U.S. Pre-Market Brief | 2026-08-21 (ET) Friday

Coverage window: 2026-08-20 16:00 ET regular-session close → 2026-08-21 08:16 ET pre-market.

Quote conventions (read timestamps are annotated item by item; this brief does not mix reads taken at different moments):

• Single-stock pre-market prices, prior closes, 52-week highs/lows and pre-market volumes are taken from the stockanalysis.com quote API, read at 07:59–08:15 ET. For key names this brief took three reads at 08:02 / 08:09 / 08:15, using "is pre-market volume still accumulating" to judge quote freshness — volume not increasing means a stale snapshot, marked "(frozen)" in the text.
• Treasury yields, VIX, commodities, FX and crypto are taken from the CNBC quote API, read at 08:05–08:16 ET.
• Gold/silver/copper are continuous futures contracts (@GC.1 / @SI.1 / @HG.1) and carry a basis versus spot — this morning gold futures are roughly $100–110 above spot; do not mix them with spot figures when citing.
• Crude is given for both WTI (@CL.1) and Brent (@LCO.1); the spread today is about $7 and the two carry different narrative implications, see §0⑥.
• Index pre-market direction is proxied by ETF (SPY/QQQ/DIA/IWM) pre-market prices, not futures quotes.
• Earnings figures are taken from the SEC 8-K Ex-99.1 originals and the conference calls; links are in the text.
• ⚠️ Convention disclosure for consensus estimates and sell-side price targets: the consensus figures cited here (ROST $1.94 / Q3 $1.74 / full year $7.78), the sell-side mean price target (FCX $71.73, 23 firms), forward P/E and CME FedWatch probabilities all come from secondary aggregator sites, with no data vendor or scrape timestamp attached. In particular, whether FCX's $71.73 already reflects post-8/6-earnings revisions could not be confirmed in this brief — if it is a pre-earnings stale value, then the argument "the share price has already run past the sell-side mean target" does not hold. Treat conclusions involving that figure as "pending verification."

---

⚠️ Read this first: today's biggest cognitive trap is "hard assets are all making new highs" — and it is wrong

This morning the dollar fell and gold, silver, copper and bitcoin all rose; the media wrote it up as "the debasement trade breaking out across the board." Trade off that framework and you will get the direction wrong.

Put the current price of the same set of assets next to each one's own historical high and the picture is the exact opposite:

Asset	Current price (pre-market)	All-time / 52-week high	Distance from high	Character
FCX (single stock)	$74.10	52-week high $72.275 (6/17 intraday spike)	+2.5%	⚠️ Alleged breakout, see below
COMEX copper @HG.1	$6.611/lb	52-week high $6.70 settlement / $6.87 intraday (8/5–8/6)	−1.3%~−3.8% (convention unsettled)	Prior high not taken out
LME 3-month copper	$6.401/lb	Record settlement $6.56 (8/6, primary-verified in FCX's 10-Q)	−2.42%	Prior high not taken out
Gold @GC.1 (futures)	$4,654.00	ATH $5,589.38 (2026-01-28)	−16.7%	Bounce off a deep decline
Silver @SI.1 (futures)	$69.65	ATH $121.62	−42.7%	Bounce off a deep decline
Bitcoin	$76,817	ATH $126,080 (2025-10-06)	−39.1%	Bear-market bounce
GLD / SLV / IBIT (ETF cross-check)	421.94 / 62.82 / 43.52	509.70 / 109.83 / 71.82	−17% / −43% / −39%	Same as above

Table scrolls horizontally

Not one commodity is at a record high. Precious metals topped on 2026-01-28 and then went through "the most violent selloff in 13 years" (gold fell $380 in 28 minutes; silver was cut in half from its peak), and bitcoin fell as much as roughly 54% from its 2025-10-06 peak.

⚠️ And FCX's "breakout," on inspection, does not hold — this is the single most important self-correction in this brief:

1. $72.275 was an intraday spike on 2026-06-17; that day closed at only $69.06, giving back 4.4% intraday. FCX has never traded above $72.275 in a regular session. Using a rejected spike as the breakout reference is invalid on its face.
2. The pre-market $74.10 is built on about 210,000 shares, just 1.7% of the 20-day average volume (12.31 million shares). Yesterday's regular-session close was $71.22, still 1.46% below that spike.
3. And copper itself is still −1.3%~−3.8% (convention unsettled) from its own record (COMEX 52-week high $6.87, 8/6); LME 3-month copper at $6.401 is −2.42% from the record settlement of $6.56 on 8/6 — a record confirmed by FCX's own 10-Q text: "on August 6, 2026, the LME copper settlement price closed at an all-time high of $6.56 per pound."

So the correct statement is: copper is not at a record today, FCX is — and that gap is not being supplied by copper. Over the same stretch, pure-copper name SCCO is still about −6.4% below its own 52-week high (that read comes from a frozen pre-market quote on 9,000 shares and is directional only). This price level has to be carried either by FCX-specific logic (the Grasberg production-recovery option, exclusive benefit from Section 232 tariffs) or it is a price level with no fundamental support underneath it. Falsifiable test: whether today's regular-session close can genuinely hold above $72.275.

