Title: PCE, Jackson Hole, Implied Bet — Weekly Positioning Decode (24–28 Aug 2026)
Author / source: Stochastic Volatility — Market insights (stochvoltrader Substack)
Source title: PCE, Jackson Hole, Implied bet | Weekly post (24-28/Aug)
Source date: Sunday, August 23, 2026, 7:34 PM ET (same clock as America/Toronto)
Source URL: https://stochvoltrader.substack.com/p/pce-jackson-hole-implied-bet-weekly
Memo date: Monday, August 24, 2026 (America/Toronto)
Week covered: Monday 24 Aug – Friday 28 Aug 2026
Source type: Substack weekly implied-bet / dealer-positioning post. Primary source is the full weekly file only. Author instructed readers to finish a prior context piece before this post; that piece is not a source of facts for this memo except where this weekly post itself points at it.
Companion (read first, per author): /workspace/pm-memos/2026-08-23-alma-bessent-put.md — Alma, The new Bessent put, Oil, Midterms and EoY regime (Sun 23 Aug 2026, 6:11 PM ET). Underlying file: /workspace/emails/alma-bessent-put-2026-08-23.md. URL: https://stochvoltrader.substack.com/p/the-new-bessent-put-oil-midterms
Underlying weekly file: /workspace/emails/sv-pce-jackson-hole-2026-08-23.md
Product: Positioning / implied-distribution map for this week. Not advice. This memo decodes the author’s implied bet. It does not recommend a trade.
Spot used in source: SPX 7674 at Friday’s close (author’s print). Do not treat as live. Live quote = External check needed.
Source discipline: Dealer gamma, charm, theta, contract counts, IVs, betas, and customer-book P&L are source-model, not observable truth. Reproduce his numbers exactly; flag as model-dependent. No invented greeks, no invented IVs, no buy/sell.
Takeaway 1 — The week’s mechanical path is mapped, not a directional call. Source-model at Friday’s 7674 close: dealers carry about $22mn of gamma per 1% move, positive ~7385–7860, and bleed just under $9mn/day of theta to customers. Conviction belongs to the positioning map (path of gamma, charm, and which day customers refuse to be short premium). It does not belong to a view that SPX “should” finish higher or lower. Live dealer books = External check needed.
Takeaway 2 — Monday is the only local short-gamma pocket; the 7800 ceiling dies at Monday settlement. Dealers are locally short gamma at spot inside a pocket reaching 7715. A 21,000-contract long-call band 7740–7860 builds a ceiling that peaks near $125mn per % around 7800 and expires with Monday’s settlement. Source: ~5,000 ES-equivalents to sell into Monday close as that book rolls off. Inference: any Monday squeeze toward 7800 is a same-day, settlement-bounded event, not a week-long pin.
Takeaway 3 — Tuesday pins; Wednesday is the one day customers will not be net short premium. Tuesday is a “clean pin day.” Wednesday (PCE) is neutralized: ~45,000 gross vs ~1,000 net — 2% of gross, versus 13–32 on every other day — leaving dealers with essentially zero gamma at spot. Source: “the one session of the week the customer aggregate refuses to be net short premium into is the data day; the respect shows up in quantity, not price.”
Takeaway 4 — Charm is a Thursday-night / Friday-morning tailwind into Jackson Hole. After a “quiet middle,” source-model has ~4,000 ES-equivalents to buy between Thursday close and Friday morning — “a charm tailwind that arrives hours before the Jackson Hole keynote.” Friday: air pocket just under 7400, “iron wall” of long puts at 7350, and enough weight at 7700 to make it the “natural closing magnet.” Warsh speaks Friday morning (source, this post).
Takeaway 5 — Front-week IV is not rich versus realized. 5-day ATM IV ~11 vs trailing 21-day RV 12.8%. Weekly customer-book breakeven 0.83%/day, 2 bps above the trailing pace. Source’s decode of the implied bet: by selling this week, customers bet it realizes like last week (9.6), not like the trailing month. The whole premium-harvest asymmetry “rest[s] entirely on realized volatility holding 2 bps a day below its trailing month.”
Takeaway 6 — The rates overlay is inflation-dominant, not a growth beta. Spot-vol beta −1.36 VIX points per 1% SPX (corr −0.81 / 63d, −0.87 / 1m). SPX −0.95% per 10 bp on the 2y (corr −0.48); 2y +26 bp over the author’s window. Source: “inflation-dominant regime, stock-bond correlation firmly negative,” with the September meeting priced near a coin-flip for a hike. This is the tape/beta the weekly post measures. It is not an import of the companion Alma geo/repression narrative.
Takeaway 7 — PCE is treated as nearly known; Friday’s interpreter is the unknown. Consensus: core PCE 3.3% YoY, ~0.3 MoM, with a known upside tilt from PPI passthrough in portfolio-management fees. Implied event budgets (source variance split): ~24 SPX pts Wednesday, ~39 for the Nvidia gap, ~43 Friday. Inverting Wednesday’s budget through his betas: effective core surprise SD only 0.02–0.04 of a point — “coherence rather than complacency.” “The market treats the data as known and the interpreter as unknown, and Fri budget is nearly twice Wed. This is key.” NVDA prints Wednesday evening; VIX weekly settlement is that same morning.
