Paulo Macro — research hub

Stock & name index

▲ Positive

► Neutral / referenced

▼ Negative

Overall thesis

• Oil — not what positioning looks like at a top; the covering short is the buyer. Sep-19 "Positioning in Oil and Nasdaq Is All Wrong." After the pullback from ~$105, the physical market still says tight: Dubai swaps over Brent "sticky bid," Dated Brent "at premiums above the worst of the Russia-Ukraine onset in 2022," landed China barrels "still notably bid" (no May-style flip to discount), and a curve "screaming at the world to bring any and all crude out of storage" while SPR releases have "slowed to 400kbpd due to physics." Vol and skew are asleep six weeks before the midterms ("Iran is never gonna really send it, are they?"). The new lens, from "The Gnome in Zug": Brent+WTI managed-money gross long / total gross is 77% vs north of 85% at a real high, the gross short is "far too big" for $100+ oil, above $100 the ratio "has only been this low once — last week," and prop/small traders are "even less committed"; WTI spec net long is below every major high of the decade. "I'm not saying oil has to roof. But I am saying that the current positioning is not what we see at tops… The most motivated buyer is the covering short seller." Aside: an uncommented QQQ chart with the 1M-3M ATM vol spread at a two-year low — the Aug-28 "near-dated vol is cheap" call, in the Nasdaq.
• Oil — the second mouse is about to get the cheese; the ETF plumbing says the crowd is short. Sep-12 "The Second Mouse Is About to Get the Cheese." The Jul-24 re-entry thesis gets a new layer of evidence. The "CTAs max long / beware TACO" worry "may have been a concern for me briefly several weeks ago, but no longer": managed money is far less stretched than earlier in the year with price "~$20/bbl from new highs." BNO is up 50%+ while AUM is only ~20% above the July lows, shares outstanding are falling (11.8→10.1mm in two weeks) and short interest is at a record ~2.2mm shares, over 20% of the float — the inverse of ARKK in 2022, when creations fed short sellers — so longs are redeeming while new shorts borrow from a shrinking pool: "the mechanical signature of 'the market is fading this rally'," with borrow "starting to look…provocative." Retail agrees: SCO takes inflows at fresh lows and is still ~50% bigger than BNO. Vols show "no fear" of an upside skid despite Bab-el-Mandeb and six months of storage depletion; managed money % of OI has room to 2017-18 levels while Brent OI is down ~25% (re-entry needs higher prices); respected oil traders have "lightened up significantly." "The sentiment and positioning is wrong here… I think oil is about to scream higher and am worried I don't own enough, and I'd like to think I'm pretty long." Macro: a stagflationary soft patch — ISM services prices paid says "CPI is about to pop," diesel above $6 at all-time highs, OBBBA rolling off, Boomer housing supply not clearing at 7% — maybe a brief Growth Scare; and a split Fed that won't hike into the midterms, so he wants to "receive the 2Y and SOFR" and is "uncomfortable in the long bond position" as 88% hike odds risk "a giant stop-out of flattener trades."
• Bonds as the hated asymmetric trade; the AI IPO as a two-sided supply signal. Sep-10 "Fly on the Wall with Shrub & Cloudbear" (with Le Shrub; speakers inferred). With the 30s near 5.3% and the 10s near 5%, Le Shrub keeps a small TLT option trade ("risking 30 bips total" for ~10×) because the whole Street is bond-bearish and Bessent is "two out of two" managing crude and the yen — "the trade is asymmetric," though fiscal and geopolitics make any fix short-term. On AI, the Anthropic IPO "has to succeed," so expect a month of bullish stunts; Paulo doubts it makes the window (DeepSeek at "1.4% of the cost," Congress, data-center NIMBY, a Democratic sweep) and reads a pulled deal as "a near-term low." His structural layer: 2026 re-runs 2023's bond script, and the West is "speaking Portuguese" — over-funded pensions defeasing into 5.5% long bonds, a Brazil-style real-yield bid that turns a dollar collapse into a slow bleed.
• Long palladium — the cheapest way to own an unpriced Russia risk. Sep-02 chat note. A trade idea from the day before, pulled forward on what "the market action could be telling us": "I think I may need to get long sooner than I thought." The mechanism is supply concentration, not demand — "Comes down to Russia on the supply side at ~40% of global supply" — married to a risk he thinks is mispriced at zero: "I think Russia is the most under-appreciated geopolitical risk on the chessboard right now particularly with elections coming up later in September (18-20th) and the uptick in noise around NATO flights and Russian hybrid warfare (Leipzig drones and diplomatic tensions). None of this is in commodity prices right here." The set-up came from FX: "USDJPY took a leg down earlier on rumors of small intervention while precious metals back on the menu" — note the adjectives, rumors and small, the same weak-sauce grading he applied to the BoJ's actual intervention in Aug-10. Carried on the index through PALL (the physical-metal instrument); he names the metal, not a fund, and keeps it labelled a trade idea — "food for thought," no size, no level. Same construction as his Nov-16 nickel thesis (concentrated Indonesia + Russia supply, rising event risk).
• Crude — the Jaws of Death have converged; the trigger is a failed Flush through $93. Sep-02. Re-running "one of my old favorites," he notes it is "interesting to see cracks and WTI back to trade directionally together" — the gap closing from the top, as it did in Jun-18/Jul-24. The confirmation is that leadership alternates: "as gasoline and diesel play daily ping pong with who leads, these products are dragging the crude complex slowly but surely higher" (one product ripping alone can be an outage; both taking turns is the refining system). The caveat and the trade are one sentence apart: "Still, everything is a Flush, which likely explains today's price action. A failure for the Flush to hold and a turn back through $93 would be quite bullish (see RBOB did something similar)." The bullish signal is not the flush — it is the flush failing and price reclaiming the level, with gasoline's completed version as the sibling read-across.
