-05 54.07 +11.9% 2008-02-06 42.81 −11.4%
26 Global financial crisis 2007-11-09 50.00 2007-11-23 9 49.84 −0.3% −9.4% 2007-12-24 52.24 +4.8% 2008-02-26 44.07 −11.6% 2008-05-27 49.05 −1.6%
27 Global financial crisis 2008-01-07 48.17 2008-01-14 5 47.87 −0.6% −13.0% 2008-02-13 44.78 −6.5% 2008-04-15 44.14 −7.8% 2008-07-15 44.24 −7.6%
28 Global financial crisis 2008-01-22 44.18 2008-01-24 2 44.91 +1.7% −18.4% 2008-02-25 43.90 −2.2% 2008-04-24 47.27 +5.3% 2008-07-24 44.69 −0.5%
29 Global financial crisis 2008-03-10 41.26 2008-03-11 1 42.83 +3.8% −22.2% 2008-04-10 45.54 +6.3% 2008-06-10 48.54 +13.3% 2008-09-09 42.45 −0.9%
30 Global financial crisis 2008-09-04 43.66 2008-09-11 5 43.60 −0.1% −20.8% 2008-10-10 31.32 −28.2% 2008-12-10 29.91 −31.4% 2009-03-13 28.74 −34.1%
31 Global financial crisis 2008-09-17 40.21 2008-09-18 1 41.57 +3.4% −24.5% 2008-10-17 32.30 −22.3% 2008-12-17 30.19 −27.4% 2009-03-20 29.18 −29.8%
32 Global financial crisis 2008-10-06 34.86 2008-10-13 5 35.13 +0.8% −36.2% 2008-11-11 30.10 −14.3% 2009-01-13 29.53 −15.9% 2009-04-15 32.40 −7.8%
33 Global financial crisis 2008-11-20 25.56 2008-11-24 2 28.30 +10.7% −46.1% 2008-12-24 29.14 +3.0% 2009-02-26 27.79 −1.8% 2009-05-28 34.95 +23.5%
34 — 2009-10-28 41.39 2009-11-05 6 42.35 +2.3% −2.2% 2009-12-07 43.91 +3.7% 2010-02-08 42.67 +0.8% 2010-05-10 47.77 +12.8%
35 — 2010-01-29 42.79 2010-02-02 2 43.65 +2.0% −6.3% 2010-03-04 45.75 +4.8% 2010-05-04 48.43 +11.0% 2010-08-03 46.47 +6.5%
36 2010 flash crash / euro debt 2010-05-07 45.41 2010-05-10 1 47.77 +5.2% −5.4% 2010-06-09 43.82 −8.3% 2010-08-09 47.08 −1.4% 2010-11-05 53.67 +12.4%
37 2010 flash crash / euro debt 2010-06-29 43.37 2010-07-07 5 43.96 +1.4% −13.0% 2010-08-05 46.83 +6.5% 2010-10-05 49.66 +13.0% 2011-01-04 55.27 +25.7%
38 — 2011-03-16 54.15 2011-03-21 3 55.50 +2.5% −5.7% 2011-04-19 56.60 +2.0% 2011-06-20 54.07 −2.6% 2011-09-19 56.61 +2.0%
39 — 2011-06-10 54.64 2011-06-14 2 55.34 +1.3% −6.6% 2011-07-14 57.09 +3.2% 2011-09-13 54.58 −1.4% 2011-12-12 56.38 +1.9%
40 2011 US downgrade / euro debt 2011-08-04 54.17 2011-08-09 3 53.03 −2.1% −11.1% 2011-09-08 54.39 +2.6% 2011-11-07 58.21 +9.8% 2012-02-08 62.46 +17.8%
41 — 2011-11-25 52.88 2011-11-28 1 54.72 +3.5% −8.2% 2011-12-28 55.59 +1.6% 2012-02-29 64.41 +17.7% 2012-05-30 62.33 +13.9%
42 2012-05 euro debt (Greece) 2012-05-14 63.58 2012-05-21 5 62.51 −1.7% −8.4% 2012-06-20 64.29 +2.8% 2012-08-20 68.42 +9.5% 2012-11-20 63.80 +2.1%
43 — 2012-10-11 66.71 2012-10-16 3 68.12 +2.1% −3.2% 2012-11-16 62.30 −8.5% 2013-01-18 67.07 −1.5% 2013-04-22 68.79 +1.0%
44 — 2012-11-14 62.24 2012-11-19 3 63.78 +2.5% −9.4% 2012-12-19 66.26 +3.9% 2013-02-21 66.47 +4.2% 2013-05-22 73.62 +15.4%
45 — 2014-10-16 91.79 2014-10-17 1 93.00 +1.3% −7.3% 2014-11-17 102.91 +10.7% 2015-01-20 101.62 +9.3% 2015-04-21 108.06 +16.2%
46 — 2014-12-16 99.98 2014-12-18 2 104.25 +4.3% −1.7% 2015-01-21 102.14 −2.0% 2015-03-23 108.32 +3.9% 2015-06-22 110.70 +6.2%
47 — 2015-01-06 100.07 2015-01-08 2 103.30 +3.2% −2.6% 2015-02-09 102.80 −0.5% 2015-04-10 107.75 +4.3% 2015-07-10 107.65 +4.2%
48 2015-16 yuan devaluation / oil crash 2015-08-21 102.40 2015-08-26 3 103.03 +0.6% −9.6% 2015-09-25 102.92 −0.1% 2015-11-24 114.05 +10.7% 2016-02-26 103.43 +0.4%
49 2015-16 yuan devaluation / oil crash 2015-09-28 99.99 2015-09-30 2 101.76 +1.8% −10.7% 2015-10-29 113.84 +11.9% 2015-12-30 113.27 +11.3% 2016-04-01 110.36 +8.5%
50 2015-16 yuan devaluation / oil crash 2015-11-13 109.84 2015-11-18 3 113.63 +3.5% −1.2% 2015-12-18 109.83 −3.3% 2016-02-22 103.28 −9.1% 2016-05-20 106.47 −6.3%
51 2015-16 yuan devaluation / oil crash 2016-01-07 104.87 2016-01-14 5 104.07 −0.8% −9.6% 2016-02-16 100.23 −3.7% 2016-04-15 110.64 +6.3% 2016-07-15 111.80 +7.4%