Takeaway 8 — What the customer implied bet is. In-line to slightly warm PCE (already treated as nearly known), a hawkish-leaning but non-shocking Warsh, and a drift back into 7570–7690, where “a third of Friday’s density and the bulk of the short-put premium both live.” Speed-profile centroid 7576.18. Weekly customer-book economics: 63% win rate, median take ~$23mn, 1% tail ~−$205mn, expected shortfall −260 (million, units carried from the prior clause). Classic premium-harvest asymmetry. Soft print is “the scenario the book is not built for.”
Takeaway 9 — Mild-hot is where the Friday book makes money; genuinely hot plus hawkish Warsh is the left wall. A 0.1 hot core print → 6–9 bp on the 2y × his beta → −0.9% to −1.3% SPX = 7570–7600, “precisely where the Friday customer book’s payoff peaks at +$60mn.” It takes a 0.2+ hot print combined with a hawkish Warsh to reach the 7450 lower breakeven; below that the book “bleeds into the 7350 wall and the loss brackets get deep.” Soft relief 7710–7760 is a 36% probability zone where the position turns mildly negative. Right tail >7800 is 14% of density, insured with ~7,600 customer-owned calls; hot tail insured via the front VIX call spread.
Takeaway 10 — Time horizons, in the source’s own calendar. Mon–Tue: short-gamma pocket then pin; Monday ceiling gone at settlement. Wednesday 8:30: PCE + VIX weekly settlement; Wednesday evening: NVDA; dealers ~zero gamma at spot. Thu close–Fri morning: +4k ES-eq charm into Warsh. Friday: 7700 magnet / 7350 wall; JH keynote. 16 Sep: customer risk-reversal struck at the forward, ~41% win odds at near-zero net premium, “financed by selling the very mean-reversion it fears.” September FOMC: hike a coin-flip (this post). Year-end SPX/VIX distribution is not priced in this file — author says he examined it in the prior piece (companion memo).
Takeaway 11 — Explicit falsifiers (source). Any one redraws the map: (1) core print outside 0.15–0.45, or >12 bp of 2y movement on Wednesday; (2) Friday close above 7760, or two sessions realizing past 0.83%/day. Author “most likely won’t be around next week” but “will still update this”; next weeklies on the 30th and Labor Day.
Takeaway 12 — What headlines miss. The non-obvious read is not “PCE week = vol event.” It is that customers already treated PCE as a computable number after CPI/PPI, refused to be net short premium only on that day (quantity, not price), and spent the larger event budget on Friday’s interpreter. The implied bet is a modest fade into 7570–7690, not a melt-up and not a crash. Front-week IV ~11 vs 21d RV 12.8 and last week 9.6 is the entire carry engine. This memo proposes hypotheses and a watchlist. It does not recommend a trade.
Conviction in the positioning map (not a price view): Medium in the author’s internal coherence, Low as observable fact. The week’s gamma path, the Wed net/gross collapse, the charm sign-flip, the PCE-as-known vs Warsh-as-unknown budget split, and the 7570–7690 centroid are consistent with each other inside his model. None of it is a listed-options reconstruction this desk has run. Dealer gamma at 7674, the 21k call band, the 45k/1k Wednesday book, and the $23mn / −$205mn customer P&L are source-model. Live levels, contract ownership, and whether Friday’s speaker is in fact Warsh = External check needed.
Chronological, attributed. No invented timestamps. Section order is the post’s own.
Availability / instruction to read the prior piece first. Source said he most likely will not be around next week; readers get the weekly on the 30th, then the next one on Labor Day, and between those he will “most likely not be able to be here,” but “will still update this.” He then pointed at “this new article of mine” in which he “explained the current Bessent measure, Warsh’s plan, its expected effects on the yen, the oil situation, and … what the SPX and the VIX are pricing in through year-end, while also explaining the current geopolitical landscape.” Instruction: “Make sure you read it — don’t even continue until you have.” Those topics live in the companion Alma memo. They are not facts of this weekly post. This post is a free post “available to everyone.”
Stated job of this post. Source said he will “scrutinize the weekly positioning, PCE expectations and implied distribution.” Header: “Coding positioning into Friday expo.”
Dealer gamma / theta at Friday’s close (source-model). Dealers “carry about 22 million dollars of gamma per 1% move at Friday’s 7674 close, positive between roughly 7385 and 7860, and they bleed just under 9 mln a day in theta to the customer side.”
Monday (source-model). “On Mon, dealers are locally short gamma right at spot, inside a pocket reaching 7715, while a 21,000 contract long call band across 7740–7860 builds a ceiling that peaks near 125 million per % around 7800 and dies with Monday’s settlement.”
Tuesday / Wednesday / Thursday / Friday shape (source-model). “Tue looks like a clean pin day.” Wednesday, “the PCE day, … is a neutralized one, carrying the second-largest gross of the week at roughly 45,000 contracts against a net imbalance of barely 1,000 — 2% of gross, where every other day runs 13 to 32 — which leaves dealers with essentially zero gamma at spot.” “The one session of the week the customer aggregate refuses to be net short premium into is the data day; the respect shows up in quantity, not price.” “Thu is modestly long.” “Fri carries an air pocket just under 7400, an iron wall of long puts at 7350, and enough weight at the 7700 line to make it the natural closing magnet.”
Charm / roll path (source-model). “Roughly 5,000 ES-equivalents to sell into Monday’s close as that book rolls off, a quiet middle, then about 4,000 to buy between Thu close and Fri morning, a charm tailwind that arrives hours before the Jackson Hole keynote.”