This distinction determines the nature of the position: gold, silver and bitcoin are on day 3 of a bounce inside a deep bear market and can only be traded as a bounce; while the copper miners' "breakout" has not been confirmed in a regular session and needs to be verified with the copper price and the closing price, not self-validated on thin pre-market quotes. Chasing a bounce as if it were a breakout is the easiest way to lose money today.

---

0. Today in one line

① Today is day 3 of the "fiscal easing vs. a hawkish Fed" mispricing, not day 1 of a new catalyst. On 2026-08-19 (Wednesday) the U.S. Treasury announced it would at least double the size of each long-end buyback operation from $2 billion to $4 billion (targeting the 10–20y and 20–30y sectors, effective 9/9). This is the master switch behind every hard-asset move.

② One premise has to be corrected first or the whole brief reads wrong: the market is currently pricing the September FOMC as "hold vs. hike," with no cut on the menu. The fed funds rate is now 3.50%–3.75%; CME FedWatch prices the 9/16 meeting at roughly 70% hold / 30% a 25bp hike / 0% cut (the prior day: 68.4% / 31.6%, essentially flat day over day). The July FOMC was a 9–3 hold, with three regional Fed presidents voting to hike; the July minutes show "further tightening would likely be needed if inflation does not come down." The current Chair is Kevin Warsh, who took office on 2026-05-22 — and it was his nomination that set off the precious-metals crash in January. Loosening on the fiscal side, hawkishness on the monetary side — that dislocation is the root of dollar weakness and the bid for hard assets.

③ This morning's genuinely new marginal news is not in the U.S., it is in Japan. Japan's July core CPI came in at +1.8% y/y (prior 1.6%), and core-core, excluding fresh food and energy, at +1.9% (prior 1.7%); the BOJ's 9/17–18 meeting is widely expected to raise rates to 1.25%. The direct counterparty to DXY at 98.681 (−0.22%) is right here: USD/JPY 158.67 (yen stronger), EUR/USD 1.1701. ①② above are carry-over narrative; ③ is what actually happened today.

④ Crypto is the most genuine leg of liquidity this morning, and the drivers are more solid than "a short squeeze." BTC $76,817 (24h +5.76%, 08:16 ET), roughly +20% on the week. The sequence is: 8/18 the SEC released the 402-page "Regulation Crypto Assets" proposed rules → 8/19 the White House crypto summit, with Trump urging the Senate to move the CLARITY Act → 8/19 the Treasury doubled buybacks → real money into spot ETFs ($517 million on 8/19, the largest since early May; another $500 million-plus on 8/20) → triggering over $3 billion of liquidations within 24 hours (of which BTC $1.45 billion, 92% shorts). The squeeze is an amplifier, not the initiator. But for today, all three catalysts are already 2–3 day-old news and should be priced as "carry-over catalyst persistence," not treated as fresh positives today.

⑤ And crypto's internal structure today is the opposite of yesterday's — the single most worthwhile verification point in this brief. Yesterday was a textbook squeeze pattern of "junk outperforming quality" (MARA +15.54% vs HOOD −0.70%); this morning it has flipped — the quality end leads with HOOD +4.74%, COIN +4.58%, CRCL +5.19%, IBIT +5.63%, while the miners lag: RIOT +2.67%, CIFR +1.63%, WULF +1.58%. And crypto is the sector with the most visible pre-market volume accumulation this morning (IBIT across three reads: 14.04 million → 14.19 million → 14.35 million shares).

⑥ One fact easy to get wrong that this brief specifically corrects: energy is not absent, you were just looking at the wrong contract. WTI @CL.1 $86.80 (−0.03%) is indeed flat, but Brent @LCO.1 is $93.83, up roughly +5% on the week, against the backdrop of the U.S./Israel conflict with Iran and the situation in Hormuz. Both the Hang Seng and DAX wraps list "oil above $93" as a factor suppressing risk appetite. Looking only at WTI produces the wrong conclusion that "energy isn't following, so the debasement narrative has a crack" — Brent says the opposite.

⑦ Only two earnings reports, one a genuine beat and one with a guidance problem. ROST +7.30%: Q2 EPS $2.66 includes roughly $0.60 of tariff refunds, but ex-refund EPS ≈ $2.06 still beats the $1.94 consensus by about 6.2%, with comps +10% driven by traffic — this is not a fake beat stacked up out of refunds (see §5①, which includes an important layer of quality discount). OSIS −9.39%, approaching but not breaking its 52-week low: FY27 revenue guidance of $1.875–1.93 billion implies only 4.7%–7.8% growth over FY26's $1.79 billion.

⑧ Pre-market state: SPY +0.43%, QQQ +0.67%, DIA +0.47%, IWM +0.71% — broadly higher but mild, not enough to recover yesterday's losses (yesterday SPY −0.84%, Nasdaq −1.00%, Dow −1.32%). The 10-year at 4.69% (prior close 4.698%), the 30-year at 5.241% (prior close 5.237%), the 2-year at 4.183% — yields have barely moved this morning. This matters enormously: today's hard-asset rally is dollar-driven, not driven by falling rates. And the Treasury buyback's "push yields lower" leg has essentially stopped working — the 30-year has gone from 5.18% right after the announcement back to 5.241%, giving back about 6bp of the 8bp decline on announcement day (about three quarters), though not yet fully erased. VIX 15.50 (−3.19%).

