Market Narratives research — where the volume washout failed: QTUM, CQQQ and IGV (2026-11-17)
Research only, 17 Nov 2026 (Asia/Jerusalem). Nothing live changed: the scan, the dashboard, the Signal and Action rules and the SOXX chart stay as they are.
The rule being checked. The volume washout entry was added to the SOXX chart on the 14 Nov dashboard post (Hybrid and Sentiment only books). While the narrative Stage is Rolling Over, a down day on volume at least 1.5× the average of the prior 50 days (today not included), closing within 3% of the lowest close of that Rolling Over stretch, is a “capitulation” day. The book then buys the first up close within the next 5 sessions, if it is not already holding. Exits are the normal ones of each book.
Why this post. Tested on 23 lead ETFs, the washout made the Hybrid basket slightly worse out of sample (2 Jan 2024 – 16 Nov 2026: +45.4% → +43.5%, 12 ETFs better, 9 worse). The three biggest misses were QTUM (quantum computing, −25.6 points), CQQQ (China tech, −21.6) and IGV (software, −14.9). In the Sentiment only book, IGV lost the most (−28.2 points). This post shows each of those trades on the chart and in tables, so you can see exactly where the points went.
Data. Yahoo daily closes through 16 Nov 2026, unadjusted (same as the SOXX chart). Stages are the rebuilt Stages for each narrative (no lookahead), shown as the chart background. The Hybrid and Sentiment only engines are the same code as the SOXX chart, run with the washout off and on. Each ETF uses its own narrative’s weekly Google AI and ChatGPT scores. Trades fill at the signal day’s close. Every window starts flat, and any open position is counted as sold at the 16 Nov 2026 close.
Jump to: QTUM · CQQQ · IGV · Why it fails · Conclusion
1. QTUM — quantum computing (Hybrid +148.7% → +123.0%)
Diagnosis. Almost the whole loss comes from one trade. On 29 Jul 2024 QTUM fell on 1.57× its prior 50-day volume, close to the low of its Rolling Over stretch. The washout bought the next up close, 31 Jul at 61.97. That wasn’t the bottom: the 5 Aug 2024 global sell-off took QTUM down to 55.33, 10.7% lower. Without the washout, the Hybrid waited for the Epstein B2 buy (the stochastic leaving its embedded-low zone) and bought on 7 Aug at 55.49.
Both books then sold on the same S2 day, 3 Sep 2024 at 59.46. So the washout trade made −4.1% where the Hybrid made +7.2%. That is an 11-point gap on one trade, but it costs −26.0 points of final return because the book had already compounded to about +149%. The other washout, 19 Nov 2025, was a draw (+0.4 points).
Sentiment only went the other way: +0.1% → +35.6%. The 19 Nov 2025 washout buy at 104.57 is still held, because that book only exits on a GC sell. It’s marked at 124.47 (+19.0%). It also blocked the 17 Jul 2026 ChatGPT buy at 141.69, which is now −12.2%. The cost is a deeper worst drop, −24.8% → −29.3%.
QTUM daily candles from 3 Jul 2023 to 16 Nov 2026 (Yahoo, unadjusted, like the SOXX chart), SMA 50/100/200 and Bollinger Bands 18,2. Panes: volume with its prior 50-day average (today excluded, as the rule uses), Stochastic 14,3,5 with 20/80 lines, and weekly Google AI and ChatGPT scores for the Quantum computing narrative from 15 Nov 2024. The default view shows the Hybrid book without and with the washout, from the out-of-sample start on 2 Jan 2024. Click any triangle for the trade details. The Zoom buttons jump to each washout trade.
QTUM: every washout buy in the Hybrid book, 2 Jan 2024 – 16 Nov 2026 (out of sample), and what it replaced
| # | Capitulation day | Volume vs prior 50-day avg | Washout buy (first up close) | Exit | Washout trade | Lowest close while held | What Hybrid without the washout did instead | Stretch with washout | Stretch without | Difference | Effect on final result |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 29 Jul 2024 close 60.98 | 1.57× | 31 Jul 2024 61.97 | 3 Sep 2024 59.46 S2 embed-high end, close below SMA 68 | −4.1% | 5 Aug 2024 55.33 (−10.7%) | B2 7 Aug 2024 at 55.49 → S2 3 Sep 2024 at 59.46: +7.2% | −4.1% | +7.2% | −11.2 pts | −26.0 pts |
| 2 | 17 Nov 2025 close 105.15 | 2.14× | 19 Nov 2025 104.57 | 17 Dec 2025 107.15 S4 waited sell, first close below SMA 68 | +2.5% | 20 Nov 2025 100.36 (−4.0%) | B2 24 Nov 2025 at 104.74 → S4 17 Dec 2025 at 107.15: +2.3% | +2.5% | +2.3% | +0.2 pts | +0.4 pts |
| Total effect = +123.0% with washout − +148.7% without (the stretches add up exactly, because each one is applied in date order to the compounded book) | −25.6 pts | ||||||||||
“Stretch” = the days where the two books held different trades. Difference = washout stretch return minus the no-washout stretch return. Effect on final result = how many points of the window’s total return that stretch added or removed, after compounding.