Spot-vol and SPX/2y betas; regime label. Spot-vol beta −1.36 VIX points per 1% of SPX, correlation −0.81 over 63 days, “softening to −0.87 over the last month as the tape calmed.” SPX “moves about −0.95% per 10 basis pts on the 2-yrs, correlation −0.48,” and “2-yrs yields have climbed 26 bps over the window.” Source: “That is an inflation-dominant regime, stock-bond correlation firmly negative, with the September meeting priced near a coin flip for a hike.”
Front-week IV vs realized; the implied vol-bet. “Against that backdrop the front week’s IV is not rich: 5-day ATM sits near 11 while trailing 21-days realized runs 12.8%, and the weekly customer book’s breakeven of 0.83% a day sits 2 basis points above the trailing pace.” “So by selling this wee,k [source typo] customers bet that it realizes like last week, which printed 9.6, not like the trailing month.”
Calendar stack (this post). “Core PCE comes on Wednesday at 8:30, ahead of the VIX weekly settlement that same morning and NVDA report that evening. Warsh speaks at Jackson Hole on Friday morning.”
PCE consensus, implied event budgets, and the known-vs-unknown split. Consensus: core PCE “holding at 3.3% YoY, roughly 0.3 on the month, with a known upside tilt from the PPI passthrough in portfolio management fees.” “Under my stated variance split, the implied event budgets are about 24 index pts for Wednesday, 39 for the Nvidia gap, and 43 for Friday.” “Inverting Wednesday’s budget through the measured betas implies the market prices an effective core surprise SD of only 0.02 to 0.04 of a point, which is coherence rather than complacency. So after CPI and PPI, the PCE is close to a computable number.” “The market treats the data as known and the interpreter as unknown, and Fri budget is nearly twice Wed. This is key.”
Hot / soft print through the customer book (source-model). A 0.1 hot core print “through the chain, 6 to 9 bps on the 2-yrs times the measured beta,” lands at −0.9% to −1.3% on SPX which is 7570 to 7600, “precisely where the Friday customer book’s payoff peaks at +60 million.” “It takes a genuinely hot print of 0.2 or more combined with a hawkish Warsh to reach the 7450 lower breakeven, below which the book bleeds into the 7350 wall and the loss brackets get deep.” “However soft print is the scenario the book is not built for. A relief toward 7710–7760 lands in the 36% probability zone where the position turns mildly negative, which is why the right tail above 7800, carrying 14% of the density, is insured with some 7,600 customer-owned calls, and the hot tail is insured through the front VIX call spread.”
Stated implied bet. “So the implied bet says that the market expects an in-line to slightly warm PCE it already considers nearly known, a hawkish-leaning but non-shocking Warsh, and a drift back into the 7570–7690 zone where a third of Friday’s density and the bulk of the short-put premium both live.”
Weekly customer-book economics (source-model); September 16 overlay. “The weekly economics are a 63% win rate and a median take around 23 million, set against a 1% tail near −205 million and an expected shortfall of −260, this is the classic premium-harvest asymmetry, resting entirely on realized volatility holding 2 bps a day below its trailing month.” “Into September 16 however a customer risk reversal struck at the forward, roughly 41% win odds at near-zero net premium, financed by selling the very mean-reversion it fears. Spot is damped, vol is bid, and the two are coupled at −1.36.”
Falsifiers (source, explicit). “The falsifiers would be 1) a core print outside 0.15 to 0.45, more than 12 basis points of 2-yrs movement on Wed; 2) a Friday close above 7760, or two sessions realizing past 0.83% a day. Any one of those, and this map gets redrawn.”
Speed profile (source-model). “Speed profile is net negative, customers are selling the downside.” Centroid 7576.18. Downside pivot 7522.58. Downside target 7478.40. Upside pivot 7706.61. Upside target 7819.09. Put/call borderline 7501.01.
Vomma model (source-model). Downside zone 7540.78–7583.66. Upside zone 7748.42–7783.43. “So the market does not expect to breaking the downside. Bet is asymmetric to the upside. Customers are ready to collect left-tail premium.”
Vol-projected sentiment ranges (source-model; selected; full table in Watchlist). SPX 7738.88–7609.86 (vol MoE 7770.08 / 7578.66). ES 7757.20–7625.30. VIX 16.09–14.17 (MoE 16.56 / 13.70). QQQ 723.33–703.55. NDX 29704.21–28913.51. NVDA 220.56–208.88 (MoE 223.38 / 206.06). TLT 82.76–81.34. GC 4771.03–4590.17. CL 89.81–84.31. These are the author’s vol-projected bands, not live quotes and not this desk’s forecasts.
This section is the week’s mechanical plumbing as the source models it. Dealer gamma / charm / theta are source-model, not tape. Ownership of the listed book, whether 7674 is still the relevant spot, and whether the 21k/45k/5k/4k figures still exist on Monday morning are External check needed.
What the system is this week. The “product” being decoded is a customer premium-harvest book overlaid on an inflation-dominant SPX/2y beta, with one data print the customers refuse to be net-short into and one speech they paid almost 2× the data-day budget for. Dealers are the residual: long gamma in a 7385–7860 band at Friday’s close, short gamma locally on Monday at spot, zero gamma on Wednesday, modestly long Thursday, then a 7350 put wall / 7700 magnet on Friday. Theta of just under $9mn/day is the carry customers pay (and dealers earn) to hold that shape.
Bottleneck 1 — Monday ceiling, dying at settlement. The 7740–7860 21k long-call band, ~$125mn/% near 7800, is a same-session lid. It dies with Monday’s settlement. The ~5,000 ES-eq to sell into Monday close is the roll-off of that book. Inference, labeled: a Monday probe toward 7715–7800 is the one window in which local dealer short-gamma (pocket to 7715) and a still-alive call ceiling coexist; after settlement the ceiling is gone and the week is no longer “short gamma at spot.” Tuesday is then described as a pin, not a trend day.