• Divergence check, second firing — buy cheap near-dated vol. Aug-28 "When Things Diverge — Subtle Shift Under the Covers": the recurring internals checklist he ran in February has flagged again. Level one, the indices: February's all-time highs were scattered across ~14 weeks (Nasdaq 100 Oct-29-25, Russell 2000 Jan-22-26, S&P 500 Jan-28-26, Dow Feb-6-26) and now "the indices have all scattered different highs again." Level two, the leadership: the Mag7 / hyperscaler / chip names are "all over the map" — META and MSFT peaked over a year ago, TSLA last December, NVDA and GOOG in May, even AVGO and MU peaked different weeks — i.e. the complex has been rolling over serially rather than topping together. He credits the dispersion tape as a genuine trader's market but rejects the bullish gloss: participation is "too quickly embraced as bullish by the 'Broadening Out' crowd here when looking simply at the S&P Equalweight Index" (the long form is his Feb-20 Broadening Out as Late Cycle). Leadership has changed, but rotation is only harmless if what you rotate out of is small — and "nearly half of the S&P remains AI-related with Mag7 over 30%" (chips near 20% at their highs). The trigger is credit: IG spreads for "what were previously considered 'asset light' businesses" are pressing higher in a reflexive loop with the US Treasury curve, into "an expected onslaught of future issuance just as managers are fully committed to equities and carrying low cash." Output is convexity, not direction — no index short, no target, no name-level call: "I think near-dated equity (and credit) volatility is grossly mispriced and cheap."
• The three regimes — pacification, claudification and MUM. Aug-10 "Fly on the Wall with Shrub & Cloudbear" (a conversation with Le Shrub, whose piece it walks through — views attributed by speaker). The organising claim: three forces all suppress volatility and, in doing so, build the instability. Pacification = passive's mechanical bid, plus the addendum that mega-IPOs now game it — "mark up in the private market with a view that you're going to put it in the index at a stupid valuation and the passive flows will maintain that stupid valuation… now we have like peak passive where we're getting trillion dollar companies added to the index on IPOs, right? Like SpaceX (SPCX)." Claudification = retail and pods asking the same models the same bottleneck questions ("find me the bottleneck in photonics… in power… in memory"), getting the same names and throwing them into the same baskets: "they think they're having alpha, but actually they're just doing momentum investing in a different word." Its first casualty was Leopold Aschenbrenner's fund at 4× leverage ("everyone sees his track record… and they just buy the same stocks at four times leverage and they all blow up together"); its institutionalisation is IBKR adding a ChatGPT portfolio function and Millennium working with Anthropic — "early innings or maybe mid… there's no free will out of all these guys." Paulo's synthesis: the mechanical/passive player and "the active slash retail monkey" are now herding identically. MUM = Markets Under Manipulation: "markets were always manipulated… but Yellen took it a step further with the QRA and the bond market in October 2023," and Bessent — "an experienced operator… he's been trading the yen all his life" — extends it to everything, because "traders just follow price and the price creates the narrative." Exhibit A is Hormuz, "the energy crisis that never happened": "20% of the world's energy flows were disrupted" and the best it produced was a two-month disruption and oil "spiking to 110 or something — below the 2008 level," amid signs of "suppression algos or active engagement… a team effort." Exhibit B removed the deniability: on the yen "they made it very explicit… we are coordinating with the Japanese treasury." The discipline (explicitly P&L-driven, "I don't want this to be like an academic paper"): step one recognise it, step two play along "90%, 95%, or 99%" — "if you went against passive over the last 10 years, you got annihilated; if you went against Claude over the last two years, you only made money after Leopold blew up" — and reserve the rest for MUM, the one discretionary regime with "a lot of precedents when central banks lose control." The dashboard: the yen (smashed 160→156 and "it's a 159 already" after the BoJ "tosses 50 yards plus and Bessent is in the mix" — "weak sauce"), the US 10-year (breaking out; a negative NFP and "bonds are flat… this never happened" — Paulo's "news failure" tell), and crude ("if there was suppression… it's like holding the ball underwater"). The damning context is the calendar: MUM should work best "when markets are most illiquid and quiet because it takes the least amount of money. And here we are in the dog days of August, and the best you can do is yen at 159… with the bond market breaking out." The credit half (Paulo): this is a debt-funded AI-capex market — last year the hyperscalers "were funding everything with their own cash," this year "there's zero buybacks. They've used all their cash flows, and they're borrowing on top," now even "sprinkling some equity in there alongside, like Google… moving down the capital stack because the market's just not cooperating" (their 2025 coinage: "AIBS — Artificial Intelligence Backed Securities. Or bullshit"). So equities and credit are tethered through the same handful of levered names that also drive "half of economic growth" — "that's where the Minsky moment always goes" — into "a ton of issuance coming in in September" with yields breaking out. His dating tools, crediting Kevin Muir and a credit-desk friend: for every "$20 billion of credit" the flippers "have to start making room sooner," so watch the ladder — "a big pop and run… turning into a pop and fade… into a weak pop… into just break deal price on day one" — plus tails in the auction and the desks' new-issue-versus-secondary tracker, which since "June, July, when momentum started to go down the stairs" has shown "existing bonds trade actually very poorly ahead of so much of this issuance… guys are starting to choke on the issuance." Rolling crack-ups (his name, "in honour of our friend Kevin Muir for his rolling bubbles"): gold/silver caned → long-short factor pairs blowing up on one day ("worst day in forever… since COVID") → the fixed-income guys — and the vector still standing is the yen: "why didn't the carry trade blow up… so far." The tail: a Saturday headline "nobody saw" that Katayama may be replaced by a little-known reflationist — "firmly Takaichi, make Japan great again"; if confirmed, "expansionary fiscal and an accommodative BOJ… depreciation big time and a rate blowout in Japan that only a few alarmists have been warning about," while consensus leans repatriation and FX vol sits near the lows despite the intervention ("not a base case for me"). The regime's own limit: "the cruiser of mum and pacification and claudification sits in front of like an aircraft carrier of policy… their mummification only works if the primary trend is going their way." Positioning: both are long and both are reducing risk into September — "it's been a good run… take chips off the table, give thanks… it's always end of August, September that bad things happen" — against "so much complacency." The expression (Le Shrub's "MUM trade," the successor to the bond vigilantes): "you buy oil calls, yen puts… and bond puts or NASDAQ puts or gold calls"; if control is lost, "commodities initially sell off… but then they rip… I'm not going to buy the S&P… they're just going to buy hard assets because what else can they buy?" — and gold and oil "just suddenly in the last week" have started trading together. The endgame: take passive "and turn it up to 11" — a narrowing "from Mag 7 to there can be only one" in NVDA, 8% of the S&P and the largest name again yet "forgotten" after missing "the big whoosh of Korea and Samsung and Hynex and Micron"; with so little volume relative to market cap in these inelastic passive-owned names, "we could see like that kind of last 50% move in a month, and then it's over" — a "trade after the trade, like later in the fall maybe," catalysed by Le Shrub's predicted Elon-Jensen "biggest circular deal of all times" (Tesla exclusive with NVIDIA; "it has to be like a trillion," "in the next few months") and marked, when it comes, by the "world's first $10 trillion company" headline — the successor to his 2021 "richest man, even for a day" (a Masayoshi-1999 rhyme) and last year's "world's first trillionaire."