QTUM: every washout buy in the Sentiment only book, 15 Nov 2024 – 16 Nov 2026
| # | Capitulation day | Volume vs prior 50-day avg | Washout buy (first up close) | Exit | Washout trade | Lowest close while held | What Sentiment only without the washout did instead | Stretch with washout | Stretch without | Difference | Effect on final result |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 17 Nov 2025 close 105.15 | 2.14× | 19 Nov 2025 104.57 | 16 Nov 2026 124.47 still open, marked sold at the 16 Nov 2026 close | +19.0% | 20 Nov 2025 100.36 (−4.0%) | ChatGPT buy 17 Jul 2026 at 141.69 → Open 16 Nov 2026 at 124.47: −12.2% | +19.0% | −12.2% | +31.2 pts | +35.5 pts |
| Total effect = +35.6% with washout − +0.1% without (the stretches add up exactly, because each one is applied in date order to the compounded book) | +35.5 pts | ||||||||||
QTUM: washout buys in 2021–2023 (in-sample, Hybrid = Epstein Live rules, no sentiment yet)
| # | Capitulation day | Volume vs prior 50-day avg | Washout buy (first up close) | Exit | Washout trade | Lowest close while held | What Hybrid without the washout did instead | Stretch with washout | Stretch without | Difference | Effect on final result |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 12 May 2021 close 44.06 | 2.42× | 13 May 2021 44.41 | 16 Jul 2021 47.50 S4 waited sell, first close below SMA 68 | +7.0% | 13 May 2021 44.41 (+0.0%) | B2 18 May 2021 at 45.17 → S4 16 Jul 2021 at 47.50: +5.2% | +7.0% | +5.2% | +1.8 pts | +2.7 pts |
| 2 | 6 Sep 2022 close 39.67 | 2.07× | 7 Sep 2022 40.18 | 27 Dec 2022 38.88 S5 Stage Rolling Over, waited sell on a close below SMA 40 | −3.2% | 14 Oct 2022 34.57 (−14.0%) | B2 9 Sep 2022 at 41.45 → S5 27 Dec 2022 at 38.88: −6.2% | −3.2% | −6.2% | +3.0 pts | +5.1 pts |
| 3 | 15 Aug 2023 close 49.29 | 3.00× | 18 Aug 2023 48.06 | 7 Sep 2023 50.19 S4 waited sell, first close below SMA 68 | +4.4% | 18 Aug 2023 48.06 (+0.0%) | B2 22 Aug 2023 at 48.76 → S4 7 Sep 2023 at 50.19: +2.9% | +4.4% | +2.9% | +1.5 pts | +2.4 pts |
| Total effect = +67.4% with washout − +57.2% without (the stretches add up exactly, because each one is applied in date order to the compounded book) | +10.1 pts | ||||||||||
QTUM: without vs with the washout, table 4 format (TQQQ-16-VIX Panel 4)
| Performance QTUM | Hybrid · 4 Jan 2021 – 29 Dec 2023 (in-sample) | Hybrid · 2 Jan 2024 – 16 Nov 2026 (out of sample) | Sentiment only · 15 Nov 2024 – 16 Nov 2026 | |||||
|---|---|---|---|---|---|---|---|---|
| Without washout | With washout | Without washout | With washout | Without washout | With washout | |||
| Window | 4 Jan 2021 29 Dec 2023 | 4 Jan 2021 29 Dec 2023 | 2 Jan 2024 16 Nov 2026 | 2 Jan 2024 16 Nov 2026 | 15 Nov 2024 16 Nov 2026 | 15 Nov 2024 16 Nov 2026 | ||
| Trades | T 15/67% | T 15/67% | T 15/80% | T 15/73% | T 3/67% | T 3/100% | ||
| Buy & Hold QTUM | +30% | +30% | +134% | +134% | +93% | +93% | ||
| Buy & Hold SPY | +29% | +29% | +57% | +57% | +26% | +26% | ||
| Strategy | +57.2% | +67.4% | +148.7% | +123.0% | +0.1% | +35.6% | ||
| Max Drawdown | -21.7% | -19.2% | -20.8% | -20.8% | -24.8% | -29.3% | ||
