Question: As of 2026-05-22 (16 calendar days / 12 trading days after the 2026-05-10 baseline), did $DRAM fire either of the two watch triggers from TASKS-RESEARCH.md:113? (a) doubled again from $52.80 ATH in <30 days → hard fade signal on memory cohort; (b) rolled over -20% with cohort holding up → ETF-specific froth blowing off.

Verdict: Neither trigger fired. $DRAM consolidated tightly at the ATH zone ($46–56 range), tracking the underlying memory cohort rather than running ahead or rolling off independently. The "OTD penny stock" / "ETF-specific froth" framing from the 5/10 read does not match the past 12 trading days of behavior: $DRAM is acting as a derivative of $MU/$HY9H.F, not as an autonomous froth signal. Standing-watch retained with refined triggers proposed below.

What we're asking

The 2026-05-10 baseline note flagged $DRAM as +101.9% off its $26.14 launch low, with a +13.4% single-day move on 39.7M volume vs 13.7M average (2.9× volume) closing at $52.80 ATH. The watch frame was:

• Up-trigger: $DRAM doubles again from $52.80 in <30 days → ETF leading the cohort means the cohort is over-extended; hard fade.
• Down-trigger: $DRAM rolls over -20% from $52.80 while $MU/$STX/$WDC/$HY9H.F hold up → ETF-specific froth blowing off, thesis intact.
• Tell: "Volume + flow are the tell."

12 trading days later, neither extreme has materialized. The investigation tests the actual path of price + volume against both triggers and against the cohort.

What we found

1 — Trigger evaluation

(Up-trigger ≥$105.60 = 2× from $52.80; down-trigger ≤$42.24 = -20% from $52.80.)

• Up-trigger: Closest approach to a new ATH was $56.38 intraday on 2026-05-13 (+6.8% from baseline ATH). Never anywhere near +100%.
• Down-trigger: Closest approach was $46.43 intraday low on 2026-05-18 (-12.1% from baseline). Did not touch -20%.

2 — Cohort-relative behavior

Indexed to 2026-05-01 close, daily ($ values; % from 5/01 base):

($SNDK pulled back -5.4% vs 5/08 and is excluded from the cohort lead/lag table — sideways.)

Three observations:

(a) $DRAM did not lead. Peak gain over the period: $HY9H.F +50.0% > $MU +46.7% > $DRAM +36.3% > $STX +14.7%. The Korean HBM primary ($HY9H.F) and DRAM-maker ($MU) both ran further than the ETF. The "ETF dragging the cohort upward" hypothesis from the 5/10 framing is not in evidence.

(b) $DRAM did not lag. Pullback 5/11 → 5/18: $DRAM -10.4% / $MU -14.2% / $HY9H.F -10.1% / $STX -11.2%. $DRAM tracks the cohort on the down-leg too — the cleanest visible pair to $MU.

(c) $DRAM correlation to $MU is tight. Across the 16 trading days, the $DRAM / $MU price spread holds in a ~5-point band; the $DRAM / $HY9H.F spread is wider (cohort leg-leader) but still bounded. No autonomous froth, no autonomous fade.

3 — Volume profile

$DRAM trading volume (daily, from stonks/data/stocks/DRAM/ohlc/3m.json):

Recent 5-day avg volume: ~30M. The 5/10 baseline noted 13.7M as "average" — that was the launch-period average; by 5/22 the trailing 30-day average is ~16M based on the 3m.json data. So today's volume runs ~2× the post-launch-normalized avg, not 2.9× the launch baseline. Volume is normalizing, not blowing off, not vanishing.

4 — $DRAM technicals (precompute snapshot)

From stonks/data/summaries/memory.json (precompute mtime 2026-05-23 03:24, data as of 2026-05-22 close):

• Price: $52.82
• RSI: 65.9 (strong but not extreme — sub-70 leaves room)
• SMA20: $47.71 (price +10.7% above SMA20)
• SMA50 / SMA200: null (insufficient history — ETF launched ~April 2026, ~36 daily bars)
• Trend: "neutral" (precompute classifier — SMA-stack incomplete)

The 20-day moving average ($47.71) is the only real technical anchor. Price has held above it for the entire post-ATH period; that's the cleanest support level for a trigger redesign.

5 — Coverage gap reminder (from 2026-05-26 weird-ETF consolidation)

DRAM remains absent from stonks/watchlists/etf-monster-universe.json (confirmed: grep -c DRAM data/summaries/etf-monster-universe.json → 0). It is in memory.json + etf-ideas.json + buffer-signals.json. The structural fix is filed as a separate follow-up under TASKS-RESEARCH.md:126 (the weird-ETF consolidation child task); this investigation does not duplicate it.

