A deeply-drawn-down stock that reverses hard intraday can be short covering. This study isolates it in PLTR — >30% below the
200-day high and a −2/+2 reversal day — across all 20 occurrences since the 2020 IPO. But the plain average buries the signal that
PLTR's construct has completely changed: the 2021–22 events were a speculative-bear dead-cat, while the modern, profitable, S&P-500 PLTR
(2023+) turned every one of these into a winner. So the headline here is the current-construct weighting — pre-2023 ×2, 2023–2025 ×6, 2026 ×9 —
which flips the intermediate-horizon read from a fade to a hold.
📄 Download the AQI brief (PDF)
🟢 The live trade & ladder
Trigger: close ≤ 70% of the trailing 200-day high (a ≥30% drawdown) ◆ intraday low ≤ prior close −2%
(traded down ≥2%) ◆ close ≥ prior close +2% (a ≥4% intraday reversal). Forward returns are measured from the signal close.
Every occurrence & its forward returns
| Signal date | Wt | Drawdown | Close |
+3d | +5d | +10d | +21d | +42d | +63d |
| 2021-11-24 | ×2 | -36% | $21.11 | -2.2 | -6.7 | -9.0 | -10.3 | -39.6 | -45.7 |
| 2021-12-17 | ×2 | -34% | $19.06 | -0.9 | -0.6 | -2.8 | -23.6 | -38.2 | -33.9 |
| 2022-01-28 | ×2 | -56% | $12.71 | +4.8 | +1.8 | +3.3 | -4.2 | +9.2 | -18.2 |
| 2022-02-24 | ×2 | -59% | $11.83 | +3.0 | -4.1 | +0.3 | +9.6 | -6.0 | -32.7 |
| 2022-03-15 | ×2 | -63% | $10.77 | +19.0 | +23.2 | +31.8 | +19.8 | -22.6 | -27.4 |
| 2022-05-02 | ×2 | -63% | $10.74 | -5.8 | -30.5 | -25.1 | -21.2 | -13.7 | +1.7 |
| 2022-05-12 | ×2 | -74% | $7.34 | +13.9 | +13.2 | +13.9 | +4.1 | +21.3 | +35.0 |
| 2022-05-26 | ×2 | -71% | $8.36 | +1.2 | +6.9 | -1.2 | +11.4 | +21.2 | -5.0 |
| 2022-06-22 | ×2 | -69% | $9.01 | +6.8 | +2.4 | +13.0 | +9.2 | -10.4 | -14.7 |
| 2022-07-05 | ×2 | -65% | $10.07 | +1.0 | -5.8 | -7.2 | +11.2 | -25.6 | -18.2 |
| 2022-07-29 | ×2 | -61% | $10.35 | +8.2 | +10.6 | -4.3 | -24.5 | -23.3 | -16.9 |
| 2022-09-14 | ×2 | -64% | $7.63 | +3.0 | +0.8 | +4.1 | +5.9 | +10.2 | -1.7 |
| 2022-10-21 | ×2 | -51% | $8.29 | +3.4 | +4.2 | -4.3 | -13.3 | -21.6 | -15.3 |
| 2022-11-11 | ×2 | -43% | $8.41 | -4.2 | -12.1 | -14.9 | -10.8 | -17.2 | +9.6 |
| 2022-12-21 | ×2 | -56% | $6.50 | -7.7 | -1.5 | -1.5 | +8.0 | +25.7 | +26.2 |
| 2023-01-25 | ×6 | -44% | $7.22 | +2.1 | +13.6 | +13.9 | +12.0 | +11.4 | +7.2 |
| 2023-03-13 | ×6 | -31% | $7.87 | +1.4 | +2.7 | +2.2 | +5.8 | +25.5 | +98.9 |
| 2025-04-07 | ×6 | -38% | $77.84 | +13.8 | +19.0 | +20.7 | +41.9 | +64.1 | +83.9 |
| 2026-07-06 partial | ×9 | -36% | $132.54 | -2.6 | -1.9 | +1.7 | +22.7 | — | — |
| 2026-07-14 LIVE | ×9 | -35% | $133.72 | -1.0 | -0.8 | -7.6 | +27.9 | — | — |
Look down the +63d column: the deep-drawdown 2022 events (−50 to −74%) are a sea of red (dead-cats); the moderate-drawdown
turns — 2023-03-13 (+99%) and 2025-04-07 (+84%) — are the launches. The live row updates from PLTR each build.