52 — 2016-04-28 106.28 2016-05-09 7 105.88 −0.4% −8.1% 2016-06-08 110.37 +4.2% 2016-08-08 116.65 +10.2% 2016-11-04 113.65 +7.3%
53 2016-06 Brexit vote 2016-06-17 106.49 2016-06-23 4 108.77 +2.1% −5.5% 2016-07-25 113.66 +4.5% 2016-09-22 119.09 +9.5% 2016-12-21 120.46 +10.7%
54 2016-11 pre-US election 2016-11-02 115.18 2016-11-07 3 116.35 +1.0% −2.7% 2016-12-07 118.36 +1.7% 2017-02-08 126.50 +8.7% 2017-05-10 138.40 +19.0%
55 2018-02 volatility blowup 2018-02-05 158.12 2018-02-12 5 158.87 +0.5% −7.1% 2018-03-14 171.68 +8.1% 2018-05-14 169.75 +6.8% 2018-08-13 180.32 +13.5%
56 2018-02 volatility blowup 2018-03-23 158.51 2018-04-04 7 159.74 +0.8% −8.2% 2018-05-03 161.80 +1.3% 2018-07-03 170.80 +6.9% 2018-10-02 185.75 +16.3%
57 2018 Q4 rate hikes + trade war 2018-10-08 179.05 2018-10-12 4 174.32 −2.6% −6.7% 2018-11-12 166.33 −4.6% 2019-01-15 162.38 −6.8% 2019-04-16 186.50 +7.0%
58 2018 Q4 rate hikes + trade war 2018-12-21 147.57 2018-12-26 2 152.46 +3.3% −18.4% 2019-01-28 163.11 +7.0% 2019-03-28 178.31 +17.0% 2019-06-27 186.50 +22.3%
59 2019-05 trade war escalation 2019-05-13 178.58 2019-05-15 2 183.09 +2.5% −4.2% 2019-06-14 182.64 −0.2% 2019-08-14 182.76 −0.2% 2019-11-12 201.54 +10.1%
60 2019-05 trade war escalation 2019-06-03 170.12 2019-06-04 1 174.91 +2.8% −8.5% 2019-07-03 191.44 +9.5% 2019-09-03 185.65 +6.1% 2019-12-02 203.00 +16.1%
61 2019-08 trade war / yield-curve inversion 2019-08-05 180.73 2019-08-08 3 188.26 +4.2% −3.6% 2019-09-09 191.19 +1.6% 2019-11-06 199.76 +6.1% 2020-02-07 229.20 +21.7%
62 COVID crash 2020-02-25 215.37 2020-03-02 4 216.42 +0.5% −8.7% 2020-03-31 190.40 −12.0% 2020-06-01 234.06 +8.2% 2020-08-28 292.53 +35.2%
63 — 2021-02-25 312.83 2021-03-01 2 323.59 +3.4% −3.8% 2021-03-30 314.32 −2.9% 2021-05-28 333.93 +3.2% 2021-08-27 376.04 +16.2%
64 — 2021-05-12 316.89 2021-05-14 2 326.39 +3.0% −4.6% 2021-06-15 342.26 +4.9% 2021-08-13 368.82 +13.0% 2021-11-11 390.59 +19.7%
65 — 2021-09-20 365.70 2021-09-23 3 372.98 +2.0% −2.4% 2021-10-22 374.10 +0.3% 2021-12-22 393.95 +5.6% 2022-03-24 359.65 −3.6%
66 — 2021-09-30 357.96 2021-10-05 3 357.38 −0.2% −6.5% 2021-11-03 393.18 +10.0% 2022-01-04 396.47 +10.9% 2022-04-05 361.10 +1.0%
67 2022 rate-hike bear market 2022-01-20 361.72 2022-01-28 6 351.80 −2.7% −12.9% 2022-03-01 341.49 −2.9% 2022-04-29 313.25 −11.0% 2022-08-01 315.27 −10.4%
68 2022 rate-hike bear market 2022-06-13 275.41 2022-06-21 5 281.08 +2.1% −30.4% 2022-07-21 307.38 +9.4% 2022-09-20 288.73 +2.7% 2022-12-19 269.75 −4.0%
69 2022 rate-hike bear market 2022-09-02 295.17 2022-09-07 2 298.97 +1.3% −26.0% 2022-10-06 279.76 −6.4% 2022-12-06 281.68 −5.8% 2023-03-09 292.66 −2.1%
70 2022 rate-hike bear market 2022-09-23 275.51 2022-09-28 3 279.94 +1.6% −30.7% 2022-10-27 272.87 −2.5% 2022-12-28 260.10 −7.1% 2023-03-30 315.68 +12.8%
71 2023-08 to 10 Treasury yield spike 2023-09-21 357.86 2023-09-29 6 358.27 +0.1% −7.1% 2023-10-30 349.20 −2.5% 2023-12-29 409.52 +14.3% 2024-04-02 441.11 +23.1%
72 — 2024-04-19 414.65 2024-04-23 2 425.07 +2.5% −4.8% 2024-05-22 455.71 +7.2% 2024-07-24 463.38 +9.0% 2024-10-22 495.96 +16.7%
73 2024-07 yen carry-trade unwind 2024-07-25 458.27 2024-07-31 4 471.07 +2.8% −6.3% 2024-08-29 470.66 −0.1% 2024-10-29 500.16 +6.2% 2025-01-31 522.29 +10.9%
74 2025 tariff selloff 2025-03-11 471.60 2025-03-14 3 479.66 +1.7% −11.1% 2025-04-14 457.48 −4.6% 2025-06-13 526.96 +9.9% 2025-09-15 591.68 +23.4%
75 2025 tariff selloff 2025-04-04 422.67 2025-04-09 3 466.00 +10.3% −13.6% 2025-05-09 487.97 +4.7% 2025-07-11 554.20 +18.9% 2025-10-09 610.70 +31.1%
76 — 2026-03-27 562.58 2026-03-31 2 577.18 +2.6% −9.2% 2026-04-30 667.74 +15.7% 2026-07-01 725.17 +25.6% 2026-09-30 739.77 +28.2%
77 — 2026-07-29 661.73 2026-07-30 1 683.55 +3.3% −8.4% 2026-08-28 716.43 +4.8% N/A N/A N/A N/A N/A N/A

