Positioning and Outlook - 2026-08-21

Published 2026-08-21

AI-linked earnings and a benign inflation print have kept equities rising and the Federal Reserve on hold, even as corporate credit health remains weak and a record buyback bid of roughly $4.5 trillion shoulders much of the market's support.

Recent Events

The clearest lift of the past few weeks came from earnings season, where a fresh round of AI-linked results pushed technology shares higher and a supportive inflation print reinforced bets that the Federal Reserve would keep rates on hold, a backdrop in which the central bank itself stayed relatively subdued, its communication tone classed as moderate with readings of 17 and 15 out of 100, and a speech from Cook on the outlook for the US and Alaskan economies the lone scheduled item. Credit told a more strained story: the average corporate survival score recovered only slightly, from 22 to 25 out of 100, while the count of companies in danger narrowed from 51 to 48 and high-yield spreads ended the stretch flat, all against a still-massive buyback bid of more than 12,000 active programs totaling roughly $4.5 trillion over the trailing twelve months. Away from the macro tape, On Holding tempered its sales outlook after missing quarterly revenue targets, Boston Scientific's neuromodulation unit and AVID Medical each drew the FDA's most serious class of device recall, Veralto reported first-quarter results, and Asian multi-strategy diversifier funds posted heavy monthly drawdowns, while commentary argued that only the Bank of Japan could arrest the yen's decline. How those cross-currents resolve is the business of the outlook that follows.

Upcoming Events

The forward calendar is back-loaded: the macro spine sits in mid-September, when four major central banks decide policy in quick succession, the ECB on the 10th, the FOMC on the 16th, and the Bank of England and Bank of Japan within twenty-four hours of each other after that, a cluster that compresses the window in which the tape can reprice. Before that, the terrain is thinner, with the PBOC's loan prime rate fixing on August 20th the main macro marker and the week's government bill and bond auctions, including a 30-year sale, the modest supply events to clear. Single-name catalysts fill the gap, led by a run of phase-two trial readouts from Takeda, Pfizer, GSK and Novo Nordisk across August 20th and 21st. Earnings season adds steady background noise, with 198 reports scheduled over the window and Workday, Affirm and Buckle among the nearest to print.

Macro Projections

The next three weeks run a dense gauntlet of catalysts, PCE and GDP on 8/26, payrolls on 9/4, then the ECB 9/10, CPI 9/11, the FOMC 9/16 and the BoE 9/17, and the inflation prints are the swing inputs: prediction markets put September CPI above 3.0% year-over-year at 92% with breakevens at 2.30% and still drifting up, so a hot number revives hike talk into a meeting priced 69% to hold, while a soft one feeds the 13% tail of a cut before 2027. The near-term leg leans calm, and carries the highest confidence, because the transmission channels are quiet: high-yield spreads at 2.73% have been flat for three weeks, the CCC-versus-BB ratio (the weakest junk credits against mid-tier ones) is stable at 1.995 with false-bottom risk low, the Fed-reaction model puts hold propensity at 0.85 against intervention at 0.16, and volatility is subdued everywhere, VIX 14.89, bond-market volatility (MOVE) at 73, G7 FX volatility (how much major exchange rates are swinging) below average though accelerating, and the VIX term structure in contango at 0.841, albeit on a stale 7/31 read that gets discounted accordingly. Confidence steps down a rung at the cyclical horizon, where the stress sits at the long end and the mechanism is term premium rather than policy shock: the 30-year at 5.237% stands at its 93.6th percentile since 2001 and briefly touched levels last seen in 2007, while the 10-year at 4.696% (58th percentile) and the 2-year at 4.19% (56th) sit unremarkable, a steepening signature consistent with the balance-sheet shrinkage and no-forward-guidance regime associated with Warsh. That long-end pressure propagates through the carry channel, where the model puts a disorderly unwind at 50.3%, decelerating from its 52.1% peak, but the 0.39-to-0.52-to-0.50 trail reads as a regime step-up rather than noise, against a market pricing essentially no equity-vol tail, with SKEW (the price of tail-risk protection) already elevated at 143 while leveraged funds sit record short the two-year and other reportables record short the S&P 500: someone is hedging what the index level denies. The honest bound is a fork: roughly 30–35% is placed on the calm-credit regime breaking by Q4, with high yield widening through 3.25% alongside MOVE above 100 as the confirming lead indicator, and the conditional branch runs through Hormuz, if two-day-old closure rhetoric and tanker seizures escalate into sustained transit disruption by mid-September, Brent clears $100, the above-3.4% CPI tail (priced at 27%) becomes the base case, and spreads blow through that tripwire, whereas absent escalation the 4.4-million-barrel crude build caps oil and the muddle-through holds. Cross-domain agreement currently endorses the no-landing path, prediction markets put 2026 recession at 7%, payrolls above 50k at 53% and GDP above 2.5% at 57%, and sentiment at 70 greed confirms rather than contradicts the calm, while the China theme (tariff escalation day one, Taiwan's record defense budget) registers as a slow-burn drag on multiples rather than a catalyst. At the structural horizon, where these mechanisms are weakest, the valuation channel still points one way: at a CAPE of 42.1 the US equity earnings yield of 3.98% sits well below 5.38% for developed ex-US and 5.98% for emerging markets, leaving domestic large-caps the secular loser of the relative-value chain. The whole framing carries explicit falsifiers, it is wrong if high yield widens through 3.0% or MOVE sustains above 100 by the 9/16 FOMC, which would mean stress is transmitting rather than contained, or if the 30-year round-trips below 4.90%, which would mark the term-premium scare as transient.

Positioning

Position weights in the model portfolio scale directly to conviction: each holding is sized by its screening-kernel composite score, a modelled blend of expected upside drift against drawdown probability, so the highest-scoring names carry the largest equity weights while lower-conviction names stay small. That conviction-led sizing is set against a calm credit backdrop with high-yield spreads near 2.73%, a regime in which the model is content to let its strongest scores concentrate at the top of the book. Beneath the individual names, a set of documented cause-and-effect channels, merger-arbitrage completion, oil-supply shocks feeding headline inflation, and commodity production losses, act as weighable inputs rather than predictions, each carried with its full historical base rate, misses counted, and discounted for what the market already prices. The resulting weights and the dossier behind each holding are set out in the model-portfolio table below.

Model Portfolio

The model portfolio is running a defensive posture, with just under a third of the book parked in money-market cash, dry powder earmarked for staged deployment as the credit and volatility backdrop resolves. What is deployed expresses a single, concentrated theme: electrification and the data-center power buildout, carried by two copper producers and a uranium miner. Each of those single-name positions sits at little more than one percent of the book, keeping company-specific risk contained while the cash reserve does the heavy lifting on the cautious side.

Track Record

The track record so far is a mixed one, and it is reported in full rather than curated: every idea the model has logged since mid-June is scored when its horizon arrives, win or lose. Over the 68 days evaluated, short-horizon calls have been close to a coin flip, about half of one-day picks beat the benchmark, with average alpha hovering near zero, while the four-week window has been weaker, with only about a third of ideas finishing ahead and mean alpha modestly negative. The counts are large enough to be taken seriously, but the sample covers barely two months in a single market regime, and the three-month horizon has yet to produce its first readings, so this record should be read as an early discipline check rather than a verdict.

Macro forecasts: 77 graded, 415 open — each call is scored against what actually happened when its horizon arrived.