Forward-return paths & distribution
| Horizon | n | Med | %+ | Wtd med | Wtd avg | Wtd %+ |
|---|
| +3d | 20 | +1.8 | 65% | +1.4 | +2.4 | 58% |
| +5d | 20 | +1.3 | 55% | +0.8 | +2.9 | 52% |
| +10d | 20 | -0.4 | 50% | +1.7 | +2.4 | 59% |
| +21d | 20 | +7.0 | 65% | +11.4 | +11.5 | 79% |
| +42d | 18 | -8.2 | 44% | +10.2 | +7.2 | 58% |
| +63d | 18 | -9.8 | 39% | +1.7 | +17.2 | 54% |
Plain median vs the regime-weighted view (pre-23 ×2, 23–25 ×6, 26 ×9). The weighting flips +42d/+63d from negative to positive
— the fade was a 2021–22 bear artifact. The purple dashed line on the chart is the weighted median; gold is the live occurrence.
The honest read — reweighted to the current construct
1 · The near-term pop is regime-agnostic
Whether you weight or not, the entry works: +1.3% median at 5 sessions (55% green) with almost no adverse excursion. Short covering is short covering. The disagreement is what happens after three weeks.
2 · The weighting flips the intermediate horizon
Unweighted, +42d is -8.2% and +63d is -9.8% (a dead-cat). Weighted to the modern construct, +42d becomes +10.2% and +63d +1.7% median / +17% average, with the +21d win-rate at 79%. The fade was a bear-market artifact the weighting deliberately down-weights.
3 · Why the weighting is justified here
The 2021–22 events (weight ×2) were a pre-profit, pre-index, story-stock in an −80% bear — every bounce failed. The 2023–2025 events (weight ×6) are a profitable, S&P-500, AI-era PLTR, and all three worked: 2023-01-25 (−44% dd) +7% at 63d, 2023-03-13 (−31%) +99%, 2025-04-07 (−38%) +84%. Crucially they fired at moderate −30 to −44% drawdowns — exactly where PLTR sits today at -4%.
4 · The caveat the weighting can't erase
A weighting is a view, not new data. The up-weighted modern cohort is only n=3, and its +average is still carried by two moonshots. If PLTR is entering a genuine downtrend (not a shakeout), it reverts to 2022 dead-cat behavior — which is why the reversal-day low stays the hard stop. This expresses conviction that the construct rhymes with 2023+; it does not prove it.
Idea & trade — reassembled
Idea. Under the current-construct weighting, the −2/+2 reversal >30% below the 200-day high is not a rent-the-bounce dead-cat — it is a
buyable turn in a name whose modern analogs (all at today's moderate drawdown depth) followed through hard. Hold it for the intermediate move,
not just the pop.
Trade — hold the turn. Long PLTR off the reversal close.
◆ Entry: the 3–5 day pop is the reliable get-in (+1.3% median, 55% green, tight risk) — but don't sell it at three weeks; the weighted edge keeps building (+21d +11% / 79%, +42d +10%).
◆ Hold / target: the 21–63 session window (weighted +63d +2% median / +17% average) — trail rather than time-stop out.
◆ Stop (unchanged): a close back below the reversal-day low = the turn failed, revert to the bear read.
◆ Press on strength: a reclaim of the 200-day puts the +80–99% 2023/2025 tail in play — add there.
◆ Size: moderate conviction — larger than a pure rent, but the up-weighted cohort is n=3, so it's a regime bet, sized accordingly.
Research, not investment advice. The weighting (pre-23 ×2, 23–25 ×6, 26 ×9) is a deliberate expression of the view that PLTR's current construct resembles the 2023+ regime; the underlying modern sample is small and tail-influenced. Verify against your own risk framework.