QQQ Table Two: 6-month drawdown after entry and relative-low determination

# Signal date T T close 6M total return (incl. dividends) 6M lowest close (date) 6M max drawdown Window low (T−20..T+126) Entry above window low Relative low Percentile (0=lowest) D0-entry 6M return Cluster
1 2000-04-17 89.62 −8.5% 74.81 (2000-05-23) −16.5% 74.81 +19.8% No 24.3 −14.6%
2 2000-08-03 90.44 −31.9% 53.44 (2001-01-02) −40.9% 53.44 +69.2% No 60.8 −22.9%
3 2000-10-13 81.25 −49.2% 34.05 (2001-04-04) −58.1% 34.05 +138.6% No 63.0 −47.6%
4 2001-02-15 58.40 −32.3% 34.05 (2001-04-04) −41.7% 34.05 +71.5% No 72.5 −28.6%
5 2001-03-05 47.55 −23.0% 34.05 (2001-04-04) −28.4% 34.05 +39.6% No 58.2 −26.1% Y (11 days after prior)
6 2001-03-13 44.45 −29.8% 31.20 (2001-09-17) −29.8% 31.20 +42.5% No 46.0 −19.4% Y (6 days after prior)
7 2001-04-05 37.31 −12.9% 28.19 (2001-09-21) −24.4% 28.19 +32.4% No 15.9 −8.1% Y (17 days after prior)
8 2001-09-24 29.61 +21.2% 28.48 (2001-09-27) −3.8% 28.19 +5.0% No 3.7 +8.7%
9 2002-02-08 36.17 −35.5% 21.44 (2002-08-05) −40.7% 21.44 +68.7% No 57.7 −33.0%
10 2002-05-08 31.77 −17.8% 20.06 (2002-10-09) −36.9% 20.06 +58.4% No 67.7 −22.7%
11 2002-06-17 28.56 −12.3% 20.06 (2002-10-09) −29.8% 20.06 +42.4% No 69.3 −8.2%
12 2002-09-25 21.87 +20.6% 20.06 (2002-10-09) −8.3% 20.06 +9.0% No 7.9 +25.6%
13 2004-03-12 35.51 +0.2% 32.47 (2004-08-12) −8.6% 32.47 +9.4% No 40.7 +0.7%
14 2004-07-20 35.43 +10.3% 32.47 (2004-08-12) −8.4% 32.47 +9.1% No 34.9 +8.7%
15 2004-08-16 32.89 +16.2% 33.26 (2004-08-17) 0.0% 32.47 +1.3% Yes 2.1 +15.4% Y (19 days after prior)
16 2005-01-12 38.58 +0.6% 34.70 (2005-04-20) −10.1% 34.70 +11.2% No 79.4 −3.8%
17 2005-01-26 37.15 +7.0% 34.70 (2005-04-20) −6.6% 34.70 +7.1% No 34.9 +7.5% Y (9 days after prior)
18 2005-04-21 35.62 +8.7% 34.75 (2005-04-28) −2.4% 34.70 +2.7% Yes 6.9 +8.8%
19 2005-07-05 37.11 +11.6% 36.93 (2005-07-06) −0.5% 36.73 +1.0% Yes 20.1 +12.0%
20 2005-10-14 38.06 +11.9% 37.91 (2005-10-18) −0.4% 37.49 +1.5% Yes 2.6 +12.3%
21 2006-05-25 39.36 +13.8% 35.70 (2006-07-21) −9.3% 35.70 +10.3% No 40.2 +6.6%
22 2006-06-15 38.65 +13.7% 35.70 (2006-07-21) −7.6% 35.70 +8.3% No 28.6 +17.4% Y (14 days after prior)
23 2006-07-19 36.62 +20.8% 35.70 (2006-07-21) −2.5% 35.70 +2.6% Yes 6.9 +26.0%
24 2007-03-08 42.97 +14.5% 42.37 (2007-03-13) −1.4% 42.15 +1.9% Yes 4.8 +13.8%
25 2007-08-07 48.31 −11.2% 42.81 (2008-02-06) −11.4% 42.81 +12.8% No 43.9 −7.6%
26 2007-11-23 49.84 −1.4% 41.26 (2008-03-10) −17.2% 41.26 +20.8% No 63.0 −1.6%
27 2008-01-14 47.87 −7.4% 41.26 (2008-03-10) −13.8% 41.26 +16.0% No 50.3 −4.6%
28 2008-01-24 44.91 −0.3% 41.26 (2008-03-10) −8.1% 41.26 +8.8% No 31.2 +1.4% Y (7 days after prior)
29 2008-03-11 42.83 −0.7% 41.48 (2008-03-17) −3.2% 41.26 +3.8% No 5.3 +5.1%
30 2008-09-11 43.60 −33.9% 25.56 (2008-11-20) −41.4% 25.56 +70.6% No 68.8 −39.8%
31 2008-09-18 41.57 −29.5% 25.56 (2008-11-20) −38.5% 25.56 +62.6% No 66.7 −26.2% Y (5 days after prior)
32 2008-10-13 35.13 −7.5% 25.56 (2008-11-20) −27.2% 25.56 +37.4% No 69.3 −9.9% Y (17 days after prior)
33 2008-11-24 28.30 +23.9% 25.74 (2009-03-09) −9.0% 25.56 +10.7% No 10.6 +36.1%
34 2009-11-05 42.35 +13.1% 42.60 (2009-11-06) 0.0% 40.96 +3.4% No 23.3 +19.0%
35 2010-02-02 43.65 +7.0% 42.47 (2010-07-02) −2.7% 42.47 +2.8% Yes 13.2 +7.1%
36 2010-05-10 47.77 +13.0% 42.47 (2010-07-02) −11.1% 42.47 +12.5% No 61.9 +18.2%
37 2010-07-07 43.96 +26.3% 43.46 (2010-08-31) −1.1% 42.47 +3.5% No 5.3 +26.1%
38 2011-03-21 55.50 +2.4% 50.03 (2011-08-19) −9.9% 50.03 +10.9% No 31.7 +2.2%
39 2011-06-14 55.34 +2.3% 50.03 (2011-08-19) −9.6% 50.03 +10.6% No 30.7 +2.7%