• Mid-year 2026 — "a state of criticality." Jul-8 "Ramblings & Ruminations": the pieces of a forced unwind are all in place. The USD is a "consensus the market is not confirming" (Vanda: DXY positioning 92nd percentile, most bullish since Apr-2022, yet price nowhere near the 105-110 that level historically implied) — a Kovner-signal dollar bear market. He expects a Risk Off / Bonds Off / USD Off stress moment before Labor Day (the correlation-flip precedent: SVB spring-2023, Liberation Day, 1Q26-Iran all flipped the USD/long-yield correlation negative), with a "Summer 2024 Yenmaggeddon" yen at 162 + compressed FX vol as the trigger. Margin debt has peaked and rolled over (precedes every bear market this century); dispersion is at record highs vs correlation, single-stock vol at an all-time high vs index vol, S&P put/call skew 0.71 (lowest on record), the VIXEQ-vs-S&P correlation never more positive (gamma squeeze → vanna/charm downside). His private Carry Unwind Risk indicator sits at Jan-2018/Jan-2020 levels into July-August crisis seasonality (LTCM '98, Quant Quake '07, Taper Tantrum '13). Macro backdrop: jobs rolling over (a possible late-summer negative payroll print → the classic 4Q midterm-year easy-liquidity rally) into classic stagflation (supercore stuck 3-4% vs Warsh's 2%). The one active trade (sized as such, not an investment): long bitcoin (BTC) upside — selling exhausted, BTC holding up while risk frays, a possible weak-USD tailwind; MSTR is "a disaster… will end in bankruptcy," the eventual fuse. Platinum ETFs draining (>500koz since January; PPLT borrow ~12%) keep PPLT/platinum on watch for a move higher.
• The supply wall meets tightening funding. ~$200B of mega-IPO supply (SpaceX/SPCX, Anthropic, OpenAI) — ~$800B counting the later unlocks — is hitting just as leverage demand has pushed equity funding costs up; SPCX alone may close the ECM window. Jun-17 update: SPCX is now "sucking the oxygen out of the equity room" — options listing + Russell inclusion (days away) + Nasdaq (early July) + the "narrowing in" that final tops show set up "something truly acrobatic," a Volkswagen-2008 squeeze "×10-20" that "doesn't end until SpaceX is the world's largest company." He also hears SPCX coming to the IG bond market with $20-25bn of debt, ~25bps tighter than ORCL ("I would rather lend to SPCX than Larry Ellison") — a direct read on how stretched credit appetite is. The melt-up may be violent and real, but it's the structural top that ends the cycle.
• Fed-day reaction function — genuinely two-sided. New Chair Kevin Warsh: consensus fades him hawkish ("Warsh 2008," limited leash under Trump), but the just-in-time oil crash hands the Fed a dovish "Transitory / wait-and-see" out that "starts dipping into policy-mistake territory." Tie-breaker: "the 2yr is the scariest chart in finance and continues to lead the Fed." Watch the desync too — US "running hot" while the PBOC ("the Bundesbank") tightened in March on the oil spike (now softening Chinese data); if the PBOC loosens, base/precious metals get "some life" (and gold's earlier sell-off "was PBOC liquidity tightening all along," not the Gulf/Turkey headlines). Jul-30 verdict — "Wishy Washy Warsh": he expected a hike ("lot of tough guy talk on 2%," near-every modern new chair uses the first meetings to shake the tree "with exception of Arthur Burns," a committee of "vocal hawks and squeaky wheels") and got nothing — earning the Cloudbear moniker. Warsh "talks a 'trader/operator' game… wants to keep the market guessing and let it do the work ('play the ball, not the ref')," but the EM lens is the correction: "when you have become Brazil like we have, if you're… BCB's Roberto Campos… a former market operator… you know that doing nothing will be read as dovish no matter what you say about the future, especially if you say nothing about the future." The tape agreed — USD down, equities down, bonds down: "yet another Triple Yasu the likes we had not really seen so visibly since Liberation Day. Bonds were the real tell." Now he must "herd the cats" with "two months before you can begin to catch up to the market which is running well ahead." The path from here: "I think we stabilize a bit, but bonds will take Warsh at their word — 'you're the market, you do the tightening' — and the bond market gives him what he wants... good and hard."
• Semis rolling over + the datacenter-capex crack. GPU rental rates "rolling over/trading heavy" since Mythos; the datacenter construction-delay / cost-explosion story is "a problem for the bull narrative into the teeth of weakening pricing," and with index crowding-out from new paper issuance "there's room for shaking the tree between now and the fall." The concrete tell: MSFT walking away from a $3bn ORCL cloud-leasing deal — early evidence capex is cracking.