⑨ The only macro event today is the August Markit Flash PMI at 09:45 ET: manufacturing expected around 53.9–54.0 (prior 53.9), services expected around 53.9–54.0 (prior 54.6), both essentially expected to be flat. There is no Michigan consumer sentiment today (the August final is on 8/28; the August preliminary was released on 8/14 at 51, a large miss). Jackson Hole 2026 is 8/27–29, not today, and Warsh's debut as Chair is on 8/28 at roughly 10:00 ET, on the theme of financial innovation's impact on payments and policy. If you see claims like "Powell speaks today" or "waiting for signals on the rate-cut path," they are recycled stale copy or contradict FedWatch pricing — do not rely on them.

---

1. News overview

Overnight Asia and Europe

⚠️ Data-collection failures and implementation details for this run:

• stockanalysis.com's bulk screener endpoint (/api/screener/s/f) is now returning 404 across the board, work/pm_mv.py is broken, and the pre-market gainers/losers list cannot be pulled in bulk. Switched to the web pages /markets/premarket/gainers|losers/ plus per-symbol /api/quotes/s/<sym> (the latter works). Recommend fixing pm_mv.py soon.
• SEC EDGAR returns 403 to WebFetch from the main process (a sub-agent did retrieve ROST's 8-K, which suggests it is intermittent/UA-related); CNBC article pages 403; the OSIS IR site times out on WebFetch (60s). OSIS therefore rests on cross-checking Seeking Alpha plus the Benzinga call transcript, with no 8-K original obtained — that is the evidence-strength weakness of the OSIS section here.
• yfinance hit YFRateLimitError on the second call from the sub-agent side, with .financials/.quarterly_financials returning empty; ROST financials were instead completed via the SEC 8-K original plus three cross-checks reconstructed from the statements.
• Because the bulk lists broke, large-cap moves were swept via "pull 62 major names one by one + threshold screen" (result: no unexplained moves other than BURL's 127-share noise). This supplementary sweep is what guarantees this brief's coverage — do not drop it in the next version.
• The FCX/copper sub-agent still had not returned at press time; the Gresik smelter's current restart status and the impact on FCX production guidance are unresolved, and are explicitly flagged as open questions in §5②.

---

2. Strongest themes, descending

---

3. Single-stock strength ranking

Pre-market changes uniformly use the last read at 08:12–08:15 ET. "(frozen)" = pre-market volume did not increase between reads, i.e. the quote is stale, the gap size is low-confidence, and it must be re-pulled before the open.

Positive side

Negative side

---

4. Single-stock scoring model (100 points)

---

5. Detailed analysis of top names

① ROST — Ross Stores | Priority deep-dive | Total 73

• Related news: 08/20 16:05 ET, SEC 8-K Ex-99.1¹ (primary source, not a secondary retelling).
• Catalyst logic — the conclusion first: this is not "a beat manufactured by a one-time refund." Q2 EPS $2.66, of which roughly $0.60 / $253M came from IEEPA tariff refunds, against original guidance of only $1.85–1.93 and consensus of $1.94. Ex-refund EPS ≈ $2.06–2.07 (after-tax $189.8M ÷ 319.45 million shares = $0.594/share, hence $2.66 − $0.59), still about 6.2% above consensus and about 6.7% above the top end of its own guidance. The company's own framing is internally consistent too: of the +610bp of operating margin, 405bp came from the refund, and ex-refund it is still +205bp, above the original plan of +130–150bp.
• The revenue-side beat cannot be explained by an accounting item: revenue of $6,264.9M (+13.3%, 1.8% above consensus), comps +10% (guidance only +6–7%), explicitly driven by traffic — traffic is the hardest metric in off-price retail to fake. New customers span income brackets and age groups.
• ⚠️ But one more layer has to be peeled back, and this is the most important reservation this brief holds on ROST: within the ex-refund +205bp, only 110bp is merchandise gross margin (the most sustainable component), while another 100bp comes from distribution costs — and two of the reasons the company itself gives are timing-related/one-off in nature (the timing mismatch on packaway expense, and the low base from last year's tariff-handling costs). Extrapolating the full +205bp as a new structural margin would be a second layer of overestimation.
• And the "clean versus clean" year-over-year is far less flattering than it looks: last year's Q2 $1.56 itself included −$0.11/share of tariff costs. So on a comparable basis it is $2.07 vs $1.67 = +24.0%, not the reported +70.5%. Of the $1.10 of year-over-year EPS growth, about 65% relates to the one-off swing in tariffs.
• Guidance (more important than the quarter's beat): Q3 EPS $1.75–1.83 (consensus $1.74, only 2.9% above); full year raised to $8.61–8.77, with the CFO volunteering an ex-refund basis of $8.01–8.17 on the call (consensus $7.78, 4.0% above). That translates into a net raise of roughly +$0.185 for the second half; measured against the company's old guidance from 5/21, the H2 raise is +$0.29 / +8.0%, and that portion contains no refund at all — a genuine second-half upgrade.
• Two-year stacked comps: Q1 17%, Q2 12%, Q3 guided 13.5%, Q4 guided 13.5%. The accurate statement is "H2 guidance is not below the Q2 just completed," not "no deceleration anywhere" — Q1→Q2 itself decelerated by 5pp — the single-quarter comp falling from +10% to +6.5%/+4.5% is a base effect (last year's Q3/Q4 were +7%/+9%), and on a stacked basis it is in fact slightly above the Q2 just completed.
• Fundamentals: H1 operating cash flow $1,711.8M, net income conversion to cash 114%; FCF $1,251.6M. Net cash of +$3.27 billion (excluding leases); but on a basis that includes $3.72 billion of operating lease liabilities it is net debt of $451 million — both conventions have to be stated, not just the flattering one. ROE roughly 39–43%. Two negative details to watch: ① inventory +18.4%, faster than revenue at +13.3% (the non-packaway portion +22.2%); ② with comps at +10%, the SG&A ratio actually deteriorated by 15bp — no operating leverage on expenses.
• Valuation — there is a denominator trap here: at the pre-market $245.70, trailing P/E on the refund-inclusive $8.27 is 29.7x, but $8.27 contains $0.60 that will not repeat; on the ex-refund $7.67 it is 32.0x, and that is the comparable number. By the same token, computing PEG on the reported basis systematically understates the valuation. 52-week position 91%, only −4.4% from the $257 high.
• Attribution of the +7.3% gap: the refund itself contributes only about +0.24% to intrinsic value (after-tax $189.8M ÷ roughly $79 billion of market cap); the Q2 operating beat about +0.05–0.5%; the net H2 guidance raise about +2.1–3.4%; the remaining 4–5pp is multiple re-rating — a bet on the durability of the "comps +10%, traffic-driven" curve, which is the most fragile segment of the gap, backed by no Q3 data at all and facing a +7% base in Q3.
• Peers not following: TJX +0.58%, BURL +2.03% (127-share noise). The market is pricing this as company alpha rather than sector beta — neither easily dragged down by the sector nor lifted by sector money.
• Final judgment: of the two companies reporting today, ROST is the one with all four boxes ticked — primary source + its own catalyst + still beating after stripping the one-off + a genuine second-half raise — and scores first. (Note: today's sample is only ROST and OSIS, so this should not be over-extrapolated.) But half of the upside is multiple re-rating rather than earnings, and it has already gapped +7.3% pre-market. The 42,000 pre-market shares are thin volume; the exact move should be taken from the open.