40 2011-08-09 53.03 +18.5% 50.03 (2011-08-19) −5.7% 50.03 +6.0% No 6.3 +14.5%
41 2011-11-28 54.72 +14.5% 54.32 (2011-12-19) −0.7% 52.88 +3.5% No 13.2 +18.9%
42 2012-05-21 62.51 +2.6% 60.41 (2012-06-01) −3.4% 60.41 +3.5% No 10.6 −0.9%
43 2012-10-16 68.12 +1.8% 62.03 (2012-11-15) −8.9% 62.03 +9.8% No 67.2 +2.1%
44 2012-11-19 63.78 +16.3% 63.78 (2012-12-28) 0.0% 62.03 +2.8% Yes 3.7 +19.4%
45 2014-10-17 93.00 +16.9% 94.39 (2014-10-20) 0.0% 91.79 +1.3% Yes 2.1 +17.2%
46 2014-12-18 104.25 +7.1% 99.65 (2015-01-15) −4.4% 99.65 +4.6% No 41.8 +10.7%
47 2015-01-08 103.30 +4.7% 99.65 (2015-01-15) −3.5% 99.65 +3.7% No 29.6 +6.0% Y (13 days after prior)
48 2015-08-26 103.03 +0.9% 96.32 (2016-02-09) −6.5% 96.32 +7.0% No 17.5 −0.8%
49 2015-09-30 101.76 +9.1% 96.32 (2016-02-09) −5.3% 96.32 +5.6% No 11.6 +9.4%
50 2015-11-18 113.63 −5.7% 96.32 (2016-02-09) −15.2% 96.32 +18.0% No 88.9 −4.0%
51 2016-01-14 104.07 +8.0% 96.32 (2016-02-09) −7.4% 96.32 +8.0% No 16.4 +5.2%
52 2016-05-09 105.88 +7.9% 102.22 (2016-06-27) −3.5% 102.22 +3.6% No 18.5 +11.4%
53 2016-06-23 108.77 +11.4% 102.22 (2016-06-27) −6.0% 102.22 +6.4% No 21.7 +13.1%
54 2016-11-07 116.35 +19.6% 114.63 (2016-11-14) −1.5% 113.65 +2.4% Yes 7.4 +19.4%
55 2018-02-12 158.87 +13.9% 155.51 (2018-04-02) −2.1% 153.45 +3.5% No 24.9 +14.6%
56 2018-04-04 159.74 +16.7% 156.63 (2018-04-06) −1.9% 155.51 +2.7% Yes 7.9 +15.9%
57 2018-10-12 174.32 +7.5% 143.50 (2018-12-24) −17.7% 143.50 +21.5% No 49.7 +3.6%
58 2018-12-26 152.46 +22.8% 149.82 (2019-01-03) −1.7% 143.50 +6.2% No 2.1 +25.3%
59 2019-05-15 183.09 +10.5% 170.12 (2019-06-03) −7.1% 170.12 +7.6% No 30.2 +12.7%
60 2019-06-04 174.91 +16.6% 176.21 (2019-06-05) 0.0% 170.12 +2.8% Yes 3.7 +20.6% Y (13 days after prior)
61 2019-08-08 188.26 +22.3% 182.07 (2019-08-23) −3.3% 180.73 +4.2% No 31.7 +25.9%
62 2020-03-02 216.42 +35.7% 169.30 (2020-03-16) −21.8% 169.30 +27.8% No 37.6 +31.7%
63 2021-03-01 323.59 +16.5% 299.94 (2021-03-08) −7.3% 299.94 +7.9% No 37.0 +19.8%
64 2021-05-14 326.39 +19.9% 322.22 (2021-05-18) −1.3% 316.89 +3.0% Yes 19.0 +24.7%
65 2021-09-23 372.98 −3.3% 318.17 (2022-03-14) −14.7% 318.17 +17.2% No 57.1 −4.3%
66 2021-10-05 357.38 +1.3% 318.17 (2022-03-14) −11.0% 318.17 +12.3% No 20.6 +1.3% Y (8 days after prior)
67 2022-01-28 351.80 −10.1% 271.39 (2022-06-16) −22.9% 271.39 +29.6% No 56.1 −16.5%
68 2022-06-21 281.08 −3.6% 260.49 (2022-11-03) −7.3% 260.49 +7.9% No 22.2 +3.7%
69 2022-09-07 298.97 −1.7% 260.10 (2022-12-28) −13.0% 260.10 +14.9% No 72.5 +0.4%
70 2022-09-28 279.94 +13.2% 260.10 (2022-12-28) −7.1% 260.10 +7.6% No 26.5 +12.1% Y (15 days after prior)
71 2023-09-29 358.27 +23.6% 343.66 (2023-10-26) −4.1% 343.66 +4.3% No 7.9 +24.7%
72 2024-04-23 425.07 +17.0% 421.52 (2024-05-01) −0.8% 414.65 +2.5% Yes 5.3 +19.2%
73 2024-07-31 471.07 +11.2% 434.77 (2024-08-07) −7.7% 434.77 +8.3% No 28.0 +12.2%
74 2025-03-14 479.66 +23.7% 416.06 (2025-04-08) −13.3% 416.06 +15.3% No 15.3 +23.1%
75 2025-04-09 466.00 +31.4% 433.11 (2025-04-21) −7.1% 416.06 +12.0% No 6.9 +43.8% Y (18 days after prior)
76 2026-03-31 577.18 +28.4% 584.31 (2026-04-01) 0.0% 558.28 +3.4% No 1.6 +30.9%
77 2026-07-30 683.55 N/A 687.99 (2026-07-31, under 6M, only 55 days) 0.0% (under 6M) N/A N/A N/A N/A N/A

VOO complete event tables (main definition, 47 events)

The two tables below list all 47 events one by one from the CSV file events_VOO_P_RSI6lt20_W10.csv, and the count matches the CSV; column definitions are the same as in the QQQ tables. All 47 events have complete 6-month forward data.