• Copper — uncomfortable but holding the juniors. "A bit uncomfortable but sticking with my juniors because drawdowns are a part of life." Two-sided: a PBOC loosening would put life in base metals, but "quite a lot of length in US copper futures" plus a datacenter-capex crack (the MSFT/ORCL tell) "could shake longs" — "a big pillar of the copper bull case is driven by datacenter rollouts." China builds its own; the sulfur issue. "Don't color me surprised if we go to $7.50 or $5.50." Gold/base metals "feel a little invisible" despite copper near all-time highs. (His copper-junior book: largest position ALDE, second SURG — Berg PFS "solid.") Framework (Oct-25 "Anatomy of a Breakout"): a real commodity breakout confirms via the whole curve — backwardation and a rising long-dated price (the 2004 oil template); LME copper is now doing this after two decades under ~$10-11k/t, RoW inventory (ex-Comex) at 2004-08/2021-23 lows, depressed copper/gold & copper/silver ratios flagging entry. Preferred expression = copper juniors into the Phase-2 M&A wave: ALDE, Taseko (TGB), and Surge Copper (SURG) — the Dec-04 origin buy at ~2-3% of NAV, moly byproduct the real draw, PFS due 1H26. Aug-03 upgrade ("Pick Your Fighter: Precious Metals vs. Copper"): the discomfort resolves into conviction — asked for his gold view, he picks copper, because "if positioning and flows are 90%+ of a story today" only copper's book is clean. The COMEX extreme (~$11bn spec net long, near record) is the wrong exchange to read: Trump's tariff premium (~3% COMEX over LME) has pulled ~700k tons in so that COMEX holds 74% of all world exchange inventory, "like uranium disappearing into the Sprott Trust… sequestered so long as Trump's threat of tariffs exists." Back it out and ex-COMEX stocks are "drawing off a cliff" against a seasonality that should flatline in 3Q; Morgan Stanley confirms Chinese inventories drawing fast, and the Yangshan cathode premium is blowing out with copper on the highs — the inversion of 3Q21/4Q22/4Q23 (premium widening then meant China buying a low), i.e. "China now appears to be short copper metal." The cash-3m spread has flipped back into backwardation, and 2026 mine-supply growth "rhymes with 'hero'" (Chile/Codelco "a tough slog", sulphur & acid post-Hormuz). The piece nobody looks at: LME open interest has crashed to 2022 bear-market territory and LME spec net long (contracts and $ notional) to nearly 2022-23 washout levels — with copper near all-time highs; combined with COMEX the aggregate is "not particularly extended." Ratio anchors: copper/gold below 4× implies ~$8/lb at $4,000 gold, and silver/copper above 9.5× reverting to its 5-7× band implies $8-11/lb at $58 silver — versus ~$6.50 spot (the Aug-2020 precedent: extended LME length didn't stop copper running another +50% to $10k/t). The squeeze template is silver mid-2025 — spec positioning peaked at $40 and fell as silver ran to $100+, "the squeeze was all commercial" — and today "specs don't have longs to sell to commercials if they have to scramble. Where is the paper or physical copper going to come from? … It's times like these when curves start to tighten that I smell an accident and like to rush in." He names the counter-case honestly: per GS the AI Data Center factor is now a primary driver of copper prices YTD, and the buildout proxies Caterpillar (CAT), Eaton (ETN) and GE Vernova (GEV) "have started to break down" amid the rotation out of semis (the 2020-21 tech-crowds-into-uranium rhyme) — "Copper is adjacent to this, and this is a real risk. The trade is not 'sleep at night'…but as a buddy once said: do the hard trade." His answer is location, not demand: "even if projected global balances were to notably soften… the metal is simply in the wrong place… and when China comes for something, they really come for it." Expression: advanced explorers / junior producers where "discounts have grown extreme vs large-cap copper multiples currently trading rather full at 15x+ forward PEs and 7-8x fwd Ebitda" — highly concentrated in ALDE and SURG, with very small tracking positions in TGB, NICU, IE "and a few others — inconsequential size for now." Illiquid: "I would not use market orders."
• Precious metals — constructive, but "not ready for primetime." Aug-03: temperamentally drawn to gold ("the ultimate 'opt out'"; Seawolf's Vinny Daniels and Porter Collins: "gold is my therapist"), but the 2Q oil near-halving taught him to refuse a trade that is "85% there" or where technicals and fundamentals conflict. The bullish half is real — COMEX gold open interest back to 2008 GFC levels, gold-miner ETF AUM (juniors especially) cut hard. The disqualifier is the capitulation test: since mid-May gold fell $4,500 → $4,000 while speculators increased their position, and spec length is longer now than after the January smash from $5,400 to $4,700 (managed money as a % of OI agrees) — "investors are buying into the decline rather than capitulating… this is not what capitulation and a longer-term low looks like," and global gold-ETF stockpiles that built all through 2025 are only just starting to roll over. Sentiment corroborates: a surge of bullish notes and inbounds, with the platinum-spring-2024 rhyme (Ferg and Le Shrub — "both worth subscribing" — were right on fundamentals and you still "had to wait nearly a year"): "when something suddenly becomes visible among people I respect, I get a little uncomfortable." Silver's book is "cleaner" (OI at lows unseen since 2008/2011, net longs near early-2024 levels when silver was in the low $20s) but as a % of OI still has room down, and the USD-index vs gold/silver ratio relationship says a Risk-Off rupture in the metals has gold outperforming silver — hurting the silver price outright. Verdict: "It's not that I don't like gold or silver, but gold in particular is not ready for primetime, and the move it had in 2024-2026 requires more digestion."
• Long nickel — the invisible, event-optionality trade. Nov-16 "The Invisible Metal": nickel is written off (4th straight surplus year, ~300kt inventory, 40% of output cashflow-negative, range-bound ~$15k/t) — but that base is the setup, and supply is dangerously concentrated (Indonesia >65% + Russia ~7% = >70%, "more than OPEC in 1973"). Indonesia is tightening its grip (illegal-tin crackdown as the template; now eyeing illegal nickel in Morowali; 1-yr vs 3-yr quotas), and laterite is oil-intensive into his bullish-oil view — any of which flips surplus to deficit. Skip the metal (contango roll); express via two sub-$2bn juniors that "pay you to wait" on their own catalysts: Talon Metals (TLO) (the only US nickel development, Rio JV, DOE/DLA, 1Q26 FS) and Magna Mining (NICU) (Sudbury; copper pays the bills, stockpiled-nickel option). Checklist: Jurisdiction ×3, Management, Alignment, Asset.