Bottleneck 2 — Wednesday zero-gamma data day, with VIX settlement and NVDA stacked on top. PCE 8:30, VIX weekly settlement that morning, NVDA that evening. Dealers ~zero gamma at spot because net is ~1k vs ~45k gross (2%). Customers will not be net short premium into the print; they expressed that as quantity (gross), not as a rich mid. Implied Wednesday budget is only ~24 SPX pts, and inverted through his betas that is a 0.02–0.04 core-surprise SD — i.e. the index is not paying for a PCE shock. The 39-pt Nvidia gap is a separate, larger budget sitting on the same calendar day after the cash session. Inference, labeled: Wednesday is the week’s unhedged (by dealers) event cluster. The print can move spot without a dealer-gamma shock absorber; the NVDA gap can then gap the same book into Thursday with charm not yet having flipped to the Friday tailwind.
Bottleneck 3 — Friday 7350 put wall / 7700 magnet, after a charm tailwind. ~4,000 ES-eq to buy Thu close → Fri morning, “hours before the Jackson Hole keynote.” Friday’s structure: air pocket just under 7400, long-put wall at 7350, weight at 7700 as closing magnet. Implied Friday budget ~43 pts, nearly 2× Wednesday. Source’s own sentence: data known, interpreter unknown. Warsh is the Friday speaker (this post).
Alma link, only where this weekly post itself connects it — labeled inference, not a fact import. This post tells the reader to finish the prior piece on “the current Bessent measure, Warsh’s plan,” yen, oil, and year-end SPX/VIX before continuing. It then prices Warsh on Friday morning as the unknown interpreter and assigns him the larger budget. Inference: Friday’s 43-pt budget is the market paying for the interpreter of a nearly known PCE, which the author previously framed as a Warsh-vs-Bessent regime question (companion memo: guidance vs term premium; “put paid in the denominator”). Do not treat companion claims — gold +8%, yen 159, diesel cracks, Bessent buyback size, EoY SPX 7830–7870 — as facts of this memo. They were not re-stated here.
Upstream → downstream chain (source, then inference).
Source: 2y +26 bp over the window; SPX −0.95% / 10 bp 2y; corr −0.48; inflation-dominant; stock-bond corr negative; Sep hike a coin-flip; spot-vol beta −1.36.
Inference, labeled: 2y / inflation regime → equity vol → dealer hedging. A Wednesday 2y move >12 bp is an explicit falsifier because it is the transmission belt his beta uses to map a PCE surprise into SPX. If the 2y does the work, SPX does not need a “PCE shock” in the print itself. Dealer gamma is then the last stage: zero on Wednesday (no absorber), charm-positive into Friday (absorber / magnet at 7700), put wall at 7350 only if the hot+hawkish branch is hit.
Customer vs dealer residual. Customers: net short premium except Wednesday; speed profile “net negative, customers are selling the downside”; “ready to collect left-tail premium”; right tail >7800 hedged with ~7,600 calls; hot tail hedged with a front VIX call spread. Dealers: the other side of that shape, plus the Monday call-ceiling that is their long-call band (a ceiling, in the author’s language). Inference: the week is a sold-vol, sold-downside, mildly long-upside-insurance customer book, with dealers providing the pin/magnet except on the data day.
Tension inside this file (not vs Alma). Vol-projected SPX sentiment 7609.86–7738.88 (MoE 7578.66–7770.08) sits around Friday’s 7674 close. The customer-book centroid 7576.18 and the implied-bet zone 7570–7690 sit below spot. Both are the author’s numbers. Inference: the vol-projected band is a realized-vol cone around spot; the implied bet is the customer book’s payoff centroid, which wants a modest fade. Do not collapse them into one “target.”
Each chain: [obs] → [2nd] → [3rd] → [investment relevance]. Investment relevance is what to watch, not a trade. Source vs inference labeled.
Chain 1 — Monday short-gamma pocket.
[obs, source-model] Dealers locally short gamma at spot, pocket to 7715; 21k long-call ceiling 7740–7860 peaking ~$125mn/% near 7800, dying at Monday settlement; ~5k ES-eq to sell into the close.
→ [2nd, inference] Intraday convexity is local and time-boxed. A probe through 7715 can run toward the 7800 ceiling while the band is still alive, then lose that lid at settlement as dealers sell ~5k into the close.
→ [3rd, inference] Tuesday’s “clean pin” is downstream of that roll-off, not an independent view. A Monday close that has already spent the ceiling is a different Tuesday than a Monday that never tested 7715.
→ [investment relevance] Watch 7674 / 7715 / 7800 into Monday settlement, then whether Tuesday actually pins. The Monday 7800 lid is not a week-long magnet. No trade implied.
Chain 2 — Wednesday PCE + zero dealer gamma + VIX settlement + NVDA.
[obs] PCE 8:30; VIX weekly settlement that morning; NVDA that evening; dealers ~zero gamma at spot (~45k gross / ~1k net); implied budgets 24 / 39 / 43 pts (Wed / NVDA gap / Fri); core surprise SD 0.02–0.04; consensus 3.3% YoY / ~0.3 MoM.
→ [2nd, inference] The cash session can move without a dealer shock absorber, but the index is not paying for a PCE surprise. The larger same-day budget is the NVDA gap (39 pts), which prints after VIX weekly settlement. Settlement then a mega-cap gap is a mechanical vol event even if PCE is “known.”