2Freeport-McMoRan | Watch onlyFCXTotal 52 · cut two notches from the initial "priority deep-dive"

This name was cut two notches over the course of writing this brief. Verification overturned both pillars of the initial read — "copper at a record high + miner breakout" — and the reasons are all primary sources (FCX's own 10-Q and SEC XBRL).

• Related news: copper futures $6.611/lb (+2.20%, 08:06 ET). Note: no company-level catalyst for FCX could be found within the 8/20–8/21 window — no new 8-K, no IR press release, no confirmable rating change. Precisely stated: FCX's last 8-K stops at 07/23/26; since the 10-Q filed on 8/06 there have been no new filings at all. ("Not found" ≠ "does not exist"; this item is marked uncertain.)

⚠️ Refutation one: the so-called "breakout" does not exist in a regular session. $72.275 was an intraday spike on 2026-06-17; the close that day was only $69.06, giving back 4.4% intraday. FCX has never traded above that price in a regular session, and yesterday's close of $71.22 is still 1.46% below it. This morning's $74.10 is built on 220,000 shares (1.7% of the 12.31 million-share 20-day average). Using a thin-volume pre-market quote to confirm a rejected spike is doubly unreliable.

⚠️ Refutation two: Gresik is not a double-edged sword; for FCX it is clearly negative — and deeper than the first draft assumed. FCX's 10-Q (filed 2026-08-06) confirms three things point by point, and in combination they are lethal:

1. The PT Smelting facility that was hit is a wholly-affiliated smelter 66%-owned by PTFI (Mitsubishi Materials holds 34%) — this is FCX's own capacity, not a competitor's.
2. PTFI's concentrate export license expired on 2025-09-16, and the 10-Q text describes PTFI as now being "a fully integrated producer" — i.e. there is no legal channel to sell concentrate externally.
3. PTFI's own Manyar smelter has not yet restarted ("expected to restart operations in the second half of 2026 at reduced rates").

So since 8/8, PTFI simultaneously has: (a) its only operating smelter down, (b) its own smelter not yet restarted, and (c) no concentrate export license. The Indonesian sales channel is ≈ completely severed. Reading this as "a competitor's output cut benefits FCX" is a factual error — this is FCX being unable to sell its own material. Roughly 100,000 tonnes of concentrate are reported to be stockpiled and unprocessed.

How large is the exposure (10-Q primary): within full-year 2026 sales guidance, Indonesia accounts for 675 million pounds of copper (22.1%) and 654 thousand ounces of gold (nearly 100%).

⚠️ Refutation three: guidance was set two days before the accident, has not been updated since, and is extremely back-loaded. Q3 guidance (copper 750 million pounds / gold 160 thousand ounces) and full-year guidance were set on 8/6, and the accident happened on 8/8; the parent has said nothing at all and has not updated guidance. Meanwhile H1 actuals were only 1,367 for copper and 244 for gold, so hitting the full year requires H2 copper +23.6% and gold +68% sequentially — guidance is heavily back-loaded, and the accident landed precisely on the back-loaded portion. The guidance already embedded roughly 100 million pounds of copper and 50 thousand ounces of gold of smelting-stage deferral; the accident can only make that number larger.