Refined trigger proposal

The original 2026-05-10 triggers were calibrated against a "$DRAM is an OTD-penny-stock-shaped autonomous flow signal" hypothesis. 12 trading days of evidence shows $DRAM tracks the cohort with high correlation. The refined triggers should reflect the actual behavior:

1. Replace "doubled in <30 days" with "$DRAM/$MU price-ratio divergence > 10%." $DRAM's autonomous-froth case is now a relative signal, not an absolute one. If $DRAM/$MU ratio expands >10% above its rolling 20-day mean, it's running ahead of the cohort = froth tell. If it compresses >10% below, it's lagging — ETF-redemption pressure or cohort doubt.

2. Replace "-20% with cohort holding" with "$DRAM closes below SMA20 for 5 consecutive sessions while $MU stays above its SMA20." The SMA20 has held as actual support across all 12 trading days; a 5-day break with cohort intact is the cleaner ETF-specific-froth-fade signal than a fixed -20% threshold. As of 2026-05-22, $DRAM is +10.7% above SMA20 — well clear.

3. Add "cohort-wide RSI > 75 across 3+ memory names" as the new up-trigger. This is the actual peak-cycle tell — the equivalent of the "doubled" trigger but expressed in cohort breadth. Today: $MU 66, $STX 61, $WDC 60, $HY9H.F 64, $SNDK 61 — all below 70. No tickers in extreme territory.

4. Retain "volume + flow are the tell" but make it operational. Today $DRAM 5d-vol-avg = ~30M vs 30d-vol-avg ~16M = ~1.9×. The flag should be: vol > 3× 30d-avg on a day where $DRAM diverges from $MU by >5% (the autonomous-flow tell that's both rare and informative).

Verdict + reasoning

• Both 2026-05-10 triggers retired as written. Neither fired in the 12 trading days they covered. They were calibrated against a frame that the subsequent data does not support ($DRAM as autonomous OTD-shaped froth).
• $DRAM remains in the standing-watch set. The thesis underlying the watch — memory-supercycle exposure via a parabolic-launch ETF — is intact. The watch frame just needs refinement against observed correlation behavior.
• Volume is normalizing, not blowing off. From 40M baseline-day to 21–30M recent — half the parabolic-launch volume, but still 2× post-launch-normalized average. Read: the parabolic-launch interest is real and persistent, but the volume velocity is decaying as expected for a maturing ETF.
• The cohort is at peak-but-not-extreme. Top RSI 66 ($MU), bottom RSI 60 ($WDC). All within 10% of 3m ATHs. The cohort holds the supercycle thesis but the breadth-extreme up-trigger has not fired. Continued monitor.
• One refined follow-up filed: retire the original triggers in the source row, replace with the 4 refined ones above, calendar-revisit in 2-3 weeks (mid-June) for the next status update.

Sources

• stonks/data/stocks/DRAM/ohlc/3m.json — 36 daily bars 2026-04-02 through 2026-05-22; price, volume.
• stonks/data/stocks/DRAM/ohlc/max.json — 8 weekly bars; weekly aggregates.
• stonks/data/stocks/MU/ohlc/3m.json — daily cohort comparison.
• stonks/data/stocks/HY9H.F/ohlc/3m.json — daily Korean HBM primary comparison.
• stonks/data/stocks/STX/ohlc/3m.json, stonks/data/stocks/WDC/ohlc/3m.json, stonks/data/stocks/SNDK/ohlc/3m.json — daily cohort comparison.
• stonks/data/summaries/memory.json — $DRAM precompute snapshot (price, RSI, SMA20, vs SMA20).
• TASKS-RESEARCH.md:113 — source row with original 2026-05-10 baseline + watch triggers.
• research/investigations/2026-05-26-weird-etf-desk-consolidation.md — companion read; $DRAM identified as standing-watch #2 of 5.
• research/investigations/2026-05-25-hbm-memory-supercycle-blind-test-verdict.md — underlying memory-supercycle thesis context.
• research/deep-dives/2026-05-25-ateyy-deep-dive-hbm-test-equipment.md — adjacent test-equipment layer; cited for cohort framing.

Carrier notes

• Data as of 2026-05-22 close (last row in 3m.json files). Precompute mtime 2026-05-23 03:24. No fresh fetch — this is a status-update read against cached state.
• Volume in the daily breakdown for 2026-05-12 marked "~47" because the precise daily figure was not surfaced in my eval slice but inferred from the surrounding weekly aggregate; the precise daily figure is in 3m.json if needed.
• $DRAM SMA50 / SMA200 remain null per the precompute; technical-trigger refinement uses SMA20 only until ~Aug 2026 (50 trading days post-launch).