→ [3rd, inference] A “known” PCE plus a hot NVDA gap (or vice versa) can still force Thursday’s book to re-mark before the charm tailwind arrives. Zero dealer gamma means the path of Wednesday is customer- and event-driven, not pin-driven.
→ [investment relevance] Leading indicators: core print vs 0.15–0.45 band and vs 0.3 consensus; 2y move vs 12 bp; NVDA vs his vol-projected 208.88–220.56 (MoE 206.06–223.38); VIX vs the −1.36 beta off whatever SPX does. Event-cluster risk is first-order. No trade implied.
Chain 3 — Thursday–Friday charm into Jackson Hole.
[obs] Thu modestly long gamma; ~4k ES-eq to buy Thu close → Fri morning; Friday 7700 magnet, 7350 put wall, air pocket just under 7400; Warsh Friday morning; Fri budget ~43 pts ≈ 2× Wed.
→ [2nd, source + inference] Source: data known, interpreter unknown. Charm is a mechanical bid into the speech, not a view that Warsh is dovish. Implied bet is hawkish-leaning but non-shocking.
→ [3rd, inference, Alma pointed-at only] If Friday’s interpreter is the Warsh-vs-Bessent question the author told readers to study first (companion memo), then a speech that “says nothing structural” vs one that validates back-end discipline is the branch the 43-pt budget is paying for. This weekly post does not re-specify those branches; do not import them as priced here.
→ [investment relevance] Watch Friday close vs 7700 / 7760 / 7450 / 7350. Close >7760 is an explicit falsifier of this map. Charm is a tailwind into the event, not insurance through a shock.
Chain 4 — Cheap front-week IV vs customer breakeven.
[obs] 5d ATM ~11 vs 21d RV 12.8%; customer BE 0.83%/day (2 bps above trailing pace); last week RV 9.6; weekly economics 63% win, median +$23mn, 1% tail −$205mn, ES −260; “resting entirely on RV holding 2 bps/day below its trailing month.”
→ [2nd, source] Customers are selling this week to match last week’s 9.6, not the trailing month. Speed profile net negative; left-tail premium being collected; right tail 14% density hedged with ~7,600 calls.
→ [3rd, inference] Two sessions >0.83%/day (explicit falsifier) is enough to break the carry engine even if PCE is in-line. IV ~11 vs RV 12.8 means the market is not paying a rich premium for the cluster; the customer book is still short it. Those can both be true.
→ [investment relevance] Track realized vs 0.83%/day session by session, not just the print. A quiet PCE and a 1% NVDA-driven SPX day still falsifies the harvest. No trade implied.
Chain 5 — Inflation-dominant stock-bond corr / September hike coin-flip.
[obs] SPX −0.95%/10 bp 2y, corr −0.48; 2y +26 bp over the window; stock-bond corr “firmly negative”; Sep meeting “near a coin flip for a hike”; into Sep 16, customer RR at the forward, ~41% win at near-zero net premium, “financed by selling the very mean-reversion it fears”; spot damped, vol bid, coupled at −1.36.
→ [2nd, source] A 0.1 hot PCE maps through 6–9 bp on the 2y into 7570–7600, which is the Friday book’s +$60mn peak. The 2y is the transmission belt, not “growth yields.”
→ [3rd, inference] If stock-bond correlation flips positive (equities up with yields), his −0.95% beta and the 7570–7690 fade-bet both misfire — that is a regime break, not a print miss. The Sep 16 RR at 41% / ~zero premium is a later book that sells the mean-reversion this week’s book is harvesting. Same author, two horizons, not the same trade.
→ [investment relevance] Watch 2y on Wednesday vs 12 bp; whether SPX/2y corr stays negative into the Sep meeting; hike odds still coin-flip. Companion Alma’s “customers betting a positive stock-yield beta” is not re-stated in this file — do not mix the two books. No trade implied.
This week is the product. Scenarios below are a decode of his implied distribution and customer-book payoff, not this desk’s forecasts and not recommendations. A second layer for 16 Sep / September FOMC is included because this file prices it. Year-end SPX/VIX is not in this file (companion only).
Spot reference in source: 7674. All index levels are his levels against that close.
