The short answer
Because memory-stock options are priced for calm while the stocks themselves are still moving violently. As of the August 10, 2026 close, at-the-money implied volatility for Micron, SanDisk, Western Digital and Seagate sat below each name's own 20-day realized volatility — by 29.5, 47.6, 20.3 and 3.4 vol points respectively.
That sign is backwards. Normally implied runs above realized, and the difference is the premium option sellers collect (the variance risk premium, or VRP). Right now all four are negative, and all four are at their most negative reading in the available history.
Key qualifier: every "record" in this piece refers to 95 observations from 2026-03-30 to 2026-08-10 (66 observations from 2026-05-07 for Seagate). That is not one year, and not all history. Full caveats in the last section.
1. The core data: implied vs realized
Figures below are on an August 10, 2026 close basis. The options snapshot is collected pre-open on 08-11 and carries the 08-10 settlement state.
| Ticker | ATM implied vol | 20-day realized vol | VRP (IV − HV20) | IV position in available range |
|---|---|---|---|---|
| Micron (MU) | 80.0 | 109.5 | −29.5 | 5.3% |
| SanDisk (SNDK) | 106.8 | 154.3 | −47.6 | 7.0% |
| Western Digital (WDC) | 88.9 | 109.2 | −20.3 | 9.5% |
| Seagate (STX) | 86.2 | 89.6 | −3.4 | 6.3% |
HV20 is computed independently: annualized 20-day standard deviation of log returns, window ending at the August 10 close, excluding the unfinished August 11 bar.
How unusual is this
| Ticker | Observations | VRP negative | Prior most-negative | Current | New low for window |
|---|---|---|---|---|---|
| Micron (MU) | 95 | 20 (21%) | −23.1 | −29.5 | Yes |
| SanDisk (SNDK) | 95 | 15 (16%) | −46.7 | −47.6 | Yes |
| Western Digital (WDC) | 95 | 14 (15%) | −16.5 | −20.3 | Yes |
| Seagate (STX) | 66 | 1 (2%) | +2.9 | −3.4 | Yes (first ever negative) |
Seagate deserves separate attention: across 66 available observations its variance risk premium had never been negative, with a floor of +2.9. This is the first negative print on record.
Speed of the flip
Micron's premium completed a full round trip in two weeks:
| Date | IV | HV20 | VRP |
|---|---|---|---|
| 2026-07-30 | 139.7 | 92.9 | +46.8 |
| 2026-08-04 | 112.3 | 109.2 | +3.1 |
| 2026-08-06 | 115.4 | 111.8 | +3.7 |
| 2026-08-08 | 91.1 | 110.3 | −19.3 |
| 2026-08-10 | 80.0 | 109.5 | −29.5 |
Note which side actually moved: realized volatility barely budged (92.9 → 109.5, in fact higher), while implied volatility collapsed (139.7 → 80.0). This is not "the stock calmed down." It is "the options market conceded calm first."
2. One explanation we tested and rejected
The easiest objection: could this just be an artifact of the contract rolling to a new expiry?
Worth checking seriously, because in the August 11 snapshot every tracked name rolled its nearest expiry:
- Micron, AMD, Broadcom, Intel: 2026-08-12 → 2026-08-17
- SanDisk, Western Digital, Applied Materials, KLA: 2026-08-14 → 2026-08-21
The roll came with a collapse in volume:
| Ticker | 08-08 contracts | 08-11 contracts | Change |
|---|---|---|---|
| SanDisk (SNDK) | 54,031 | 14,575 | −73% |
| Micron (MU) | 29,855 | 10,913 | −63% |
| Western Digital (WDC) | 29,722 | 7,495 | −75% |
| AMD | 15,988 | 3,432 | −79% |
Result: implied vol survives the test, put/call ratio does not
Grouping the full history of the four memory names by days-to-expiry, average ATM implied volatility:
| Days to expiry | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
|---|---|---|---|---|---|---|---|
| Average IV | 116.2 | 118.5 | 115.8 | 117.3 | 118.7 | 118.2 | 114.9 |
Flat. There is no systematic difference between 3-day and 9-day ATM implied vol (range 114.9–118.7). So Micron's 91.1 → 80.0 decline is not a roll artifact — it is a genuine repricing. The core finding holds.