VOO Table One: signal date, entry price and prices 1/3/6 months later

# Context D0 (first RSI6<20) D0 close Signal date T D0→T days T close T vs D0 From 252-day high +1M date +1M close +1M return +3M date +3M close +3M return +6M date +6M close +6M return
1 — 2011-06-08 117.40 2011-06-14 4 118.22 +0.7% −5.3% 2011-07-14 119.80 +1.3% 2011-09-13 107.64 −8.9% 2011-12-12 113.64 −3.9%
2 2011 US downgrade / euro debt 2011-08-02 114.84 2011-08-09 5 107.34 −6.5% −14.0% 2011-09-08 108.86 +1.4% 2011-11-07 115.46 +7.6% 2012-02-08 123.60 +15.1%
3 — 2011-11-23 106.62 2011-11-28 2 109.48 +2.7% −12.3% 2011-12-28 114.30 +4.4% 2012-02-29 125.18 +14.3% 2012-05-30 120.54 +10.1%
4 — 2012-04-10 124.26 2012-04-12 2 127.00 +2.2% −2.1% 2012-05-11 124.02 −2.3% 2012-07-12 122.10 −3.9% 2012-10-10 131.18 +3.3%
5 2012-05 euro debt (Greece) 2012-05-14 122.70 2012-05-21 5 120.72 −1.6% −7.0% 2012-06-20 124.60 +3.2% 2012-08-20 130.12 +7.8% 2012-11-20 127.46 +5.6%
6 — 2012-11-14 124.62 2012-11-19 3 127.32 +2.2% −5.5% 2012-12-19 132.10 +3.8% 2013-02-21 137.88 +8.3% 2013-05-22 152.06 +19.4%
7 — 2013-08-19 150.96 2013-08-22 3 152.10 +0.8% −2.9% 2013-09-23 155.74 +2.4% 2013-11-20 163.55 +7.5% 2014-02-24 169.37 +11.4%
8 — 2014-01-27 163.08 2014-02-06 8 162.57 −0.3% −3.9% 2014-03-10 172.45 +6.1% 2014-05-08 171.98 +5.8% 2014-08-07 175.13 +7.7%
9 — 2014-07-31 176.96 2014-08-08 6 177.16 +0.1% −2.7% 2014-09-09 182.78 +3.2% 2014-11-06 186.34 +5.2% 2015-02-09 187.66 +5.9%
10 — 2014-12-15 182.93 2014-12-17 2 185.18 +1.2% −2.9% 2015-01-20 185.26 0.0% 2015-03-20 193.85 +4.7% 2015-06-19 193.85 +4.7%
11 — 2015-01-06 183.27 2015-01-08 2 188.82 +3.0% −1.4% 2015-02-09 187.66 −0.6% 2015-04-10 192.54 +2.0% 2015-07-10 190.24 +0.8%
12 — 2015-06-29 188.35 2015-07-07 5 190.75 +1.3% −2.5% 2015-08-05 192.64 +1.0% 2015-10-05 181.96 −4.6% 2016-01-05 184.64 −3.2%
13 2015-16 yuan devaluation / oil crash 2015-08-21 181.04 2015-08-26 3 178.36 −1.5% −8.9% 2015-09-25 176.76 −0.9% 2015-11-24 192.02 +7.7% 2016-02-26 178.98 +0.3%
14 2015-16 yuan devaluation / oil crash 2015-09-28 172.31 2015-09-30 2 175.71 +2.0% −10.2% 2015-10-29 191.53 +9.0% 2015-12-30 188.80 +7.4% 2016-04-01 189.77 +8.0%
15 2015-16 yuan devaluation / oil crash 2015-11-13 185.67 2015-11-18 3 191.32 +3.0% −2.2% 2015-12-18 184.40 −3.6% 2016-02-22 178.59 −6.7% 2016-05-20 188.49 −1.5%
16 2015-16 yuan devaluation / oil crash 2016-01-07 177.86 2016-01-14 5 175.92 −1.1% −10.1% 2016-02-16 174.01 −1.1% 2016-04-15 190.56 +8.3% 2016-07-15 198.01 +12.6%
17 — 2016-09-09 195.68 2016-09-15 4 196.66 +0.5% −2.2% 2016-10-14 195.63 −0.5% 2016-12-14 207.46 +5.5% 2017-03-17 218.56 +11.1%
18 2016-11 pre-US election 2016-11-02 192.49 2016-11-07 3 195.68 +1.7% −2.7% 2016-12-07 206.20 +5.4% 2017-02-08 210.52 +7.6% 2017-05-10 220.20 +12.5%
19 — 2017-04-13 213.47 2017-04-17 1 215.34 +0.9% −2.2% 2017-05-16 220.48 +2.4% 2017-07-17 225.50 +4.7% 2017-10-13 234.14 +8.7%
20 — 2017-08-10 223.93 2017-08-14 2 226.43 +1.1% −0.5% 2017-09-13 229.85 +1.5% 2017-11-10 237.04 +4.7% 2018-02-13 244.56 +8.0%
21 2018-02 volatility blowup 2018-02-05 242.79 2018-02-12 5 243.74 +0.4% −7.4% 2018-03-14 252.97 +3.8% 2018-05-14 250.75 +2.9% 2018-08-13 259.14 +6.3%
22 2018-02 volatility blowup 2018-03-22 243.27 2018-03-29 5 242.08 −0.5% −8.1% 2018-04-30 242.92 +0.3% 2018-06-28 248.88 +2.8% 2018-09-27 267.08 +10.3%
23 2018 Q4 rate hikes + trade war 2018-10-10 255.60 2018-10-16 4 257.58 +0.8% −4.5% 2018-11-14 248.28 −3.6% 2019-01-17 241.67 −6.2% 2019-04-18 266.34 +3.4%