| What it is in his book | PCE / Warsh / NVDA | SPX zone (his levels) | Customer Friday book (source-model) | How you’d know in real time | |
|---|---|---|---|---|---|
| Base — the implied bet | In-line to slightly warm PCE, already treated as nearly known; hawkish-leaning but non-shocking Warsh; modest fade, not a crash. ~1/3 of Friday density lives here. Weekly economics: 63% win, median +$23mn. Speed centroid 7576. | Core in ~0.15–0.45 (consensus ~0.3); 2y not >12 bp; NVDA gap inside the 39-pt SPX budget; Warsh hawkish-leaning, not a shock. | 7570–7690 (implied-bet zone). Friday magnet 7700. Vol-projected SPX 7609.86–7738.88 can still print around spot even if the book wants a fade — do not force those together. | Bulk of short-put premium lives in 7570–7690. Mild-hot (0.1) maps to 7570–7600, payoff peaks +$60mn. | Mon ceiling dies at settlement; Tue pin; Wed dealers ~0 gamma; Thu–Fri charm into 7700. VIX tracks −1.36 per 1% SPX. |
| Bull — relief the book is not built for | Soft PCE / risk-on. Source: “soft print is the scenario the book is not built for.” 36% probability zone where the position turns mildly negative. Right tail >7800 = 14% of density. | Core soft vs ~0.3; 2y down or unmoved; NVDA beats/gaps up; Warsh non-event or dovish vs hawkish-leaning expectation. | Relief 7710–7760. Right tail >7800 (Mon ceiling was ~$125mn/% near 7800, but that lid dies Monday). Upside pivot 7706.61, upside target 7819.09. Vomma upside zone 7748.42–7783.43. | Mildly negative in 7710–7760. ~7,600 customer-owned calls sit on >7800. Friday close >7760 is an explicit falsifier. | Two sessions >0.83%/day also falsify, even without a soft print. Watch whether the Monday 21k band still exists before settlement. |
| Bear — genuinely hot + hawkish interpreter | Not the 0.1 hot (that is the peak of the harvest). The left wall is 0.2+ hot combined with hawkish Warsh. | Core ≥0.2 hot (and/or print outside 0.15–0.45); >12 bp 2y on Wed; hawkish Warsh Friday. NVDA can add a same-day gap on a zero-gamma book. | 7450 lower breakeven; then bleed into 7350 wall; air pocket just under 7400. Downside pivot 7522.58, target 7478.40, put/call borderline 7501.01. Vomma downside zone 7540.78–7583.66 (author: market does not expect to break the downside). | Loss brackets “get deep” below 7450 into 7350. Hot tail insured with front VIX call spread. 1% tail of the weekly book ~−$205mn, expected shortfall −260. | 2y is the first tell (6–9 bp maps a 0.1 hot; >12 bp is a falsifier). VIX via −1.36. Friday 7350 is a wall in his model, not a bid this desk can see. |
Author’s own “does not expect a downside break.” Vomma + speed profile: customers selling the downside, collecting left-tail premium, bet “asymmetric to the upside.” That is their wager, consistent with last-week RV 9.6 vs trailing 12.8. It is also why a genuine left tail (0.2+ and hawkish) is where the harvest dies. Premium-harvest asymmetry is the structure, not a reason to be short vol.
No buy/sell. Thesis column is the author’s implied-bet decode, not a recommendation. Metrics are his unless marked External check needed. USDJPY is not discussed in this weekly post (yen thesis lives in the companion); it is omitted. Gold and crude appear only as vol-projected ranges in this file — tabulated, not thematically argued.
| Asset | Why it’s on the board (this post) | Thesis / what the implied bet says | Metrics (source) | Catalyst timeline | Risks / invalidation |
|---|---|---|---|---|---|
| SPX / ES | Primary underlier of the dealer-gamma and customer-book map. | Implied bet: fade/drift 7570–7690, not melt-up, not crash. Mon local short-gamma to 7715; Fri 7700 magnet / 7350 wall. ES vol-projected 7625.30–7757.20 (MoE 7593.40–7789.10). | Spot used 7674. Gamma ~$22mn/% at that close, +ve ~7385–7860. Centroid 7576.18. Speed pivots/targets as in Summary §15. | Mon settlement (ceiling dies); Tue pin; Wed PCE/zero-gamma; Thu–Fri charm; Fri JH. | Fri close >7760; two days >0.83%; core outside 0.15–0.45; 2y >12 bp Wed. Dealer map is source-model. |
| VIX | Spot-vol beta; VIX weekly settlement Wednesday morning; hot tail hedged with front VIX call spread. | Coupled to SPX at −1.36. Settlement before NVDA gap. Vol-projected 14.17–16.09 (MoE 13.70–16.56). | Beta −1.36 VIX pts / 1% SPX; corr −0.81 / 63d, −0.87 / 1m. 5d ATM SPX IV ~11. | Wed morning settlement; Wed NVDA gap; Fri speech. | A VIX move that does not track −1.36 is a beta-regime break. Call-spread line items not specified (strikes/size = not in file). |
| NVDA | Report Wednesday evening; 39 index pts of SPX event budget assigned to the gap. | Largest same-day budget after Friday’s speech. Prints onto a zero dealer-gamma SPX book, after VIX settlement. Vol-projected 208.88–220.56 (MoE 206.06–223.38). | 39 SPX pts (not NVDA points). No IV, no expected move in NVDA pts, no positioning in this file. | Wed evening. | Gap that spends a large fraction of the 39 SPX pts can falsify the 0.83%/day harvest without PCE helping. NVDA-specific options book = External check needed. |
| 2y UST | Transmission belt of the inflation-dominant regime. | SPX −0.95% per 10 bp; corr −0.48; 2y +26 bp over his window. 0.1-hot PCE → 6–9 bp. Sep hike coin-flip. | No 2y yield level stated in this file. TLT range 81.34–82.76 (MoE 80.99–83.11); ZB 107.98–109.77 (MoE 107.55–110.20) are the rates proxies he tabulated. | Wed 8:30 pass-through; Sep FOMC. | >12 bp Wed = falsifier. Corr flip to positive stock-bond = regime risk (Chain 5). Live 2y = External check needed. |
| QQQ / NDX / NQ | In the vol-projected table; mega-cap/NVDA channel. | No separate implied-bet narrative. Ranges only. | QQQ 703.55–723.33 (MoE 698.77–728.11). NDX 28913.51–29704.21. NQ 28988.45–29787.05. | Wed NVDA; week’s SPX path. | Treat as beta to SPX/NVDA, not a second thesis. |
| Other names in his range table (context, not a thesis) | Author printed vol-projected bands for SPY, IWM, AMZN, AAPL, GOOG, META, MSFT, TSLA, TSM, AMD, PLTR, MU, SNOW, CAT, SMH, SEMI, XLE, USO, CL, BNO, BZ, GC, SI, TLT, ZB. | No per-name implied bet in this post except NVDA’s calendar slot. Do not upgrade a range print into a view. | See Appendix table at end of this section. | — | — |
| GC (gold) | Range printed; no gold thesis in this weekly post. | Companion has the gold/denominator argument. This file: GC 4590.17–4771.03 (MoE 4546.42–4814.78) only. | Range only. | — | Do not import the companion gold thesis as a fact here. |
| CL / USO / XLE | Ranges printed; no oil thesis in this weekly post. | Companion has diesel/SPR. This file: CL 84.31–89.81; USO 130.38–138.90; XLE 62.52–64.76. | Range only. | — | Same: oil narrative is companion-only. |
Vol-projected sentiment ranges (source, complete as printed) — not forecasts, not live.
