⬇ Download PDF
On 2026-07-07 IWD printed a fresh all-time high, then closed in the bottom 10.6% of its daily range;
the next session gave back −1.03%. That three-part sequence — new high → weak close → downside confirmation
— is the operational analog to a classic island-top reversal. Ten events satisfy all three conditions since 2000. This note lays
out every analog and what the tape actually did next. The live event's forward returns fill in from live IWD as sessions elapse (through August 31, 2026).
Trigger (t0): intraday high exceeds the prior 200-session high (a new 200-day high) and the close lands in the bottom
20% of the day's range. ◆ Confirmation (t+1): the next session closes ≤ −1%. ◆
Forward returns are measured from the t0 close.
Event triggers & forward returns
| Trigger date (t0) | Close | Close-in-range |
t+1 | t+5 | t+10 | t+20 | t+30 | t+60 |
| 2003-06-06 | 30.66 | 12.4% | -1.45 | -0.03 | +0.33 | +1.07 | -1.35 | +2.76 |
| 2006-05-11 | 47.49 | 15.3% | -1.32 | -3.75 | -2.74 | -3.94 | -4.59 | +0.68 |
| 2007-05-23 | 57.72 | 8.1% | -1.09 | +0.28 | -2.59 | -0.54 | -0.44 | -6.70 |
| 2009-09-23 | 38.81 | 6.7% | -1.45 | -0.82 | -0.63 | +1.41 | -2.98 | +2.02 |
| 2010-04-26 | 44.81 | 8.5% | -2.64 | -0.58 | -4.52 | -12.14 | -12.93 | -12.40 |
| 2014-04-04 | 73.95 | 4.4% | -1.17 | -2.62 | +0.17 | +0.74 | +0.75 | +5.60 |
| 2021-05-10 | 146.26 | 2.7% | -1.28 | -0.63 | -0.66 | +0.19 | -2.18 | -1.05 |
| 2021-10-26 | 151.09 | 2.9% | -1.37 | +0.22 | +0.91 | +0.13 | -0.59 | -1.84 |
| 2026-06-16 partial | 242.96 | 4.1% | -1.06 | -0.81 | +0.38 | +2.68 | +3.19 | — |
| 2026-07-07 LIVE | 247.74 | 10.6% | -1.03 | -0.55 | -0.12 | +3.59 | +3.63 | — |
The live 2026-07-07 row (shaded) updates from live IWD. Blank cells are windows not yet elapsed. The 2026-06-16 event
is a near re-fire only ~15 sessions earlier, so the two are not independent.
Forward-return paths & distribution
| Horizon | Mean | Median | Min | Max | % pos |
|---|
| t+1 | -1.39 | -1.30 | -2.64 | -1.03 | 0% |
| t+5 | -0.93 | -0.60 | -3.75 | +0.28 | 20% |
| t+10 | -0.95 | -0.38 | -4.52 | +0.91 | 40% |
| t+20 | -0.68 | +0.46 | -12.14 | +3.59 | 70% |
| t+30 | -1.75 | -0.97 | -12.93 | +3.63 | 30% |
| t+60 | -1.37 | -0.18 | -12.40 | +5.60 | 50% |
Across the 8 events with complete windows. The gold line is the live 2026-07-07 event through today's close.
The honest read
1 · The t+1 column is definitional, not predictive
The −1% drop is baked into the trigger, so every event shows a negative t+1 by construction. The actual forecasting content starts at t+5.
2 · A weak, fat-left-tailed signal — not a clean reversal
Beyond the mechanical t+1 drop, the median path is a shallow drift: median t+20 is slightly positive (+0.16%) and 62% of events are green by t+20. What poisons the mean is the tail — the 2010-04-26 event sits atop the May-2010 sovereign-debt selloff / flash crash and runs to −12% by t+30–60. Strip that one and the pattern is close to noise.
3 · What the pattern really warns about
The trade this signals isn't a reliable grind lower — it's tail insurance against the roughly 1-in-8 event where the weak close was the first crack.
4 · Sample caveats
n = 10 (only 8 with a complete t+60), and two pairs are non-independent — the 2021 pair, and more importantly 2026-06-16 vs 2026-07-07, which overlap by ~15 sessions. The current signal is partly a re-fire of one already received three weeks ago; that prior instance went −0.81% at t+5, then recovered to +0.38% by t+10.
Prepared by Brad Sullivan, Founder & CIO, AlphaQuery Intelligence. For informational and research purposes only; not investment advice. Small, partially overlapping sample — verify all figures against source data before acting.