• Long platinum — a Dornbusch reversion off a real shortage. Jan-14 "A Dornbusch Moment" / Jan-18 update: platinum has doubled, but the driver is a genuine ~year-long physical shortage — Chinese premiums sky-high, 1-month lease rates ~19% annualized, and a profoundly backwardated London curve ("unheard of" in precious metals) — while positioning stays complacent (ETF holdings below the 2021 peak, spec net-long subdued, open interest flat = an unstable price-discovery condition). The margin-of-safety anchor is the platinum/gold ratio: ~1.2x over 40 years but at a ~70% discount now, so platinum "should double to $4,800 even if gold goes nowhere." Invoking Dornbusch ("longer than you think, then faster than you thought"), he expects the reversion in months/quarters. Expression: very long Valterra (VAL.JO) (Sibanye the higher-torque, more-Trump-adjacent alt), diversified with PPLT calls and hedged with platinum futures.
• Long natural gas — LNG + datacenter demand, traded around the weather. Jan-18 update: the durable thesis is tighter winter balances + growing LNG export capacity + summer datacenter gas-turbine demand; Antero (AR) is the largest, most-liquid, lightly-hedged expression (screen RRC, EQT [hedges a lot], CRK [growing], CNX). Near-term he trades the weather: a second Sudden Stratospheric Warming should drive cold blasts (~Feb 12 / ~Feb 23 / early-Mar) after a Feb 6-12 warm dip — take near-dated option profits fast, buy the dip; watch Permian/Oklahoma shut-ins (the Feb-2021 Texas-freeze analog). "Friends don't let friends trade natgas" — vol will be extreme; own the fundamental through the noise.
• Private credit has entered "Snapcount." Nov-17 "Credit Enters Snapcount, Part 3": private credit just had its April-2007 Bear Stearns-funds moment — Blue Owl (OWL), rather than meet 6% redemptions on its unlisted BOCCII fund, is folding it into listed OBDC (which trades at a -20% NAV discount), imposing a ~-20% "exit tax." "Public is the exit for private." The reflexive fear (if BOCCII, why not my Fund XYZ?) is corrosive because private marks — and AI-capex funding — hinge on fresh inflows, not just avoiding a run; the public/private mark divergence is the pressure gauge. Watch gate-or-dilute events and the discount-to-NAV on listed proxies. Dec-19 update ("An Emerging Credit Risk with Contagion Potential"): record-tight HY spreads (~275bps vs a model-implied ~600bps) are a flow artifact — private credit has crowded out HY issuance, so scarce paper + strong inflows pin spreads tight (HY CDS even trades a positive basis to cash). The catalyst to break it may be a supply shock, not defaults: ORCL (BBB / negative at S&P and Moody's, ~$101bn of bonds, its 30Y already ~275bps over Treasuries) downgraded to BB+ would drop $101bn into a ~$1.7T HY market (+6% overnight), forcing managers to "make room." Tail risk: a single ~$12bn HY fund reportedly levered 10-15x selling CDS (12-20% of the market — a Bear-Stearns-2007 analog). Jul-21 update ("An Emerging Credit Risk with Contagion Potential Revisited"): the ORCL call is "tipping over" — S&P downgraded ORCL to BBB-, debt grew $100bn→$120bn (coupon 4.4→4.8%), mcap $500bn→$365bn, equity below the March lows, while its 5Y CDS blew out to ~2× the BB CDX (bond ~200bps wide vs IG ~117, wider than BB corp at 154) even as BB spreads tightened — "on its own axis," the market "running well ahead of the agencies." A junk downgrade drops ORCL's $120bn into the HY market at ~7% of the index, and its 7.6yr duration vs HY's 2.9yr forces a steep-discount "air pocket" (UK-2061-Gilt analog) that "annihilates" the paper as funds make room — widening the whole HY complex into extended dealer balance sheets. The "Who's Next" contagion: another hyperscaler, or a BB-spread-driven refi issue. (ORCL is the hyperscaler most levered to OpenAI — $300bn+ Stargate/compute, ≥$90bn FY2027 capex.) Jul-30 update ("Important Credit Comment"): the follow-up adds the market-structure proof, via a credit specialist's diagnostic he had not been watching — new-issue bonds vs outstandings. The setup: fixed income is far more passive than equities (even "active" IG managers hug 6.3-7.0yr against a 6.6 index duration — "anything outside of that is too much career risk"), and new issues only enter the index at month end, so the marginal buyer of new paper is "active guys and fast money hedge funds" — making primary the cleanest read on real risk appetite. The rule that names the regime: in credit bull markets new-issue spreads tighten towards secondary; in bear markets secondary widens towards new-issue levels. UBS's weekly stats say it already flipped: new-issue premiums 7.0bps in July vs a 3.1bps 2026 YTD average (3.0 in 2025), oversubscription 3.1× vs 4.0× YTD — "falling off the cliff" — and July issuance $53.45bn against an initial estimate of $111.6bn, even with $83.4bn of net IG fund inflows YTD (the strain is in price, not demand). On the ground: data-center deals pulled because "the all-in yields are just not economical (can't make their IRRs work)", and a supposedly A+ credit (pseudonymized "Sopaipilla") forced to price at T+285bps, repricing the whole data-center space ~50bps. "Capital is unable to keep up with the pace of the exponential spending… investors are running out of balance sheet room to continue to fund this, and that is really quite scary." OWL confirms the money is leaving: Q2 fundraising $12.1bn→$7.6bn YoY with the credit business collapsing $5.8bn→$1.8bn, a three-year low — against a queue of project-finance debt on bank balance sheets still hoping to be refinanced into public/private credit and securitization: "the market is not closed, but it's sure going to extract a lot of concession to continue to stay open." And the base layer confirms it: NVDA's $250bn "Great Circularity" financing "has not been taken well and its CDS has joined the party" — widening on a risk-on day, the funding short moving from the income statement onto the balance sheet.