| Ticker | Range (as printed) | Vol MoE bounds |
|---|---|---|
| SPX | 7738.88 – 7609.86 | 7770.08, 7578.66 |
| ES | 7757.20 – 7625.30 | 7789.10, 7593.40 |
| SPY | 772.34 – 759.10 | 775.54, 755.90 |
| VIX | 16.09 – 14.17 | 16.56, 13.70 |
| IWM | 303.65 – 296.27 | 305.43, 294.49 |
| QQQ | 723.33 – 703.55 | 728.11, 698.77 |
| NDX | 29704.21 – 28913.51 | 29895.43, 28722.29 |
| NQ | 29787.05 – 28988.45 | 29980.18, 28795.32 |
| NVDA | 220.56 – 208.88 | 223.38, 206.06 |
| AMZN | 264.05 – 253.21 | 266.66, 250.60 |
| AAPL | 314.88 – 303.82 | 317.55, 301.15 |
| GOOG | 348.22 – 335.28 | 351.35, 332.15 |
| META | 563.97 – 535.83 | 570.78, 529.02 |
| MSFT | 492.19 – 474.29 | 496.51, 469.97 |
| TSLA | 373.64 – 352.08 | 378.85, 346.87 |
| TSM | 429.87 – 408.03 | 435.15, 402.75 |
| AMD | 491.39 – 455.11 | 500.17, 446.33 |
| PLTR | 185.99 – 173.89 | 188.91, 170.97 |
| MU | 1013.13 – 920.43 | 1035.55, 898.01 |
| SNOW | 348.81 – 316.75 | 356.56, 309.00 |
| CAT | 851.62 – 804.18 | 863.09, 792.71 |
| XLE | 64.76 – 62.52 | 65.30, 61.98 |
| USO | 138.90 – 130.38 | 140.95, 128.33 |
| CL | 89.81 – 84.31 | 91.14, 82.98 |
| BNO | 55.61 – 51.99 | 56.48, 51.12 |
| BZ | 97.57 – 91.21 | 99.10, 89.68 |
| SMH | 575.75 – 545.09 | 583.17, 537.67 |
| SEMI | 12096.47 – 11438.53 | 12255.58, 11279.42 |
| GC | 4771.03 – 4590.17 | 4814.78, 4546.42 |
| SI | 72.69 – 68.01 | 73.83, 66.87 |
| TLT | 82.76 – 81.34 | 83.11, 80.99 |
| ZB | 109.77 – 107.98 | 110.20, 107.55 |
P1 — do before Wednesday 8:30. Desk: vol + event + rates.
Independently reconstruct dealer gamma from listed options, or flag that we cannot. Source-model: ~$22mn/% at 7674, +ve ~7385–7860, Mon short-gamma pocket to 7715, 21k long calls 7740–7860 (~$125mn/% near 7800, dies Mon settlement), Wed ~45k gross / ~1k net, Fri 7350 put wall / 7700 magnet, ~5k ES-eq sell Mon close / ~4k buy Thu close–Fri morning, theta just under $9mn/day. We do not have his code, his “customer vs dealer” assignment, or his expiry mix. If the vol desk cannot rebuild a same-sign map from the listed book as of Monday morning, treat the entire plumbing section as unverified source-model. Do not trade it. (Vol desk.)
Verify Friday’s 7674 close and Monday’s live spot. His greeks are anchored to that print. If spot has already left the Mon pocket / 7385–7860 band, the map is stale before the week starts. (Vol / cash.)
PCE: consensus vs his implied distribution. Confirm BLS/BEA consensus for core MoM and YoY vs his ~0.3 / 3.3%. Confirm whether portfolio-management-fee PPI passthrough is still the known upside tilt. Map a 0.1-hot and 0.2-hot against 0.15–0.45 falsifier band. Compare implied surprise SD 0.02–0.04 to a street distribution (Bbg/Street account). (Event / economics.)
Confirm Jackson Hole Friday speaker is Warsh. Author states it as fact in this post. Calendar/agenda = External check needed. If the keynote is not Warsh, the “interpreter unknown / Fri budget 2× Wed” sentence survives, but the Warsh-vs-Bessent pointer to the companion memo does not. (Event.)
September hike odds. “Near a coin flip” — pull the live Fed-funds / FOMC-dated odds. If already 70/30 either way, his rates overlay is stale. (Rates.)
P2 — Wednesday cluster and NVDA. Desk: vol + single-stock.
NVDA positioning and expected move. This file gives 39 SPX index pts for “the Nvidia gap” and a vol-projected NVDA band. It does not give NVDA IV, straddle, dealer gamma, or call/put splits. Single-stock vol desk needs its own book. Question: does a typical NVDA gap historically spend ~39 SPX pts, or is that budget large/small vs last 4 reports? (Single-stock vol / delta-one.)