| Max DD Recovery | 116 bars / 116 days | 111 bars / 111 days | open 28 bars / 29 days | open 28 bars / 29 days | open 64 bars / 65 days | open 115 bars / 116 days | ||
| Profit Factor | 5.41 | 7.83 | 7.18 | 5.39 | 1.11 | ∞ | ||
| Sharpe (trade) | 0.57 | 0.67 | 0.73 | 0.63 | 0.05 | 1.86 | ||
| Best Trade | +13.3% | +13.3% | +29.4% | +29.4% | +7.3% | +19.0% | ||
| Worst Trade | -6.2% | -3.2% | -12.4% | -12.4% | -12.2% | +6.1% | ||
| Avg Win | +5.9% | +6.2% | +9.6% | +9.9% | +6.7% | +10.8% | ||
| Avg Loss | -2.2% | -1.6% | -5.4% | -5.0% | -12.2% | – | ||
| Max Win Streak | 4 | 4 | 5 | 4 | 2 | 3 | ||
| Max Lose Streak | 2 | 2 | 1 | 1 | 1 | 0 | ||
| Washout buys | 0 | 3 | 0 | 2 | 0 | 1 | ||
| Loss Buckets | Loss / Win | Loss / Win | Loss / Win | Loss / Win | Loss / Win | Loss / Win | ||
| Loss/Win 0-2% | 3 / 4 | 3 / 4 | 1 / 1 | 1 / 1 | 0 / 0 | 0 / 0 | ||
| Loss/Win 2-5% | 1 / 1 | 2 / 1 | 1 / 1 | 2 / 1 | 0 / 0 | 0 / 0 | ||
| Loss/Win 5-10% | 1 / 2 | 0 / 2 | 0 / 7 | 0 / 6 | 0 / 2 | 0 / 2 | ||
| Loss/Win 10-15% | 0 / 3 | 0 / 3 | 1 / 1 | 1 / 1 | 1 / 0 | 0 / 0 | ||
| Loss/Win 15-20% | 0 / 0 | 0 / 0 | 0 / 1 | 0 / 1 | 0 / 0 | 0 / 1 | ||
| Loss/Win 20-25% | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | ||
| Loss/Win 25-30% | 0 / 0 | 0 / 0 | 0 / 1 | 0 / 1 | 0 / 0 | 0 / 0 | ||
| Loss/Win >30% | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | ||
Panel 4 rows as in the TQQQ-16-VIX Pine script. Open positions are counted as sold at the 16 Nov 2026 close in every number. Each window starts flat. Max Drawdown is close-to-close on the strategy equity. Loss / Win buckets count trades by size.
2. CQQQ — China tech (Hybrid +11.1% → −10.5%)
Diagnosis. There were two washout buys, and both came too early in a decline that kept going. 10 Jan 2024 (1.70× volume): it bought 11 Jan at 33.26, and CQQQ kept falling to 27.95 on 2 Feb (−16.0%). The Hybrid waited for B2 on 24 Jan at 31.94. Both sold on S2, 28 Feb at 30.97: −6.9% against −3.0%, so −4.4 points.
5 Feb 2026 (2.84× volume, the heaviest day in this post): it bought 6 Feb at 53.80. The slide lasted almost eight more weeks, to 44.72 on 30 Mar (−16.9%). The Hybrid’s B2 came on 2 Apr at 45.15, near the low. Both sold on S2, 23 Apr at 49.23: −8.5% against +9.0%, so −17.2 points. The worst drop went from −37.8% to −47.4%. In CQQQ, heavy selling volume marked the middle of a fall, not its end.
Sentiment only (−4.0% → −9.4%): on 5 Nov 2025 the washout bought at 55.35 while that book was flat. It fell to 50.41 and was closed by the 26 Dec 2025 GC sell at 52.23, −5.6%.
CQQQ daily candles from 3 Jul 2023 to 16 Nov 2026 (Yahoo, unadjusted, like the SOXX chart), SMA 50/100/200 and Bollinger Bands 18,2. Panes: volume with its prior 50-day average (today excluded, as the rule uses), Stochastic 14,3,5 with 20/80 lines, and weekly Google AI and ChatGPT scores for the China tech de-risking narrative from 15 Nov 2024. The default view shows the Hybrid book without and with the washout, from the out-of-sample start on 2 Jan 2024. Click any triangle for the trade details. The Zoom buttons jump to each washout trade.