• The AI narrative has turned — GOOG long, NVDA funding-short. Nov-24 "The Worm Has Turned": Google's Gemini 3 "caught up and passed OpenAI" without NVDA GPUs or a Stargate/Oracle/Blue Owl financing morass — GOOGL (Berkshire bought it at ATHs; ~$70bn FCF/4Q) is the long, NVDA the funding short (revenue-vs-receivables divergence = rising cash burn straining the balance sheet; even an H200-to-China headline couldn't lift it — a "news failure"). ORCL is the datacenter-credit posterchild (CDS blown out, ABS off-balance-sheet debt, "the WorldCom of this cycle"). "Once you are on the backside of the narrative and flow mountain, it's a long way down." Positioning still has "room to go" (fear far from prior-crisis extremes), so beware the low-quality small-cap rotation that can "converge on the downside."
• Uranium — buy the AI-crossfire correction. Nov-24: uranium got hit as an AI/power-adjacent momentum trade (OKLO-meme flows via the URA ETF), but spot stabilized (Cameco (CCJ) + a utility bought $76-77/lb vs ~$85 term; Cameco borrowing pounds to deliver "can be played only so long"). Paulo "was buying again," betting a Strategic Uranium Reserve + a triple-digit US floor price (Mountain Pass template) becomes the "wedge" that separates uranium equities from the AI selloff. "Too much smoke for there not to be a fire." Jan-10/Jan-18 update ("Uranium Move Higher May Be Imminent"): reading SPUT (SRUUF)'s tape — a premium-to-NAV, 2x-volume day with only an $11mn ATM raise and zero pounds bought — as Sprott deliberately staying quiet to line up a large brokered follow-on before its base shelf expires ~Feb 3 (~$420mn "use it or lose it"). By Jan-18 SPUT is pro-rating January buys into an OSC shelf renewal and trading a persistent premium to NAV (largest since 2021), spawning long-spot/short-SPUT arb that pushed spot $77→$85; once reloaded it can hit the thin spot market in size. Yellowcake (YCA.LN) near a +5% NAV premium with a $100mn Kazatomprom call unused. "Uranium should continue to press higher."
• Leverage feedback loop. Gross leverage = f(funding cost, volatility); the first begets the second. Top-of-book S&P-futures liquidity at $3.5M and levered-ETF negative gamma at ~-$20B per 1% move (2× year-end) mean the derivatives tail wags the cash-equity dog until the position is rinsed — "hard to de-gross into a liquidity void." Jun-29 update: Friday's opex swung dealer gamma to its 3rd-most-negative reading on record (negative gamma → dealers chase price, amplifying two-way vol); paired with the funding squeeze, "it really is rare to see a setup so full of risk like this one" — he "really don't see how we get to Labor Day without a serious drawdown in equities, calls for Rotation notwithstanding."
• Equity funding is the tell. The cost of funding equity leverage (the total-return-swap rate; AXW the listed proxy) has exploded to new tights — and the diagnostics (swap spreads not more negative, dealer gross shorts flat) say it's raw leverage demand, not dealer-capacity. Levered-ETF swap capacity buckling (HK 2× Hynix at KIBOR +850; the MSTR-2× 2024 analog), PBs charging 3%+ on Asian-semi longs, record SPX call volume — "it's always funding constraints that end the party," so Momentum and Beta are at risk. Jun-29 update: into quarter-end (Jun-30) dealer balance sheets are behaving like extreme year-end tightness; the active S&P AIR TRF future (AXW1) sits ~120 on the Sept roll while the July contract is "screaming toward 180+" — leverage demand (levered ETFs/futures/margin) meeting IPO/ADR paper supply. He looks for some relief after Jun-30 (as after late-May) but expects a "reckoning… particularly around leveraged ETFs related to semiconductors."
• The Great Circularity — "Mark to Myth Goes Public." The AI complex's "cashflow" (LLMs and the picks-and-shovels hardware names alike) is VC-sourced cash recycled through capital raises — splendid until someone asks for their money back. OpenAI mulling a price war is one tell. May-15 ("Mark to Myth Goes Public"): the circularity has climbed from the revenue line into the earnings line. Hyperscalers (GOOG, AMZN, MSFT, ORCL, META) invest in private AI labs, mark up the stakes as each funding round prints higher, book the gains as "Other Income," and book cloud commitments from the same companies — Alphabet reported $38bn and Amazon ~$16bn of 1Q Other Income ($53bn ≈ 60% of their combined net income; $49bn from private-stake revaluations), while Anthropic ran $183bn→$380bn→a mooted $900bn round to fund $200bn/$100bn compute commitments to Google/Amazon. It's the private-equity overmarking playbook run through public companies ("Hyperscalers to PE: Hold my beer") — a self-validating ponzi flywheel ("if not expanding, collapsing"). What breaks it: reference marks the players can't set themselves — once OpenAI/Anthropic go public the quarter-end price becomes the "one true mark" and hyperscaler earnings reverse on any markdown; meanwhile they sit in a Prisoner's Dilemma where any single sponsor pausing an up-round "stops the music on Other Income growth." Morgan Stanley ("AI: Off Balance Sheet, On the Hook"): >$1.3tn of commitments, META ~1.7x / ORCL >7x fwd operating cash flow, most obligations off balance sheet, contracts underpinning asset-backed private credit. The memory leg is the same flow (Micron's parabola briefly topped Berkshire in the Russell 1000 Value; the neocloud node is CoreWeave); NVDA remains the base-layer funding short.