24 2018 Q4 rate hikes + trade war 2018-10-24 243.78 2018-10-25 1 248.15 +1.8% −8.0% 2018-11-26 245.80 −0.9% 2019-01-29 242.13 −2.4% 2019-04-30 270.01 +8.8%
25 2018 Q4 rate hikes + trade war 2018-12-17 233.23 2018-12-26 6 225.95 −3.1% −16.2% 2019-01-28 242.43 +7.3% 2019-03-28 257.88 +14.1% 2019-06-27 267.67 +18.5%
26 2019-05 trade war escalation 2019-05-13 258.03 2019-05-16 3 264.21 +2.4% −2.2% 2019-06-17 265.86 +0.6% 2019-08-15 261.39 −1.1% 2019-11-13 283.93 +7.5%
27 2019-05 trade war escalation 2019-06-03 252.12 2019-06-04 1 257.67 +2.2% −4.6% 2019-07-03 274.32 +6.5% 2019-09-03 267.11 +3.7% 2019-12-02 286.30 +11.1%
28 2019-08 trade war / yield-curve inversion 2019-08-05 260.88 2019-08-08 3 269.58 +3.3% −2.8% 2019-09-09 273.81 +1.6% 2019-11-06 282.11 +4.6% 2020-02-07 305.33 +13.3%
29 — 2019-10-02 264.57 2019-10-04 2 270.37 +2.2% −2.5% 2019-11-04 282.32 +4.4% 2020-01-06 297.35 +10.0% 2020-04-06 243.53 −9.9%
30 COVID crash 2020-02-24 296.24 2020-03-02 5 283.87 −4.2% −8.7% 2020-03-31 236.82 −16.6% 2020-06-01 280.97 −1.0% 2020-08-28 322.15 +13.5%
31 COVID crash 2020-03-12 226.48 2020-03-13 1 247.64 +9.3% −20.4% 2020-04-14 260.92 +5.4% 2020-06-12 279.65 +12.9% 2020-09-11 306.98 +24.0%
32 2020-09 to 10 correction / pre-US election 2020-10-28 300.07 2020-11-03 4 308.66 +2.9% −6.1% 2020-12-03 336.91 +9.2% 2021-02-04 355.02 +15.0% 2021-05-06 385.28 +24.8%
33 — 2021-09-20 400.43 2021-09-23 3 408.78 +2.1% −1.9% 2021-10-22 416.59 +1.9% 2021-12-22 429.99 +5.2% 2022-03-24 414.02 +1.3%
34 2022 rate-hike bear market 2022-01-19 415.33 2022-01-28 7 406.26 −2.2% −7.5% 2022-03-01 395.34 −2.7% 2022-04-29 378.70 −6.8% 2022-08-01 377.65 −7.0%
35 2022 rate-hike bear market 2022-04-26 382.54 2022-04-28 2 393.14 +2.8% −10.5% 2022-05-27 381.80 −2.9% 2022-07-29 378.79 −3.7% 2022-10-27 349.23 −11.2%
36 2022 rate-hike bear market 2022-06-13 344.46 2022-06-21 5 346.23 +0.5% −21.2% 2022-07-21 366.59 +5.9% 2022-09-20 354.55 +2.4% 2022-12-19 350.81 +1.3%
37 2022 rate-hike bear market 2022-09-06 359.24 2022-09-07 1 365.68 +1.8% −16.7% 2022-10-06 342.96 −6.2% 2022-12-06 362.03 −1.0% 2023-03-09 359.84 −1.6%
38 2022 rate-hike bear market 2022-09-23 339.63 2022-09-28 3 340.48 +0.3% −22.5% 2022-10-27 349.23 +2.6% 2022-12-28 346.17 +1.7% 2023-03-30 370.89 +8.9%
39 2022 rate-hike bear market 2022-09-30 328.30 2022-10-03 1 336.98 +2.6% −23.3% 2022-11-01 353.52 +4.9% 2023-01-03 349.99 +3.9% 2023-04-04 375.43 +11.4%
40 2023-08 to 10 Treasury yield spike 2023-08-17 400.72 2023-08-23 4 406.95 +1.6% −3.3% 2023-09-22 396.96 −2.5% 2023-11-21 416.34 +2.3% 2024-02-23 466.78 +14.7%
41 2023-08 to 10 Treasury yield spike 2023-09-22 396.96 2023-10-06 10 394.56 −0.6% −6.2% 2023-11-06 400.21 +1.4% 2024-01-08 436.13 +10.5% 2024-04-09 477.27 +21.0%
42 2023-08 to 10 Treasury yield spike 2023-10-27 377.32 2023-10-30 1 381.86 +1.2% −9.2% 2023-11-29 417.78 +9.4% 2024-01-31 443.82 +16.2% 2024-05-01 459.93 +20.4%
43 — 2024-04-18 458.94 2024-04-23 3 464.84 +1.3% −3.4% 2024-05-22 487.06 +4.8% 2024-07-24 497.29 +7.0% 2024-10-22 536.16 +15.3%
44 2024-07 yen carry-trade unwind 2024-08-05 475.20 2024-08-08 3 487.73 +2.6% −6.0% 2024-09-09 502.23 +3.0% 2024-11-06 543.30 +11.4% 2025-02-10 556.15 +14.0%
45 — 2024-12-18 539.14 2024-12-24 4 552.82 +2.5% −1.1% 2025-01-28 555.85 +0.5% 2025-03-28 510.80 −7.6% 2025-06-30 568.03 +2.8%
46 2025 tariff selloff 2025-04-04 465.52 2025-04-09 3 499.10 +7.2% −11.5% 2025-05-09 518.65 +3.9% 2025-07-11 573.22 +14.9% 2025-10-09 617.10 +23.6%
47 — 2026-03-27 582.96 2026-03-31 2 597.55 +2.5% −6.6% 2026-04-30 660.58 +10.5% 2026-07-01 685.46 +14.7% 2026-09-30 700.86 +17.3%