CQQQ: every washout buy in the Hybrid book, 2 Jan 2024 – 16 Nov 2026 (out of sample), and what it replaced
| # | Capitulation day | Volume vs prior 50-day avg | Washout buy (first up close) | Exit | Washout trade | Lowest close while held | What Hybrid without the washout did instead | Stretch with washout | Stretch without | Difference | Effect on final result |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 10 Jan 2024 close 32.64 | 1.70× | 11 Jan 2024 33.26 | 28 Feb 2024 30.97 S2 embed-high end, close below SMA 68 | −6.9% | 2 Feb 2024 27.95 (−16.0%) | B2 24 Jan 2024 at 31.94 → S2 28 Feb 2024 at 30.97: −3.0% | −6.9% | −3.0% | −3.8 pts | −4.4 pts |
| 2 | 5 Feb 2026 close 52.15 | 2.84× | 6 Feb 2026 53.80 | 23 Apr 2026 49.23 S2 embed-high end, close below SMA 68 | −8.5% | 30 Mar 2026 44.72 (−16.9%) | B2 2 Apr 2026 at 45.15 → S2 23 Apr 2026 at 49.23: +9.0% | −8.5% | +9.0% | −17.5 pts | −17.2 pts |
| Total effect = −10.5% with washout − +11.1% without (the stretches add up exactly, because each one is applied in date order to the compounded book) | −21.6 pts | ||||||||||
“Stretch” = the days where the two books held different trades. Difference = washout stretch return minus the no-washout stretch return. Effect on final result = how many points of the window’s total return that stretch added or removed, after compounding.
CQQQ: every washout buy in the Sentiment only book, 15 Nov 2024 – 16 Nov 2026
| # | Capitulation day | Volume vs prior 50-day avg | Washout buy (first up close) | Exit | Washout trade | Lowest close while held | What Sentiment only without the washout did instead | Stretch with washout | Stretch without | Difference | Effect on final result |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 4 Nov 2025 close 54.78 | 2.10× | 5 Nov 2025 55.35 | 26 Dec 2025 52.23 GC sell: start of both-models +≥40 run (G 50, C 45) | −5.6% | 21 Nov 2025 50.41 (−8.9%) | Flat (no trade) | −5.6% | +0.0% | −5.6 pts | −5.4 pts |
| Total effect = −9.4% with washout − −4.0% without (the stretches add up exactly, because each one is applied in date order to the compounded book) | −5.4 pts | ||||||||||
CQQQ: washout buys in 2021–2023 (in-sample, Hybrid = Epstein Live rules, no sentiment yet)
| # | Capitulation day | Volume vs prior 50-day avg | Washout buy (first up close) | Exit | Washout trade | Lowest close while held | What Hybrid without the washout did instead | Stretch with washout | Stretch without | Difference | Effect on final result |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 8 Jul 2021 close 76.56 | 1.60× | 9 Jul 2021 78.76 | 10 Sep 2021 71.62 S2 embed-high end, close below SMA 68 | −9.1% | 19 Aug 2021 63.94 (−18.8%) | B2 24 Aug 2021 at 70.44 → S2 10 Sep 2021 at 71.62: +1.7% | −9.1% | +1.7% | −10.7 pts | −4.2 pts |
| Total effect = −64.3% with washout − −60.1% without (the stretches add up exactly, because each one is applied in date order to the compounded book) | −4.2 pts | ||||||||||
CQQQ: without vs with the washout, table 4 format (TQQQ-16-VIX Panel 4)
| Performance CQQQ | Hybrid · 4 Jan 2021 – 29 Dec 2023 (in-sample) | Hybrid · 2 Jan 2024 – 16 Nov 2026 (out of sample) | Sentiment only · 15 Nov 2024 – 16 Nov 2026 | |||||
|---|---|---|---|---|---|---|---|---|