But run the same cut on the put/call ratio (full US sample, since July):
| Days to expiry | 3 | 5 | 9 |
|---|---|---|---|
| Average P/C | 1.274 | 0.942 | 1.269 |
| Average volume | 25,933 | 10,990 | 3,994 |
The put/call ratio depends strongly on days-to-expiry, and the far contract carries roughly one-sixth the volume of the near one. Any P/C comparison spanning a roll date is therefore unreliable — including narratives about "memory put/call falling while non-memory semis jump." Within that very group, Applied Materials fell from 1.51 to 0.40 while KLA rose from 0.32 to 1.66 — opposite directions, which is what noise looks like, not structure.
Methodological takeaway (citable): the put/call ratio of a single nearest-expiry contract is not comparable across a roll date. Align days-to-expiry first, or use a full-chain weighted measure instead.
3. Intraday August 11: which side are the stocks confirming?
Figures below are live intraday readings as of 10:37 ET on August 11, 2026. US markets are still open; these are not closing prices.
| Ticker | Price | vs prior close |
|---|---|---|
| Micron (MU) | 860.30 | −0.08% |
| SanDisk (SNDK) | 1,260.54 | +1.83% |
| Western Digital (WDC) | 442.65 | +0.98% |
| Seagate (STX) | 813.50 | +1.56% |
| Semiconductor ETF (SMH) | 577.26 | +1.38% |
| Nasdaq 100 ETF (QQQ) | 721.76 | +0.12% |
Scaled by each name's own HV20, every one of these moves sits inside a single daily standard deviation (roughly 6.9% for Micron, 9.7% for SanDisk). One hour of trading can neither confirm nor refute the options market's judgment — a move inside the normal band carries no information by construction.
What is worth noting is the sign pattern: three memory names higher, Micron alone marginally lower, with the semiconductor ETF up in line. That is sector beta, not a memory-specific move.
4. How this divergence can be falsified
A negative premium can only converge in two ways, and both are observable:
- Realized volatility falls toward implied — Micron's HV20 moving from ~109 toward 80 across three consecutive sessions.
⇒ The options market was right, the turbulent phase is over, and the negative premium closes on its own. - Implied volatility rises back — any memory name's IV position moving from single digits back above 40%.
⇒ The options market was wrong and is repricing the next leg.
Until one of those happens, "options are cheap" is a reading, not a conclusion. It says the disagreement between the options market and the cash market about the next one to two weeks is the widest on record for this window. It does not indicate direction — a negative variance risk premium tells you who is mispricing volatility, not whether the stock goes up or down.
5. Caveats and limitations
The boundaries matter more than the finding:
- The history window is short. Options data begins 2026-03-24: 95 usable observations for Micron/SanDisk/Western Digital, 66 for Seagate. Every "record low" and "never negative" is a statement about that window — roughly 4.5 months, not one year.
- Beware the field-name trap. The underlying database field is named
iv_rank_252, implying 252 trading days, but it is actually a min-max normalization over available observations (verified: Micron 80.0 → 5.3 matches (x−min)/(max−min) across all 95 observations exactly). This piece calls it "position in available range" throughout. - Snapshots carry a collection-date vs settlement-date offset. The snapshot labeled 08-11 is collected pre-open and holds the 08-10 close. All figures here are aligned to settlement date.
- The window is too short for backtesting. 95 observations cannot validate any predictive power of a negative premium, and these months represent a single market regime. This is a current-state assessment only, with no historical hit-rate claim.
- Intraday numbers move. Section 3 is a live reading and may be stale on publication; use closing prints for any settled comparison.
FAQ
Q: What does a negative variance risk premium mean?
A: Implied volatility is below realized volatility. In plain terms, options are priced for smaller moves than the stock is actually making. Historically the reverse is normal — options run slightly rich, and that difference compensates sellers for bearing risk.
Q: Does this mean options are cheap and should be bought?
A: This piece gives no trade recommendations. A negative premium only says the disagreement between options and cash about future volatility is the widest on record for this window. It does not indicate direction, and 95 observations cannot support any hit-rate claim.
Q: Why can't the put/call ratio be used as corroboration?
A: Because every name rolled its nearest expiry on 08-11, volume fell 63%–79%, and the P/C ratio is systematically dependent on days-to-expiry (see Section 2). Comparisons spanning a roll date do not hold.
Q: Which of the four is the most extreme reading?
A: By absolute size, SanDisk (−47.6). But by rarity, Seagate — it had never printed negative across 66 available observations, and this is the first.
Sources: options snapshot (Tiger, August 10, 2026 settlement basis), independently computed HV20 from daily bars, live quotes (Tiger, 10:37 ET August 11, 2026). Public-layer market-structure analysis; contains no position or trade recommendations.