VIX weekly settlement vs the cash PCE print. Same morning. How much of Wednesday’s 24-pt SPX budget is mechanically VIX-settlement related vs PCE? Source does not split them. (Vol.)
Front VIX call spread that “insures the hot tail.” Strikes, expiry, size: not in this file. Identify whether a liquid front-week VIX call spread is actually sitting in the tape, or whether that is a qualitative hedge comment. (Vol.)
P3 — map integrity and cross-horizon. Desk: vol + rates.
Customer-book P&L reconstruction. 63% win, median +$23mn, 1% tail −$205mn, ES −260, Friday peak +$60mn at 7570–7600, 7,600 customer calls above 7800. Without contract-level assignment these are not auditable. Flag as source-model and do not feed them to risk. (Vol / QR.)
Sep 16 RR vs this week’s harvest — are they the same customers? 41% win at ~zero net premium, struck at the forward, “selling the mean-reversion it fears.” Different horizon, possibly different book. Do not net them. (Vol.)
Author away next week. He will “still update this” but will “most likely not be around.” Process risk: this map is a Sunday-night snapshot. Assign an owner to refresh greeks after Monday settlement and after Wednesday 8:30, rather than waiting for his next weekly (30 Aug). (PM process.)
Model risk (first). Every gamma, charm, theta, contract count, and customer-P&L figure is source-model. “Dealers” vs “customers” is an assignment convention, not a clearinghouse file. Speed profile, vomma zones, and vol-projected ranges are additional model layers on top of the same un-audited book. If listed-options reconstruction disagrees on sign (e.g. dealers not short gamma on Monday, Wednesday not ~zero), the plumbing section is void. Do not treat $22mn/%, 21k, 45k/1k, 5k, 4k, $125mn/%, $23mn, −$205mn as observables.
Timing / staleness. Snapshot is Friday’s 7674 close, written Sunday night. Monday’s 21k call band dies at Monday settlement — i.e. a large piece of the map expires on day one. Author says he will “most likely” be away next week but will still update this post. That is a promise, not a process. If spot, 2y, or the listed book move Sunday night/Monday morning, the map can be wrong before PCE.
Event-cluster risk (Wednesday). PCE 8:30, VIX weekly settlement that morning, NVDA that evening, dealers ~zero gamma at spot. Three distinct shocks on the one day the pin is off. His 24 / 39 split says the market is not paying for PCE and is paying for an NVDA gap, but those are variance-budget numbers, not a guarantee that the gap will be 39 pts or that PCE cannot surprise. A known PCE and a large NVDA gap still produce a realized day that can clear 0.83%.
Regime risk. The weekly post’s measured world is inflation-dominant, stock-bond corr negative, SPX −0.95%/10 bp 2y, spot-vol −1.36. If corr flips (equities and yields up together), his PCE→2y→SPX chain and the 7570–7690 fade both misfire. A Sep hike that is no longer a coin-flip similarly rewrites the rates overlay. This file does not re-state the companion’s “customers betting a positive stock-yield beta.” Mixing those two would be a regime error of our own.
Asymmetry the book is not built for. Soft/relief 7710–7760 (36% zone, mildly negative) and Friday close >7760 (falsifier) are upside breaks of a sold-vol, sold-downside book. The left wall needs 0.2+ hot and hawkish Warsh to reach 7450/7350. Mild-hot (0.1) is where the harvest peaks. Reading “inflation-dominant” as “the desk should be short equities” is a category error: the implied bet is a modest fade with left-tail premium collected and the right tail insured, not a crash call.
Interpreter risk (Friday). Fri budget ~43 pts ≈ 2× Wed. Charm +4k ES-eq into the speech is a mechanical bid, not protection through a shock. If Warsh is not the speaker, or is a shock vs “hawkish-leaning but non-shocking,” Friday is the larger event and the 7700 magnet is just a prior.
Process / key-person. Author likely off between 30 Aug and Labor Day weeklies. Map will not auto-refresh unless he updates this post or the desk rebuilds it.
What would invalidate the memo’s usefulness (not a P&L stop). Any one of his stated falsifiers; a listed-options rebuild that cannot find the Monday short-gamma pocket or the Wednesday ~zero net; 7674 no longer near spot; Jackson Hole keynote ≠ Warsh and no other “interpreter” event on Friday; stock-bond corr flip. On any of those, stop using this map. Do not invent a hedge from it.
Bottom line for the PM. This is an implied-bet decode of a sold-this-week-to-match-last-week’s-9.6 customer book, sitting on an inflation-dominant SPX/2y beta, with dealers long-gamma most of the week except a Monday local short pocket and a Wednesday zero-gamma data/NVDA cluster, and with charm buying into a Friday Warsh the market paid almost 2× the PCE budget for. The bet is in-line/slightly warm PCE, non-shock hawkish Warsh, drift 7570–7690. Conviction is in that map, if and only if the vol desk can recognize its sign in the listed book. No trade recommendation.
Source: Stochastic Volatility weekly post, 23 Aug 2026, 7:34 PM ET. Companion Alma piece is context the author required readers to finish first; its geo, oil, yen, Bessent-buyback, and EoY SPX/VIX claims are not facts of this memo. Dealer positioning = source-model. External check needed for live levels.
Desk copy · not a trade recommendation · Stochastic Volatility · Substack · 23 Aug 2026