| Without washout | With washout | Without washout | With washout | Without washout | With washout | |||
| Window | 4 Jan 2021 29 Dec 2023 | 4 Jan 2021 29 Dec 2023 | 2 Jan 2024 16 Nov 2026 | 2 Jan 2024 16 Nov 2026 | 15 Nov 2024 16 Nov 2026 | 15 Nov 2024 16 Nov 2026 | ||
| Trades | T 12/42% | T 12/33% | T 13/38% | T 13/31% | T 3/67% | T 4/50% | ||
| Buy & Hold CQQQ | -57% | -57% | +6% | +6% | -9% | -9% | ||
| Buy & Hold SPY | +29% | +29% | +57% | +57% | +26% | +26% | ||
| Strategy | -60.1% | -64.3% | +11.1% | -10.5% | -4.0% | -9.4% | ||
| Max Drawdown | -62.7% | -66.7% | -37.8% | -47.4% | -35.3% | -38.2% | ||
| Max DD Recovery | open 712 bars / 713 days | open 712 bars / 713 days | open 90 bars / 91 days | open 279 bars / 280 days | open 90 bars / 91 days | open 256 bars / 257 days | ||
| Profit Factor | 0.14 | 0.11 | 1.45 | 1.03 | 1.21 | 1.03 | ||
| Sharpe (trade) | -0.60 | -0.70 | 0.13 | 0.01 | 0.09 | 0.01 | ||
| Best Trade | +8.4% | +8.4% | +30.7% | +30.7% | +20.7% | +20.7% | ||
| Worst Trade | -34.3% | -34.3% | -29.3% | -29.3% | -33.7% | -33.7% | ||
| Avg Win | +2.7% | +2.9% | +15.2% | +16.8% | +20.3% | +20.3% | ||
| Avg Loss | -13.3% | -12.8% | -6.6% | -7.2% | -33.7% | -19.7% | ||
| Max Win Streak | 2 | 2 | 3 | 3 | 2 | 2 | ||
| Max Lose Streak | 3 | 3 | 3 | 6 | 1 | 2 | ||
| Washout buys | 0 | 1 | 0 | 2 | 0 | 1 | ||
| Loss Buckets | Loss / Win | Loss / Win | Loss / Win | Loss / Win | Loss / Win | Loss / Win | ||
| Loss/Win 0-2% | 0 / 4 | 0 / 3 | 4 / 0 | 4 / 0 | 0 / 0 | 0 / 0 | ||
| Loss/Win 2-5% | 1 / 0 | 1 / 0 | 1 / 1 | 0 / 1 | 0 / 0 | 0 / 0 | ||
| Loss/Win 5-10% | 3 / 1 | 4 / 1 | 2 / 1 | 4 / 0 | 0 / 0 | 1 / 0 | ||
| Loss/Win 10-15% | 1 / 0 | 1 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | ||
| Loss/Win 15-20% | 1 / 0 | 1 / 0 | 0 / 2 | 0 / 2 | 0 / 1 | 0 / 1 | ||
| Loss/Win 20-25% | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 1 | 0 / 1 | ||
| Loss/Win 25-30% | 0 / 0 | 0 / 0 | 1 / 0 | 1 / 0 | 0 / 0 | 0 / 0 | ||
| Loss/Win >30% | 1 / 0 | 1 / 0 | 0 / 1 | 0 / 1 | 1 / 0 | 1 / 0 | ||
Panel 4 rows as in the TQQQ-16-VIX Pine script. Open positions are counted as sold at the 16 Nov 2026 close in every number. Each window starts flat. Max Drawdown is close-to-close on the strategy equity. Loss / Win buckets count trades by size.
3. IGV — software (Hybrid +47.4% → +32.5%)
Diagnosis. IGV shows the pattern most clearly. The washout bought five times out of sample. Three times it paid more than the Hybrid paid a few weeks later, and those three cost 22.7 points. The two that bought lower added back 7.8. The IGV washouts fired on the first oversold days (%K 11–25, stochastic not yet embedded low), so they arrived before the selling was done.
The worst one: capitulation on 24 Feb 2025 (1.73× volume), buy on 26 Feb at 98.95. IGV then fell to 80.15 on 8 Apr 2025 (−19.0%) in the April 2025 tariff sell-off. The Hybrid bought B2 on 17 Mar at 92.87. Both left on the same GC sell, 2 May at 99.47: +0.5% against +7.1%, so −9.1 points. Next were 7 Nov 2025 (−8.0 points: bought at 109.48, while the Hybrid’s B2 came on 25 Nov at 103.19) and 11 Apr 2024 (−5.6 points). The one clear win was 31 May 2024 (+6.8 points), when the heavy-volume day really was the low. The 9 Nov 2026 buy is still open (+1.0 point so far).