⚠️ One more reason has to be added that the first draft of this brief missed entirely, and it is more fundamental: the Grasberg mud rush of 2025-09. Copper sales −27.6%, gold sales −62.5% and the "heavily back-loaded guidance" are primarily explained not by the smelting channel but by the capacity ramp following that accident. The 10-Q text gives an explicit path: "PTFI's overall production rates are expected to approximate 65% of capacity in H2 2026, 80% by mid-2027." In other words, the back-loading of H2 is a ramp curve the company has already disclosed, not a hidden fragility — the first draft of this brief wrote it as the latter, which was wrong and is corrected here. The correct framing of Gresik is: an incremental risk layered on top of that already-known ramp — what it threatens is precisely whether "65% in H2'26" can be delivered.

Downgrade note on the "historical precedent": the first draft used "concentrate piled up to about 400,000 tonnes in early 2025, forcing PTFI mining down to about 40% of capacity" to rebut the PTFI CEO's claim that "Grasberg output is unaffected." On review, neither the timing nor the figures in that passage could be confirmed against a primary source (the 10-Q records the Manyar fire as October 2024 rather than early 2025, and the word "Manyar" does not appear anywhere in the 10-Q; the 400,000 tonnes, the 40% capacity figure and the current 100,000-tonne backlog all lack a linked source). This brief downgrades that entire passage to "disputed, pending verification" and no longer uses it as an argument. But "the export license expired on 2025-09-16" and "PTFI is now a fully integrated producer" remain primary 10-Q records, and the conclusion does not depend on the downgraded passage.

Fundamentals: volumes are down double digits, and the profit growth also contains a one-off (SEC XBRL + 10-Q primary)

	2026H1	2025H1	Change
Revenue	$13,263M	$13,310M	−0.4%
Copper sales (million pounds)	1,367	1,888	−27.6%
Gold sales (thousand ounces)	244	650	−62.5%
Realized copper price ($/lb)	6.04	4.48	+34.8%
Net profit attributable to parent	$1,865M	$1,124M	+65.9% (includes a one-off, see below)

Table scrolls horizontally

⚠️ The +65.9% in this table cannot be used at face value. The 10-Q income statement carries a standalone line, Gain on PTFI mud rush incident insurance settlement $699M, entirely in Q1'26. An assumption-free falsification: Q2 standalone net profit attributable to parent was $984M vs $772M last year, only +27.5% — H1's +65.9% was lifted by that one-time insurance settlement in Q1 (the portion attributable to the parent roughly estimated at $200–260M, about 27–35% of H1's $741M increase). Therefore: ① this brief withdraws the claim that "all profit growth came from price" — it also came from a one-time settlement; ② TTM EPS of $2.04 likewise contains that gain, and ex-gain is about $1.86, implying a P/E of about 38 rather than 34.9. Until a precise decomposition is done, treat the earnings and valuation multiples in this section as "pending verification" in their entirety.

• The revenue-side facts are unaffected: TTM revenue growth is only +0.19%, while volumes are down double digits — revenue only held flat because price (realized copper +34.8%, gold +44.3%) offset the −27.6% drop in copper volumes and the −62.5% drop in gold volumes.
• Unit costs are deteriorating, and the non-GAAP convention hides part of it: Q2 unit net cash cost $1.97/lb vs $1.13 last year (+74%); and that convention already excludes $0.40/lb of idle-facility and repair costs, so adding those back the true figure is $2.37 vs $1.13, i.e. +110%. This is negative operating leverage: fixed costs spread over a third less output, with price alone holding the margin up.
• Minority interests take roughly 36% of after-tax profit (the Indonesian government's MIND ID holds 51% of PTFI), and this is the main reason for the chasm between a 27.26% operating margin and an 11.38% net margin.
• The balance sheet is not the risk point: net debt $6.28B, debt/EBITDA 1.06×, interest coverage 16.55×. But the quick ratio of 0.79 deserves attention — current assets depend heavily on inventory, and a smelter shutdown is precisely what makes inventory keep swelling and tie up working capital.

Valuation: no cushion at all

• P/E (TTM) 34.9 @$71.22 → 36.3 @$74.10; P/B 5.09; EV/FCF 61.6.
• A forward P/E of 21.1 implies forward EPS of about $3.37, 65% above the TTM $2.04 — the market has already priced in a full Grasberg ramp plus current copper and gold prices simultaneously. Neither has happened yet, and one of them (reaching the 65% ramp in H2'26) is being directly threatened by the Gresik event.
• A PEG of 0.66 is a convention error here: its "growth" comes from a production recovery off a disaster low plus an assumed commodity price, neither of which is compoundable growth. Applying PEG to a cyclical miner is misreading a cycle as growth.
• ⚠️ The share price has already run past the sell-side mean target: the mean across 23 firms is $71.73, and the pre-market $74.10 is above it. (A statement of fact, not a recommendation.)
• An overlooked tension: Chile's Cochilco officially forecasts copper at $5.10/lb for 2027, 23% below today's spot.

The parts of the bull case that still hold (not to be ignored): ① 77.9% of copper (2,382 million pounds) has zero connection to Gresik yet fully enjoys the copper price increase — every $0.10/lb rise in copper is worth roughly +$238 million of revenue, nearly all of which drops into gross profit; ② the Indonesian portion is deferred (the metal is still there and may be sold at higher prices), not destroyed, and FCX's phrasing "timing of sales realizations" means exactly that; ③ Section 232 is FCX's exclusive option — the 10-Q states that FCX supplies roughly 70% of all U.S. refined copper output, and the refined-copper tariff step-up of 15% from 2027 and 30% in 2028 is still open. This cleanly explains "why FCX is stronger than SCCO," and SCCO does not have it.