• Carry & the yen — the last un-blown-up leverage vector. Apr-27 ("JGBs, USD/JPY, US 10Yr, and Volatility"): despite a 2026 parade of quant blowups and factor unwinds, "carry has not had a chance yet to get in on all the fun." The JGB 10Y quietly broke to a new high while USD/JPY sat frozen 158-160 and its implied vol collapsed near 4-year lows. Paulo flags the long-standing USD/JPY–US 10Y relationship (diverging since Liberation Day) and the analogy to gold vs US real yields — a decades-long lockstep that broke at the 2022 reserve freeze. The tail he's open-minded to: a multi-year capital-flow regime change where a carry unwind in USD/JPY coincides with US bonds selling OFF (not rallying), implying USD/JPY vol (or UST 10yr vol, or both) is mispriced — "maybe 2022-23 was to gold vol what 2025-26 is to yen and 10Y vol." A musing, not a trade. Jul-30 update — the musing arrives: the BoJ "has finally gotten into the game after the US open", blasting USD/JPY with "a good old fashioned 'PKO' (Price Keeping Operation)" as the yen rallies over 5 big handles "and is still going." The timing was a policy choice, not luck: "makes sense they would wait for Warsh on rates before firing their bullets… why intervene in a sleepwalking Japanese rate regime if the Fed was going to rip rates in your face after you spend billions on an FX move?" The consequence is the point — FX vol was the one asset-class volatility still asleep ("one of these is not like the others"), and "the inherent instability layered in as FX vol finally — FINALLY — wakes up is only going to reflexively feed back into the already rising volatility in other major asset classes in one of the most illiquid months on the calendar": The Era of Rolling Blowouts. The first casualty is already down — Situational Awareness, a "$10bn+ fund" that went "from up double/triple digits to down on the year in a single month" in the largest Momentum factor unwind in history, has dumped all its listed investments "through one enormous trade" (~two-thirds of AUM was public equities; FT reports approaches to LPs to take positions directly). That is a "cleanup print" in old NYSE language, "so expect semis/AI to cop a bounce" — but an "'oversold' deadcat" one. The base rate he cites against calling it a top: "GS Global Alpha quants blew up billions 3% from all time highs in July 2007 — and we still made new highs thereafter."
• The convex oil trade. His largest position by a wide margin remains long Brent (BNO calls, Jul-26→Jan-27, for roll yield in backwardation): historic US inventory draws are "inevitable AND imminent" — Cushing toward minimum operating inventory (~17-20mmbbls) by month-end, an estimated -8mmbbl commercial draw last week (lowest for the time of year in over a decade), SPR flow slowing (~10→-6mmbpw). Price has flushed -30% (-20% on roll yield) while fundamentals went the other way, so the tape now looks like a capitulation: USO short interest at ~145% of shares outstanding is a synthetic retail short (~-9k WTI futures) the CoT data misses, spec net-long sits at 20-year capitulation lows with managed money adding shorts, and dealers short calls into a hollowed-out prompt market = negative-gamma kindling for an "upside crash." Dark transits are "a distraction"; the North Star is ~12mmbpd of shut-ins. The "Checkmate in Chinese" what-if — Xi lifting the product-export ban — could send crude to $150. Position remains long: "fire on the upside." Jun-17 update: stands by the -8mm commercial-draw call at the EIA (API agrees), but warns of a perverse "-$3 in the Great Upside Down"; "the real story is in cracks" — refiners minting, gasoline/diesel Jul-Aug spreads tightening despite a Gulf tropical-storm excuse: "products lead." Jun-18 update: refiner cracks are rallying while crude flat price collapsed — the "jaws of death" he expects to close as China lifts its product-export ban and SOE refiners run hard ("my catalyst… and I'm sticking to it"); gasoline front spreads (Jul–Aug) "twitching," PADD1 scarce, PADD2 tight, "draws everywhere." Crude sentiment is "ABYSMAL" but positioning hasn't capitulated to match (Monday's delayed CoT will tell) — "the setup for whiplash is there." Jun-30 update ("Oil & The China Syndrome"): after taking a drawdown he owns being wrong — an April "Kovner signal" (a bullish commodity consensus the market stopped confirming) he ignored — and names the two visible causes: China stepping >4mmbpd out of the market since April and specs collapsing to the 11th percentile of net positioning since 2011 (Brent long/short 2nd percentile). But he reads it as a bear trap and added to his oil exposure over the two-week decline: the world still draws >35mm bbl/week (Kpler -49mmbbl last week) even without China, and 3-2-1 cracks are the widest vs Brent since 1988 (a $56 crack implies $125+ crude) while flat price collapses into contango — a "never happened before" divergence where products lead. The swing factor is China's idle refining capacity (SOEs at ~upper-60s utilization vs low-80s; teapots as a "product SPR") — re-entry would tighten crude "almost immediately"; the mystery is why China won't buy despite every green light (an embargo trial-run à la WW2 Japan? an Iran/midterms play?). Gold is now "inextricably linked to Chinese PBOC liquidity" (Howell), and the PBOC ("21st-century Bundesbank") is injecting again since ~May 21st. "More questions than answers." Jul-24 update ("The Second Mouse Gets the Cheese"): the re-entry is on, and he names the archetype — "The Second Mouse Gets the Cheese": a second chance into a big move where the fundamentals are the same or better but the first crowd "was run over and killed in the mouse trap," so "positioning is cleaner and investors are more scared… a classic wall of worry." The Jul-14→21 COT is the proof: Brent $85→$91 (+7%) yet only ~30% of the record managed-money short cut by contract count and essentially none in notional US$ (flat price rallied +30% too), leaving the $ short "larger than the price collapses of 2024-25." Combined WTI+Brent net long ran ~US$64bn at the 1Q peak → $9bn in early July → only ~US$26bn now — a level "consistent with Brent mid-$70s in 2025 and low $80s in 2024," i.e. "speculators are 'lighter' at $91 than they were at $75-85" — the textbook Kovner "consensus the market is not confirming." Open interest confirms no fresh money (WTI OI at late-2025 lows; Brent OI -30% YTD "barely grown on the bounce"; Brent managed money 7% of OI = "upper $70s, not Tuesday's $91") — "existing (underwater?) longs selling to shorts who are covering." Physical is confirming in tandem (Dated Brent led the 2Q crash and revealed China's exit; the Dubai 2m swap over Brent beats the Hormuz dark-transit obsession), and backwardation in the 2-3m/2-5m/1yr Brent spreads sits at 2Q22 Russia-Ukraine levels when Brent traded $100-125 — "screaming for crude to come to market… the crude flat price and speculative positioning are too low." The "Jaws of Death" closes from the top: cracks lead, crude follows (March-April precedent: "crude caught up to cracks…and then continued to rally"), with cracks staying elevated on brutally tight product inventories — "as long as refiners are earning a $50+ crack, they will run every last barrel as hard as they can." He owns the full YTD scorecard (the January call + the Purim war call vs "crappy options and cash trades," missing the China move and under-appreciating how far positioning could reset). Conviction in a "Guns of August" blowout "has never been greater" into Trump's political vulnerability and the midterms ("the North Star persists"; "TACO remains absurd"). Oil is his largest position by a wide margin on a delta- and volatility-adjusted basis — calls and call spreads on BNO and related futures, expiries August→January. "As always, I would not listen to me." Jul-30 update: "Something is different about the oil market… I think it is about to scream out of control." The Middle East has gone to "a total black box" (soldiers' phones confiscated, media muzzled), so he reads the board instead of the tape: Dated Brent for immediate delivery and timespreads exploded higher as the front ripped overnight, while "the back end got smoked" as if a European producer deal had gone through. The unpriced supply stack nobody printed: Russia banned diesel exports through January; Iran claims another Hormuz crossing "caught fire"; a CPC facility and a Caspian tanker were struck ("Kazakh exports are again compromised"); a US-owned floating LNG storage vessel was struck off Egypt — "I see almost none of this in the mainstream press." The conclusion is a vol call as much as a price call: the market "feels like a beach ball under water… if I didn't know better I'd say there was a concerted effort to keep it riiiight where it is and pretend nothing is happening, because mechanically this thing looks like a volatility rupture in the making." Same framing extended to the long bond ("speaking of submerged beachballs everywhere").