Sentiment only (+21.0% → −7.2%, −28.2 points) is worse because that book has no stop: only a GC sell closes a trade. The 26 Feb 2025 washout held the position, so the 4 Apr 2025 ChatGPT contrarian buy at 81.30 was skipped. That was the best trade of the period (+22.3%), so the cost was −21.6 points. Then the 7 Nov 2025 washout at 109.48 was held for nine months, as low as 74.67 (−31.8%) on 10 Apr 2026, and it blocked the 16 Jan 2026 ChatGPT buy (+4.5%). That cost −10.5 points. The 22 Aug 2025 washout helped (+3.9 points).
IGV daily candles from 3 Jul 2023 to 16 Nov 2026 (Yahoo, unadjusted, like the SOXX chart), SMA 50/100/200 and Bollinger Bands 18,2. Panes: volume with its prior 50-day average (today excluded, as the rule uses), Stochastic 14,3,5 with 20/80 lines, and weekly Google AI and ChatGPT scores for the Software sector narrative from 15 Nov 2024. The default view shows the Hybrid book without and with the washout, from the out-of-sample start on 2 Jan 2024. Click any triangle for the trade details. The Zoom buttons jump to each washout trade.
IGV: every washout buy in the Hybrid book, 2 Jan 2024 – 16 Nov 2026 (out of sample), and what it replaced
| # | Capitulation day | Volume vs prior 50-day avg | Washout buy (first up close) | Exit | Washout trade | Lowest close while held | What Hybrid without the washout did instead | Stretch with washout | Stretch without | Difference | Effect on final result |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 10 Apr 2024 close 83.59 | 1.54× | 11 Apr 2024 84.14 | 24 May 2024 82.90 S2 embed-high end, close below SMA 68 | −1.5% | 19 Apr 2024 78.53 (−6.7%) | B2 24 Apr 2024 at 80.94 → S2 24 May 2024 at 82.90: +2.4% | −1.5% | +2.4% | −3.9 pts | −5.6 pts |
| 2 | 29 May 2024 close 82.11 | 1.51× | 31 May 2024 78.09 | 24 Jul 2024 82.89 S4 waited sell, first close below SMA 68 | +6.1% | 3 Jun 2024 77.94 (−0.2%) | B4 13 Jun 2024 at 81.85 → S4 24 Jul 2024 at 82.89: +1.3% | +6.1% | +1.3% | +4.9 pts | +6.8 pts |
| 3 | 24 Feb 2025 close 99.20 | 1.73× | 26 Feb 2025 98.95 | 2 May 2025 99.47 GC sell: start of both-models +≥40 run (G 40, C 52) | +0.5% | 8 Apr 2025 80.15 (−19.0%) | B2 17 Mar 2025 at 92.87 → GC sell 2 May 2025 at 99.47: +7.1% | +0.5% | +7.1% | −6.6 pts | −9.1 pts |
| 4 | 6 Nov 2025 close 109.31 | 1.64× | 7 Nov 2025 109.48 | 15 Dec 2025 105.39 S2 embed-high end, close below SMA 68 | −3.7% | 20 Nov 2025 100.80 (−7.9%) | B2 25 Nov 2025 at 103.19 → S2 15 Dec 2025 at 105.39: +2.1% | −3.7% | +2.1% | −5.9 pts | −8.0 pts |
| 5 | 5 Nov 2026 close 99.01 | 1.74× | 9 Nov 2026 101.46 | 16 Nov 2026 99.00 still open, marked sold at the 16 Nov 2026 close | −2.4% | 16 Nov 2026 99.00 (−2.4%) | B2 11 Nov 2026 at 102.23 → Open 16 Nov 2026 at 99.00: −3.2% | −2.4% | −3.2% | +0.7 pts | +1.0 pts |
| Total effect = +32.5% with washout − +47.4% without (the stretches add up exactly, because each one is applied in date order to the compounded book) | −14.9 pts | ||||||||||
“Stretch” = the days where the two books held different trades. Difference = washout stretch return minus the no-washout stretch return. Effect on final result = how many points of the window’s total return that stretch added or removed, after compounding.