• Final judgment: watch only. The real question is not "good or bad" but "will deferral turn into lost production" — and only three things decide it: (i) whether the PT Smelting repair completes within Q3; (ii) whether Manyar restarts on schedule at the end of August/September; (iii) how long the concentrate stockyard holds out. Until those three are clarified, at a level that is already past the sell-side target, on a P/E of 34.9, a P/B of 5.09 and volumes down 27.6%, the risk/reward of building a position on a "breakout" does not stand up. ⚠️Special note: if news later appears that "PTFI is applying for a concentrate export license," that is itself confirmation the stockyard is running tight, not a positive.

⚠️ Two sub-agents gave mutually contradictory numbers on LME inventories; this brief could not settle it and has avoided the indicator in the text:

• The macro agent (citing Business Recorder 8/21): inventories +11% week over week to 235,975 tonnes, squeeze easing;
• The FCX agent: inventories 214,550 tonnes, down more than 35,000 tonnes (−14%) since 7/31, backwardation widening, tightness intensifying. The two point in completely opposite directions, possibly because of different read dates or conventions (LME registered warrants vs total inventory). The next version must pull the official LME daily report itself to settle this and not rely on either secondary retelling. Also: the COMEX copper record high has three mutually exclusive secondary versions ($6.87/8-6, $6.77/8-7, $6.714/8-12), and the official CME endpoint returns 403; the likely cause is differences in contract month and intraday vs settlement conventions. This brief's conclusion is insensitive to that (today is below the record under all three versions), but where a specific figure is cited, the CNBC exchange feed's $6.87 was used.

③ HOOD — Robinhood | Watch closely | Total 64

• Related news: no news of its own, crypto sector beta. BTC $76,817 (+5.76%).
• Catalyst logic: recovering crypto volumes map directly into commission and matching revenue; the SEC rules and the CLARITY Act's progress favor compliant platforms' product expansion.
• ⚠️ A precedent from just yesterday that must be written into today's judgment: yesterday HOOD opened +5.31% ($100.85) and closed −0.70% ($95.10), open→close −5.70%. Today it is gapping again by almost exactly the same amount at +4.74% ($99.61) — the second occurrence of the same pattern.
• But there is one key structural difference today: yesterday the crypto sector was "junk outperforming quality" (MARA +15.54% vs HOOD −0.70%), a textbook squeeze; this morning it is reversed — the quality end HOOD/COIN/CRCL/IBIT leads while the miners lag, RIOT +2.67%, WULF +1.58%. And today has something yesterday did not: two consecutive days of real spot-ETF inflows above $500 million each. If that structure holds during the session, it says the character of the money has shifted from "short covering" to "active allocation," and yesterday's open-high-close-low will not simply repeat. This is the verification point most worth watching today.
• Pre-market and technicals: +4.74%, pre-market volume of 1.5 million shares and still accumulating, the most solid among crypto single stocks. Still −35% from the 52-week high of $153.86.
• Final judgment: watch closely. Yesterday's pattern suggests the price at the instant of the open is not necessarily the more favorable level of the day. Yesterday's lesson was "the open was the high"; the reasonable approach is to wait 30 minutes after the open and see whether the quality end keeps its lead over the miners before deciding.

④ IBIT / ⑤ COIN / ⑥ CRCL — the crypto complex | Watch closely

• Driver chain (the order matters): 8/18 the SEC's 402-page rules → 8/19 the White House summit + the Treasury doubling buybacks → real ETF inflows ($517 million on 8/19, the largest since early May; another $500 million-plus on 8/20, over $1 billion across the two days) → triggering over $3 billion of liquidations in 24 hours (BTC $1.45 billion, 92% shorts). The squeeze is an amplifier, not the initiator — which, compared with yesterday's recap characterizing it as "pure squeeze beta," is a view that needs correcting.
• But three things to stay clear-eyed about: ① BTC is still −39.1% below the 2025-10-06 ATH of $126,080, with a maximum drawdown of roughly −54% in this bear market — this is a bear-market bounce, not a bull-market breakout; ② all of the catalysts above are already 2–3 day-old news, with nothing new today; ③ the CLARITY Act is far from passed — the Senate procedural vote is on 9/15, the threshold is 60 votes and Republicans hold only 53 seats, still 7 votes short; do not write it up as "about to pass."
• Internal divergence: BTC +5.76% clearly outperforms ETH +2.08%, whereas yesterday ETHA led at +10.52%. The main line has switched to the BTC side; ETH is yesterday's story.
• Liquidity evidence: IBIT pre-market 14.35 million shares, CRCL 2.57 million, COIN 1.05 million, and all still accumulating between reads — crypto is the sector with the highest "both price and volume are real" quality this morning (GDX 1.01 million, MU 818,000, HMY 484,000 and AG 281,000 pre-market shares are also accumulating, but with less magnitude and persistence).
• Judgment: watch closely. Its character is a bounce rather than a reversal, and that should show up in the holding period and risk budget, not be loosened because "the debasement narrative is grand."