VOO Table Two: 6-month drawdown after entry and relative-low determination

# Signal date T T close 6M total return (incl. dividends) 6M lowest close (date) 6M max drawdown Window low (T−20..T+126) Entry above window low Relative low Percentile (0=lowest) D0-entry 6M return Cluster
1 2011-06-14 118.22 −2.9% 100.34 (2011-10-03) −15.1% 100.34 +17.8% No 58.7 −1.6%
2 2011-08-09 107.34 +16.5% 100.34 (2011-10-03) −6.5% 100.34 +7.0% No 13.2 +5.5%
3 2011-11-28 109.48 +11.3% 109.80 (2011-11-29) 0.0% 106.30 +3.0% Yes 13.2 +13.4%
4 2012-04-12 127.00 +4.4% 117.26 (2012-06-04) −7.7% 117.26 +8.3% No 60.8 +7.3%
5 2012-05-21 120.72 +6.7% 117.26 (2012-06-04) −2.9% 117.26 +3.0% Yes 4.8 +2.8%
6 2012-11-19 127.32 +20.9% 127.46 (2012-11-20) 0.0% 124.14 +2.6% Yes 3.7 +22.7%
7 2013-08-22 152.10 +12.5% 149.62 (2013-08-27) −1.6% 149.62 +1.7% Yes 24.3 +11.1%
8 2014-02-06 162.57 +8.7% 164.67 (2014-02-07) 0.0% 159.61 +1.9% Yes 5.3 +11.2%
9 2014-08-08 177.16 +6.4% 170.72 (2014-10-16) −3.6% 170.72 +3.8% No 16.9 +3.4%
10 2014-12-17 185.18 +5.2% 182.54 (2015-01-15) −1.4% 181.52 +2.0% Yes 23.8 +5.5%
11 2015-01-08 188.82 +1.7% 182.54 (2015-01-15) −3.3% 181.52 +4.0% No 39.2 +2.3% Y (14 days after prior)
12 2015-07-07 190.75 −2.1% 171.55 (2015-08-25) −10.1% 171.55 +11.2% No 47.1 −0.1%
13 2015-08-26 178.36 +1.5% 167.70 (2016-02-11) −6.0% 167.70 +6.4% No 23.3 −2.6%
14 2015-09-30 175.71 +9.2% 167.70 (2016-02-11) −4.6% 167.70 +4.8% No 13.2 +9.7%
15 2015-11-18 191.32 −0.4% 167.70 (2016-02-11) −12.3% 167.70 +14.1% No 87.8 +1.2%
16 2016-01-14 175.92 +13.7% 167.70 (2016-02-11) −4.7% 167.70 +4.9% No 10.6 +9.7%
17 2016-09-15 196.66 +11.8% 191.48 (2016-11-04) −2.6% 191.48 +2.7% Yes 21.7 +11.6%
18 2016-11-07 195.68 +13.8% 196.58 (2016-11-08) 0.0% 191.48 +2.2% Yes 5.8 +14.3%
19 2017-04-17 215.34 +9.8% 214.28 (2017-04-19) −0.5% 213.47 +0.9% Yes 15.9 +9.6%
20 2017-08-14 226.43 +9.1% 222.87 (2017-08-18) −1.6% 222.87 +1.6% Yes 31.2 +7.4%
21 2018-02-12 243.74 +7.3% 236.48 (2018-04-02) −3.0% 236.48 +3.1% No 22.8 +7.8%
22 2018-03-29 242.08 +11.3% 236.48 (2018-04-02) −2.3% 236.48 +2.4% Yes 6.3 +10.9%
23 2018-10-16 257.58 +4.5% 215.07 (2018-12-24) −16.5% 215.07 +19.8% No 56.1 +4.3%
24 2018-10-25 248.15 +10.0% 215.07 (2018-12-24) −13.3% 215.07 +15.4% No 25.4 +10.7% Y (7 days after prior)
25 2018-12-26 225.95 +19.8% 224.50 (2019-01-03) −0.6% 215.07 +5.1% No 2.1 +15.4%
26 2019-05-16 264.21 +8.5% 252.12 (2019-06-03) −4.6% 252.12 +4.8% No 32.3 +10.0%
27 2019-06-04 257.67 +12.2% 259.85 (2019-06-05) 0.0% 252.12 +2.2% Yes 7.4 +14.5% Y (12 days after prior)
28 2019-08-08 269.58 +14.4% 260.57 (2019-08-14) −3.3% 260.57 +3.5% No 32.3 +15.9%
29 2019-10-04 270.37 −9.1% 204.27 (2020-03-23) −24.4% 204.27 +32.4% No 27.0 −12.5%
30 2020-03-02 283.87 +14.6% 204.27 (2020-03-23) −28.0% 204.27 +39.0% No 38.6 +5.3%
31 2020-03-13 247.64 +24.6% 204.27 (2020-03-23) −17.5% 204.27 +21.2% No 9.5 +35.4% Y (9 days after prior)
32 2020-11-03 308.66 +25.8% 315.61 (2020-11-04) 0.0% 299.82 +2.9% Yes 12.7 +27.8%
33 2021-09-23 408.78 +2.3% 382.70 (2022-03-08) −6.4% 382.70 +6.8% No 45.0 +2.4%
34 2022-01-28 406.26 −6.3% 337.07 (2022-06-16) −17.0% 337.07 +20.5% No 53.4 −11.7%
35 2022-04-28 393.14 −10.4% 327.64 (2022-10-12) −16.7% 327.64 +20.0% No 70.4 −7.5%
36 2022-06-21 346.23 +2.2% 327.64 (2022-10-12) −5.4% 327.64 +5.7% No 15.3 +6.5%
37 2022-09-07 365.68 −0.7% 327.64 (2022-10-12) −10.4% 327.64 +11.6% No 61.9 +2.0%
38 2022-09-28 340.48 +9.9% 327.64 (2022-10-12) −3.8% 327.64 +3.9% No 7.9 +7.2% Y (15 days after prior)
39 2022-10-03 336.98 +12.4% 327.64 (2022-10-12) −2.8% 327.64 +2.9% Yes 5.8 +15.0% Y (3 days after prior)
40 2023-08-23 406.95 +15.6% 377.32 (2023-10-27) −7.3% 377.32 +7.9% No 40.7 +14.6%
41 2023-10-06 394.56 +21.9% 377.32 (2023-10-27) −4.4% 377.32 +4.6% No 9.5 +20.4%
42 2023-10-30 381.86 +21.3% 384.17 (2023-10-31) 0.0% 377.32 +1.2% Yes 1.1 +22.3% Y (16 days after prior)
43 2024-04-23 464.84 +16.1% 459.93 (2024-05-01) −1.1% 455.10 +2.1% Yes 16.9 +16.6%
44 2024-08-08 487.73 +14.7% 489.82 (2024-08-09) 0.0% 475.20 +2.6% Yes 11.1 +16.9%
45 2024-12-24 552.82 +3.4% 456.74 (2025-04-08) −17.4% 456.74 +21.0% No 77.8 +3.8%
46 2025-04-09 499.10 +24.4% 472.37 (2025-04-21) −5.4% 456.74 +9.3% No 6.9 +32.6%
47 2026-03-31 597.55 +17.9% 602.30 (2026-04-01) 0.0% 580.93 +2.9% Yes 1.6 +20.7%