IGV: every washout buy in the Sentiment only book, 15 Nov 2024 – 16 Nov 2026
| # | Capitulation day | Volume vs prior 50-day avg | Washout buy (first up close) | Exit | Washout trade | Lowest close while held | What Sentiment only without the washout did instead | Stretch with washout | Stretch without | Difference | Effect on final result |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 24 Feb 2025 close 99.20 | 1.73× | 26 Feb 2025 98.95 | 2 May 2025 99.47 GC sell: start of both-models +≥40 run (G 40, C 52) | +0.5% | 8 Apr 2025 80.15 (−19.0%) | ChatGPT buy 4 Apr 2025 at 81.30 → GC sell 2 May 2025 at 99.47: +22.3% | +0.5% | +22.3% | −21.8 pts | −21.6 pts |
| 2 | 20 Aug 2025 close 106.33 | 1.57× | 22 Aug 2025 107.46 | 12 Sep 2025 111.67 GC sell: start of both-models +≥40 run (G 45, C 48) | +3.9% | 26 Aug 2025 106.57 (−0.8%) | Flat (no trade) | +3.9% | +0.0% | +3.9 pts | +3.9 pts |
| 3 | 6 Nov 2025 close 109.31 | 1.64× | 7 Nov 2025 109.48 | 7 Aug 2026 102.69 GC sell: start of both-models +≥40 run (G 55, C 62) | −6.2% | 10 Apr 2026 74.67 (−31.8%) | ChatGPT buy 16 Jan 2026 at 98.31 → GC sell 7 Aug 2026 at 102.69: +4.5% | −6.2% | +4.5% | −10.7 pts | −10.5 pts |
| Total effect = −7.2% with washout − +21.0% without (the stretches add up exactly, because each one is applied in date order to the compounded book) | −28.2 pts | ||||||||||
IGV: washout buys in 2021–2023 (in-sample, Hybrid = Epstein Live rules, no sentiment yet)
| # | Capitulation day | Volume vs prior 50-day avg | Washout buy (first up close) | Exit | Washout trade | Lowest close while held | What Hybrid without the washout did instead | Stretch with washout | Stretch without | Difference | Effect on final result |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 4 Mar 2021 close 66.80 | 3.91× | 5 Mar 2021 67.30 | 3 May 2021 71.21 S4 waited sell, first close below SMA 68 | +5.8% | 8 Mar 2021 65.36 (−2.9%) | B2 10 Mar 2021 at 67.78 → S4 3 May 2021 at 71.21: +5.1% | +5.8% | +5.1% | +0.7 pts | +0.9 pts |
| 2 | 4 Oct 2021 close 78.57 | 2.31× | 5 Oct 2021 79.91 | 23 Nov 2021 84.06 S4 waited sell, first close below SMA 68 | +5.2% | 5 Oct 2021 79.91 (+0.0%) | B4 13 Oct 2021 at 83.22 → S4 23 Nov 2021 at 84.06: +1.0% | +5.2% | +1.0% | +4.2 pts | +5.2 pts |
| Total effect = +31.2% with washout − +25.1% without (the stretches add up exactly, because each one is applied in date order to the compounded book) | +6.1 pts | ||||||||||
IGV: without vs with the washout, table 4 format (TQQQ-16-VIX Panel 4)
| Performance IGV | Hybrid · 4 Jan 2021 – 29 Dec 2023 (in-sample) | Hybrid · 2 Jan 2024 – 16 Nov 2026 (out of sample) | Sentiment only · 15 Nov 2024 – 16 Nov 2026 | |||||
|---|---|---|---|---|---|---|---|---|
| Without washout | With washout | Without washout | With washout | Without washout | With washout | |||
| Window | 4 Jan 2021 29 Dec 2023 | 4 Jan 2021 29 Dec 2023 | 2 Jan 2024 16 Nov 2026 | 2 Jan 2024 16 Nov 2026 | 15 Nov 2024 16 Nov 2026 | 15 Nov 2024 16 Nov 2026 | ||
| Trades | T 13/62% | T 13/62% | T 19/58% | T 19/47% | T 3/67% | T 4/50% | ||
| Buy & Hold IGV | +17% | +17% | +26% | +26% | -2% | -2% | ||
| Buy & Hold SPY | +29% | +29% | +57% | +57% | +26% | +26% | ||
| Strategy | +25.1% | +31.2% | +47.4% | +32.5% | +21.0% | -7.2% | ||
| Max Drawdown | -40.1% | -40.1% | -33.2% | -36.2% | -24.4% | -33.2% | ||
| Max DD Recovery | 533 bars / 533 days | 533 bars / 533 days | open 233 bars / 234 days | open 289 bars / 290 days | 85 bars / 85 days | open 254 bars / 255 days | ||
| Profit Factor | 1.85 | 2.00 | 2.47 | 1.99 | 5.03 | 0.39 | ||
| Sharpe (trade) | 0.23 | 0.27 | 0.27 | 0.21 | 0.62 | -0.42 | ||