⑦ GDX and the precious metals miners (NEM/AU/HMY/AG/KGC) | GDX watch closely, the single stocks mostly watch only

• Drivers: gold futures $4,654.00 (+1.81%), silver futures $69.65 (+2.27%), the dollar −0.22%. Mind the basis: spot gold is roughly $4,530–4,554, about $100–110 away from futures; do not mix them when citing.
• A characterization that has to be spelled out: gold is still −16.7% from the 2026-01-28 ATH of $5,589.38, and silver is still −42.7% from its ATH of $121.62. The fuse for January's crash was precisely Kevin Warsh's nomination as Fed Chair, and he is now sitting in the Chair's seat (took office 2026-05-22), with the September FOMC still carrying roughly a 30% probability of a hike. This bounce is fighting a structural pressure that has not gone away, not riding it.
• And these names already rallied once yesterday: AU +6.08%, KGC +5.26%, HMY +4.41%, GDX +2.59% (against a backdrop of the S&P falling 0.87% yesterday). Today is the second straight day.
• Judgment: the sector's direction is right and the position is bad, so most get "watch only." Comparing within the sector, GDX (an ETF, diversified, and with the best liquidity at 1.01 million pre-market shares) beats the individual names on liquidity and diversification; HMY (+7.13%) has the most elasticity but is a high-cost South African mine and will fall faster once gold pulls back.

---

6. Negative/avoid list

Two counterpoints on shorting OSIS (not to be ignored):

1. This is a "timing" problem, not a "demand" problem. The company explicitly stated that the delayed deliveries have not lost any orders, remain in backlog, and have had their delivery schedules updated, and the record backlog of roughly $1.9 billion — a figure higher than the entire FY27 revenue guidance itself — constitutes a fundamental floor.
2. The pre-market quote is thin. OSIS pre-market volume was frozen at 13,203 shares between 08:01 and 08:09, only rising to 17,683 shares by 08:13; the −9.39% rests on extremely thin turnover, and the real decline after the open could deviate materially. Do not build a position straight off that number.

⚠️ The OSIS 8-K original was not obtained (SEC 403 + IR site timeout), so the figures above come from cross-checking Seeking Alpha and the Benzinga call transcript, with evidence strength below ROST's. Before actually trading the name, verify the 8-K yourself.

---

7. Within-theme ranking

Theme 1: Crypto/digital assets

Theme 2: Copper/industrial metals

Theme 3: Precious metals miners

---

8. Opening verification signals

Pre-market (09:00–09:30 ET)

1. Re-pull prices: the pre-market quotes for SCCO / TECK / HBM / ERO / KGC are frozen and must be re-pulled before the open; do not trade off the numbers in this brief.
2. The only valid test for FCX is "can the regular-session close hold above $72.275" — the pre-market price does not count (just 1.7% of average daily volume), and $72.275 itself is a spike that gave back 4.4% that day. Not closing above it = the "breakout" never happened.
3. Whether BTC can hold $76,000; a break below downgrades the entire crypto line.

Intraday (09:30–10:00 ET)

4. ⚠️ Today's first verification point — the crypto sector's internal structure: can the quality end (HOOD/COIN/CRCL) hold its lead over the miners (RIOT/CLSK/WULF)?
   • Holds → the character of the money has shifted from "short covering" to "active allocation" (two consecutive days of $500 million-plus ETF inflows support that reading), yesterday's open-high-close-low is less likely to repeat, and the quality end's relative strength stands.
   • Reverts to yesterday's pattern (miners and MARA leading) → it is still the tail end of a squeeze, the quality end will most likely fade from the open again, and HOOD/COIN's pre-market gaps should be treated as the day's high zone.
5. ⚠️ Today's second verification point — FCX's excess return relative to SCCO. ⚠️ Convention reminder: SCCO's −6.4% rests on a frozen pre-market quote of 9,000 shares and cannot serve as a precise denominator; the inheritable benchmark is yesterday's regular session — on 8/20 FCX +3.08% (16.06 million shares) vs SCCO +2.08%, an excess of about 1pp. Compare today's regular-session gap between the two against that 1pp.
   • If FCX continues to clearly outperform SCCO today → the market is pricing FCX's unique logic (the Grasberg ramp option, exclusive benefit from Section 232 refined-copper tariffs), which is legitimate, but be aware Gresik is directly threatening the ramp.
   • If FCX underperforms SCCO or both pull back together → the premium is starting to be taken back by Gresik's volume problem, the premium is being reclaimed, and FCX's relative position within the sector is then the most fragile.
   • Reference: on 8/20 in the regular session FCX +3.08% (16.06 million shares) vs SCCO +2.08%, an excess of about 1pp — this morning's pre-market excess is far larger than that, but on thin volume.
6. ROST gap-fill test: if it breaks below the opening price within the first 30 minutes, the market has read the layer that "half the upside is multiple re-rating."
7. 09:45 ET Flash PMI — the only U.S. macro event today (manufacturing/services both expected around 53.9–54.0, essentially flat):
   • Strong PMI → reinforces the 30% probability of a September hike, the dollar bounces, and the damage is greater to precious metals/crypto than to copper (copper has independent support from tariffs and demand).
   • Weak PMI → the dollar keeps weakening and the hard-asset complex accelerates; but watch for equities flipping to recession pricing (the August Michigan preliminary of 51 was already a large miss).
   • Note the asymmetry: the services prior of 54.6 is above the top of the expected range, so services is the one more likely to "miss."