Limitations and the latest status as of 2026-10-16

Overlapping holding periods, same-day signals and ignored costs all overstate the strength of the evidence

This backtest has four limitations that must be stated plainly. First, the samples are small and overlap: QQQ has 77 events and VOO 47, and the 126-day holding periods often overlap (for example, 3 signals in a row in 2001-03 and 2 to 3 in 2022-09), so the events are not independent samples; the t-statistics in the text are rough estimates and the true uncertainty is larger. Second, the two funds' signals are highly synchronized: 24 of VOO's 47 signals fired on the same day as QQQ (for example 2015-08-26, 2018-02-12, 2020-03-02, 2022-01-28, 2025-04-09 and 2026-03-31), and 24 of QQQ's 40 post-2010 signals also fell on the same day as VOO's, so the two funds cannot be counted as double the independent evidence. Third, commissions, slippage and taxes are not included, and entry assumes execution at the signal-day close; in practice the crossover is only confirmed at the close, so judging on intraday prices would differ, and dividend withholding tax on foreign-held ETFs would also make actual total returns lower than the TR in the tables. Fourth, there is a single data source: only Yahoo Finance, with no day-by-day cross-check against a second source; RSI is calculated on split-adjusted closes, and if your charting software uses dividend-adjusted prices or an SMA-based RSI, a few borderline events may enter or drop off the list (Wikipedia). In addition, the "down more than 20% from the 252-day high" grouping was cut after seeing the results and is an in-sample finding.

QQQ has entered a new oversold stretch but has not yet produced a crossover

As of the 2026-10-16 close, QQQ closed at 709.80 with RSI6 = 15.5 and RSI10 = 27.8, its first break below 20 since 2026-07-29 and a new oversold stretch. Under the main definition, RSI6 must cross above RSI10 within the next 10 trading days (roughly through 2026-10-30) for this to become signal #78; that has not happened yet, so it is not a signal. Based on the same Yahoo data, QQQ was about −6.6% from its 252-day high (the 2026-10-06 close of 759.66) that day, which falls in the "decline under 10%" group that historically had the highest 6-month win rate, but that is only a historical grouping and says nothing about this particular outcome. VOO closed at 701.61 the same day with RSI6 = 30.6, not yet oversold; its last signal was on 2026-03-31. QQQ #77 (2026-07-30) does not yet have complete 6-month forward data and is not included in the 3M/6M statistics. This article only describes historical statistics and is not investment advice.

Conclusion

What this signal really offers is not "the low" but "a discount": it gets you in at a relatively low price, at the cost of paying 1% to 2% more on average than on the first oversold day, and it cannot protect you from a subsequent large decline. For someone who wants to buy relative lows, the data carries two practical implications, both counterintuitive. First, if you have already decided to buy the dip, buying directly on the day RSI6 breaks below 20 has historically beaten waiting for the crossover in both funds. Second, what really decides success or failure is the broader environment, not the shape of RSI; the decline from the 52-week high on the signal day is more useful information than the crossover.

The difference between QQQ and VOO is also a reminder that the attractive post-2010 numbers were earned in a single long bull market. VOO has never been through a bear market on the scale of 2000–2002 or 2008, and its 85% win rate has not been tested in that kind of environment; the first half of QQQ's history clearly shows that the same signal, in a market down more than 20%, buys halfway down the mountain again and again. To actually use this signal in capital allocation, the next step should be to add a trend condition that can be judged in advance (such as the 200-day moving average or the decline from the high) and to test it out of sample on new events from 2026 onward, rather than relying only on the in-sample results of this backtest.

Frequently asked questions

If I wait for RSI6 to cross above RSI10 after RSI6 drops below 20, can I buy at the lowest point?

Most of the time, no. Using "an entry price within 3% of the lowest close from 20 days before to 126 days after the signal," only 14/76 (18.4%) of QQQ signals and 18/47 (38.3%) of VOO signals qualify. The signal pushes the entry price toward the low end of the range, but roughly 4/5 of QQQ signals and 3/5 of VOO signals are followed by lower prices.

Is it safer to wait for the RSI crossover to confirm?

Not in this data. The signal-day close is on average 1.75% (QQQ) and 1.19% (VOO) more expensive than on the first day it broke below 20. Buying the same events on the day it first broke below 20 instead produced better 1/3/6-month returns across the board, and the median 6-month max drawdown was not smaller.

Does this RSI oversold signal perform the same in VOO and QQQ?

No. VOO (from 2010) had 47 signals with an 85.1% 6-month win rate and an average of +8.5%, slightly ahead of random entry; QQQ's 77 full-history signals had a 69.7% 6-month win rate and an average of +4.4%, actually trailing random entry, mainly because of the 2000–2002 and 2008 bear markets. Post-2010 QQQ, like VOO, edged out random entry.

When is the RSI oversold signal most likely to fail?

When the price is already more than 20% below its 52-week (252-day) high on the signal day. QQQ had 20 such signals, with a 6-month win rate of only 20% and an average of −12.7%, all within the 2000–2002, 2008 and 2022 bear markets. This grouping was cut after seeing the results and needs out-of-sample validation.

What are the limitations of this backtest?

The samples are small and holding periods overlap; QQQ and VOO had 24 same-day signals that cannot be treated as independent evidence; commissions, slippage and taxes are not included; only a single data source, Yahoo Finance, is used; and the "down more than 20% from the high" grouping is an in-sample finding. This article only describes historical statistics and is not investment advice.

More articles

← All articles Open today's market dashboard →

Contents