• The Anatomy of a Crash — the equity endgame framework. Mar-14 "The Anatomy of a Crash" / Mar-23 revisit: markets crash from hyperextended positioning and unstable conditions, not euphoria — a four-stage pattern (initial correction + relief bounce → retest with a minor breach/intraday tail → a 1-2 day "all clear" final bounce → collapse; 1987, 1929, the 2021 bitcoin crash, the 1997 Hang Seng, now the Nasdaq 100 at 40% Mag7). The mechanism is a Rolling VAR Shock converging on every hedge-fund strategy at once: the last-unscathed Dispersion trade buckling as correlation rips (his own AirBNB put order re-racked above the mid at 55+ vol — options illiquidity worse than the $3mn top-of-book cash tape), the Basis trade leaking (deeply negative swap spreads), credit vol now LEADING Treasury vol (a credit and a rate issue), and no marginal buyer left (retail quiet, private wealth/long-only at record-low cash, sovereign wealth + Asian surplus economies potential sellers). He bought his first near-dated VIX calls since April 2010 and shorted Airbnb (ABNB) via June puts. Companion frameworks: the "North Star" (Apr-11) — a Nash-equilibrium read that Iran's optimal strategy to constrain Hormuz doesn't change regardless of Trump, so the market's "TACO"/look-through reflex is a Misconception awaiting a "Tom Hanks Has Covid" flip when "Can't Print Molecules" overruns the print/spend/negotiate narrative — while the Apr-15 TGA rebuild removes the liquidity tailwind ("Sell in May"); and "crack spreads are the tell" (Apr-07) — refiner margins are the second-derivative that, inverted from the Covid-2020 playbook (a supply shock, not a demand shock), point to an oil "upside crash" rather than a collapse. His retail tells stayed on: ~$1bn crowding the broken -2× SCO (+ a lopsided USO short) and the Avis (CAR) single-stock squeeze jamming the Dow Transports (a false-breadth 2021-repeat top tell).
• Near-term: another skid lower. A broadly "massacred" week (long/short, long-only, retail, China/HK, EM all hit "even if the S&P didn't show it"); opex reset positioning, quarter-end looms, liquidity is "dreadful" — "we are close to another skid lower." His topping tell: "new highs into weak divergence" on thin breadth/volume — Exhibit C = MSCI Korea (EWY). "Feels like we are close to something." Jul-23 ("everyone getting the business"): the dispersion rotation ("semis on, Mags off") is over — everything sells off at once. GOOG "beats" and raises capex but prints FCF -$6bn with 87% of 2Q "income" from marking up its own Anthropic/private stakes (the Great Circularity confirmed), while the Cembalest/JPM "train" warns the front end (hyperscaler FCF plummeting) rolls over before the caboose (still-soaring semis) "gets the memo." No Risk-Off hiding spot — the 30Y knocking on 5.20%, gold/metals sold, credit widening and BDCs back in the crosshairs; TSLA -13% / SPCX ~80x sales the "two disasters." "This tape feels ugly… a lot of chop and whiplash."
• Positioning is out of control — grab protection, revisit the Rollover Syndrome. Jan-18/Jan-20: the Jan BofA survey is "eye popping" — cash 3.2% (record low), a net 48% carrying no downside protection (highest since pre-Volmageddon Jan-2018), 9% recession odds (lowest since Jan-2022), "no landing" the base case for the first time in three years; GS shows fundamental L/S gross at the 100th percentile while systematic L/S draws down; Citadel shows retail options above 2021's meme peak. With a JGB blow-up (30Y +27bps overnight) and a "Triple Yatsu" (equities/bonds/dollar down), he grabs protection and expresses downside via the Rollover Syndrome (ROS): "tops are a process," so short S&P names within ±1% of a flattening 200-dma acting sickly ("shoot them when they're running away"). Candidates: EQIX, MCD, QCOM, ETN, BRK.B, META, a software basket (IGV — WDAY/NOW/DDOG/INTU/ADSK, +CRWD/PANW — after Anthropic's Claude Cowork "demolished" SaaS), payment rails V/MA, and CLO/BDC "Kingmakers" APO/BX/KKR. The trigger he watches: rising S&P correlation into Tech earnings + rising bond vol squeezing collateral-levered gross. "A few snowflakes start an avalanche."

Notes

One dated page per note — each has its stock table, talking points, and the saved note text. Newest first.

To process — backlog

Paulo Macro notes not yet processed — newest first. None queued yet.

For personal study — not investment advice. Source material © Paulo Macro.