Options and sentiment

8. VIX 15.50 (−3.19%): already back at lows, so there is limited room for further risk-appetite repair, and this should not be positioned as a "panic-bottom big rebound."
9. Post-earnings IV crush in ROST: going long ROST via options today means eating the cost of IV collapse, and option longs bear that additional IV-collapse cost — not something to ignore when comparing position instruments.

Risk signals

10. Divergence alert: if SPY's pre-market +0.43% is quickly erased after the open while hard assets keep rising, that says money is rotating out of equities into hard assets rather than risk appetite recovering broadly, and hard-asset strength should not then be read as the broad market turning strong.
11. Oil is an underrated suppressant: Brent $93.83, roughly +5% on the week, and both the Hang Seng and DAX wraps list it as a suppressing factor. If Brent keeps rising, it is simultaneously negative for duration (inflation expectations) and for consumer names while being positive for energy — do not ignore this line just because WTI is flat.
12. Friday effect: not a triple-witching day, but weekend-holding appetite is low, and late-day profit-taking pressure is especially visible in precious metals/copper names on their second straight up day.
13. Next week's calendar pressure: Jackson Hole 8/27–29, Warsh's debut as Chair on 8/28, and the 9/16 FOMC arrive in sequence. The only sentence the market will trade is "is the hiking cycle paused or over" — hard-asset positions built today need to be able to survive next Thursday and Friday's event risk.

---

9. Final conclusions

① The 5 names most worth watching today

② Today's 3 strongest themes

Copper has been moved out of the top three. The initial read had it as the number one theme; after verification it was cut to number 4: copper is still −1.3%~−3.8% (convention unsettled) from the COMEX record and LME 3-month copper is −2.42% from the record settlement, while FCX's "new high" is a thin-volume pre-market confirmation of a rejected spike. What this theme lacks today is not direction but position and confirmation.

③ What to avoid today

• OSIS: guidance miss + approaching the 52-week low. But note for shorts: the record $1.9 billion backlog (higher than the entire FY27 revenue guidance) is a fundamental floor, and the −9.39% rests on thin volume.
• MARA / BMNR / SBET and the other squeeze protagonists from yesterday: yesterday's biggest gainers are no longer in the lead today. Chasing yesterday's champion on day two of a squeeze is the most textbook way to lose money.
• ⚠️ The copper sector as a whole (including FCX): this is the most heavily revised item in this brief. Copper is −1.3%~−3.8% (convention unsettled) from the COMEX record; SCCO/TECK/HBM/ERO pre-market quotes are broadly frozen; and while FCX has the best liquidity, it carries the highest fundamental risk in the sector — the Gresik shutdown severs its Indonesian sales channel (22% of copper, nearly 100% of gold), Q3 guidance was set two days before the accident and has never been updated, H1 copper volumes are already −27.6% with true unit cash costs +110%, and the share price has already run past the $71.73 mean target across 23 sell-side firms. "Copper going up" and "buying FCX" are not the same trade today.
• Semiconductors (MU/SMH): pure sentiment follow-through today, no catalyst of their own.
• Chasing any precious metals name: the direction is right but the position is bad, it is already the second straight up day, and next week's Jackson Hole and September hike pricing lie ahead.

④ The final one-line judgment

Today is day 3 of the "fiscal easing (the 8/19 doubling of buybacks) vs. a hawkish Fed (a 30% September hike probability still, with Warsh in office)" mispricing, and what actually happened this morning is that Japan's stronger CPI lifted the yen and pushed the dollar down — so split the tape into three parts: crypto is the segment with the most solid evidence (the SEC's 402-page rules + two consecutive days of $500 million-plus ETF net inflows, with the squeeze only an amplifier rather than the initiator), but it is still −39% from its ATH and can only be traded as a bounce; ROST is the only one of today's two reporting companies that carries an SEC primary source and still beats consensus by about 6% after stripping the one-time tariff refund, except that 4–5 percentage points of its +7.3% gap are valuation re-rating rather than an earnings upgrade; and copper — the most important self-correction in this brief — was wrongly ranked as the number one theme in the initial read: copper is still −1.3%~−3.8% (convention unsettled) from the COMEX record, FCX's "new high" is a 220,000-share pre-market quote (1.7% of average daily volume) confirming a spike that gave back 4.4% that same day, and more importantly the Gresik shutdown severs FCX's own Indonesian sales channel (22% of its copper, nearly 100% of its gold), while Q3 guidance was set two days before the accident and has never been updated, H1 copper volumes are already −27.6%, and the share price has already run past the sell-side mean target. The only macro gate today is the 09:45 ET Flash PMI (the services prior of 54.6 makes services the one more likely to miss); the two structural verification points are: crypto's "can the quality end keep leading the miners" — if it can, yesterday's open-high-close-low does not repeat, and if it cannot, this batch of +5% pre-market gaps turns into the day's high again; and FCX's "can the regular-session close genuinely hold above $72.275" — if it does not close above it, today's so-called copper breakout never happened.

---

⚠️ Risk disclaimer: this list is pre-market information triage and observation only and does not constitute investment advice. U.S. equities are volatile and pre-market gaps are risky; post-earnings IV crush and guidance reversals occur; automatically generated content may contain stale information or factual errors. Rely on company disclosures/SEC filings, and do not use this directly as a basis for trading.