| Best Trade | +16.0% | +16.0% | +28.3% | +28.3% | +22.3% | +3.9% | ||
| Worst Trade | -20.7% | -20.7% | -21.4% | -21.4% | -5.3% | -6.2% | ||
| Avg Win | +7.9% | +8.5% | +7.2% | +8.1% | +13.4% | +2.2% | ||
| Avg Loss | -6.8% | -6.8% | -4.0% | -3.7% | -5.3% | -5.8% | ||
| Max Win Streak | 4 | 4 | 3 | 3 | 2 | 2 | ||
| Max Lose Streak | 3 | 3 | 2 | 4 | 1 | 2 | ||
| Washout buys | 0 | 2 | 0 | 5 | 0 | 3 | ||
| Loss Buckets | Loss / Win | Loss / Win | Loss / Win | Loss / Win | Loss / Win | Loss / Win | ||
| Loss/Win 0-2% | 1 / 3 | 1 / 2 | 5 / 1 | 6 / 1 | 0 / 0 | 0 / 1 | ||
| Loss/Win 2-5% | 2 / 0 | 2 / 0 | 2 / 5 | 3 / 3 | 0 / 1 | 0 / 1 | ||
| Loss/Win 5-10% | 1 / 2 | 1 / 3 | 0 / 3 | 0 / 3 | 1 / 0 | 2 / 0 | ||
| Loss/Win 10-15% | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | ||
| Loss/Win 15-20% | 0 / 3 | 0 / 3 | 0 / 1 | 0 / 1 | 0 / 0 | 0 / 0 | ||
| Loss/Win 20-25% | 1 / 0 | 1 / 0 | 1 / 0 | 1 / 0 | 0 / 1 | 0 / 0 | ||
| Loss/Win 25-30% | 0 / 0 | 0 / 0 | 0 / 1 | 0 / 1 | 0 / 0 | 0 / 0 | ||
| Loss/Win >30% | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | 0 / 0 | ||
Panel 4 rows as in the TQQQ-16-VIX Pine script. Open positions are counted as sold at the 16 Nov 2026 close in every number. Each window starts flat. Max Drawdown is close-to-close on the strategy equity. Loss / Win buckets count trades by size.
Why it fails on these ETFs
- Only the entry changes. In the Hybrid, every washout trade on these three ETFs ended on the same day as the trade it replaced. That’s because Epstein’s sells (S1, S2, S4, S5) only start once the stochastic gets back above 80. So the washout gains only when it buys lower than the Hybrid’s own next buy, and loses when it buys higher.
- It buys before Epstein would. Every washout buy came below the 18-day middle band, his “line in the sand,” with %K still between 10 and 25. The Hybrid instead waits for B2 (the stochastic leaving its embedded-low zone) or another B rule. On these ETFs that wait came after the real low: 6 of the 9 out-of-sample washouts bought above the price the Hybrid paid later.
- The heavy-volume day wasn’t the low. After the losing washout buys, prices fell roughly another 7% to 19% (QTUM −10.7%, CQQQ −16.0% and −16.9%, IGV −6.7%, −19.0% and −7.9%). The rule only asks for a close within 3% of the Rolling Over low, so a heavy day in the middle of a long decline qualifies.
- Compounding makes one miss look big. QTUM’s book had already more than doubled, so one trade 11 points worse took 26 points off the final result.
- Sentiment only has no exit except the GC sell. A washout buy there can sit through the whole decline, and while it sits, it blocks the better contrarian ChatGPT buys (IGV April 2025 and January 2026).
- Why SOXX was different. On SOXX the heavy-volume days happened to fall at or near the lows, so the earlier entry paid (+11.6 points in 2021–2023, +10.9 in 2024–2026). In 2021–2023 the washout also helped QTUM and IGV. It’s a small number of trades that can go either way, not a lasting edge.
Conclusion
The washout’s only edge is buying a few days before the Epstein B2 buy. That pays when the heavy-volume day is the real low, and it costs when the decline keeps going, which is what happened in QTUM (Aug 2024), CQQQ (Jan 2024 and Feb 2026) and IGV (Feb 2025 and Nov 2025). Making it wait for the stochastic to turn or for a close back above the 18-day line would just turn it into the existing B2/B4 rules. So there is no version worth adding to other ETFs.
Recommendation: keep the washout entry on the SOXX chart only, as it is now. Nothing was changed on any other post, the live scan, routines, the Page or any live rule.