Positioning and Outlook - 2026-08-25

Published 2026-08-25

Markets remain split between AI-driven earnings strength and rate-hold support on one side and rising Middle East oil-supply risk on the other, with firm buybacks and modestly improving credit health offering some cushion ahead of a central-bank-heavy September calendar.

Recent Events

Markets spent the past few weeks caught between a supportive earnings tape and a tense geopolitical backdrop: tech rallied on AI-driven results and an inflation print that reinforced rate-hold expectations, even as Washington's "economic D-Day" rhetoric toward Iran and a reported Houthi attack on a Saudi tanker in the Red Sea kept oil-supply risk firmly in focus, and pressure on the yen drew attention back to the Bank of Japan. Beneath the surface, corporate activity remained brisk, with more than 12,000 active buyback programs and roughly $4.5 trillion in trailing repurchases, while internal credit-health readings improved modestly as the average survival score ticked up from 22 to 25 and the count of names in the danger zone slipped from 51 to 48. Fed communication through the period registered as moderate in intensity, and the calendar was light, headlined by Governor Cook's remarks on the U.S. and Alaskan outlooks.

Upcoming Events

The forward calendar tilts heavily toward mid-September, when central banks take over the tape: the European Central Bank's Governing Council meets on September 10, and the Federal Reserve, the Bank of England and the Bank of Japan then deliver policy decisions on three consecutive days, with China's loan prime rate fixing closing out the run on September 21. Each of the four major rate decisions carries high expected impact, and their compression into barely more than a week leaves a narrow stretch in which policy expectations across the major economies get repriced at once. The run-up is quieter by comparison, a handful of short-dated Treasury bill auctions on August 24 and 25, plus a batch of clinical-trial readouts from Regeneron, Roche and AstraZeneca that are single-name events rather than macro movers. Earnings supply the background rhythm, with roughly 200 reports scheduled across the window and the nearest cluster including Box and JOYY.

Macro Projections

The near-term view hangs on a packed calendar that begins with tomorrow's US GDP and PCE print (Aug 26) and runs through payrolls (Sep 4), the ECB (Sep 10), CPI (Sep 11), the FOMC (Sep 16), the Bank of England (Sep 17) and the Bank of Japan (Sep 18), and the first print lands directly against a sticky-inflation consensus, prediction markets put a 95% chance on CPI holding above 3.1% through September and only 12.5% on a rate cut before 2027, so a soft core PCE would upend that pricing while a hot one would feed the global bond sell-off. For now the regime reads calm but only thinly confirmed: the macro kernel's confirmation score is just 30/100, high-yield spreads (HY OAS, the extra yield low-rated corporates pay over Treasuries) have sat at 2.70–2.75% for two weeks, and equity volatility is benign with the VIX at 15.13 and its term structure in contango, yet the SKEW index of demand for left-tail equity protection has climbed to 143.9 and is still accelerating, and rate-market volatility (the MOVE index) is ticking up at 74bp from unusually low levels, meaning hedgers are quietly paying for protection the surface calm does not justify. The cyclical driver behind that caution is a global term-premium repricing rather than a US fiscal story: long-end yields sit at extreme percentiles nearly everywhere, the US 30-year at 5.231% (93.6th percentile, now backing off), the JGB 30-year at 4.04% (99.9th), the Gilt 10-year at 5.05% and 20-year at 5.56% (both 99.8th), and Bunds at 3.27% and 3.72%, while the US 10-year itself is unremarkable at the 58th percentile and the US–Japan and US–Germany 10-year differentials of 1.82% and 1.43% confirm the move is worldwide. The inflation channel supports that reading, because the 10-year breakeven (the market's implied inflation rate) at 2.34% is elevated at the 71st percentile but decelerating, which points to sovereign risk premia and geopolitical uncertainty doing the work rather than runaway inflation expectations. The fork in the outlook is quantified: the carry-unwind model assigns a 44% probability to a disorderly unwind (having spiked to 52% before settling) while prediction markets price a 2026 recession at only 8% and 2027 at 25%, a 36-point gap between model-implied tail risk and crowd conviction that sits atop classic crowded positioning, leveraged money heavily short front-end Treasuries at a +3.43 reading, smaller traders short the yen at -2.93, so if global yields turn lower that short base is squeezed violently, and if yields resume rising the unwind propagates across currencies and emerging markets. Geopolitics loads the same tail: the Iran/Hormuz situation is in a near-closure state two days running, the geopolitical-risk index reads 165.2, and the model's energy-closure and commodity-flow-disruption mechanism edges are firing, yet Kalshi prices Brent above $93 a barrel at only 11%, an apparent underpricing of a physically material disruption. Structurally, the base case is muddle-through, credit stays calm, no recession, but the equilibrium is fragile: US equities at a cyclically adjusted price/earnings multiple of 42 and a 2.38% earnings yield are extremely expensive while credit spreads discount almost no risk, long-duration bonds are historically cheap precisely because positioning is so crowded short, and commodities carry Hormuz upside that has not been priced as a crisis. The boundaries that would settle the argument are explicit: if high-yield spreads widen above 3.0% and the US 30-year climbs back above 5.28% before the FOMC, the calm-credit regime is breaking toward the disorderly path, whereas a 30-year below 5.0% with Brent under $85 would say the geopolitical premium is receding and the unwind probability should fall, these are weighable, base-rate-grounded channels rather than predictions, and confidence in them legitimately thins as the horizon stretches from weeks into years.

Positioning

With high-yield credit spreads near 2.70% pointing to a calm credit backdrop, the model portfolio sizes each position to conviction, where conviction is a screening-kernel composite score blending expected upside drift against drawdown probability, so the highest-scoring names carry the largest equity weights while lower-conviction candidates stay small. Overall risk appetite is further calibrated by a set of documented structural mechanisms, spanning merger-arbitrage completions, oil-supply shocks feeding through to headline CPI, and commodity production losses, each carried with its historical base rate across every prior firing and discounted for what the market already prices, making them weighable inputs rather than predictions. 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 light, holding most of its capacity in cash as dry powder while the deployed capital does a few targeted jobs. The largest commitment sits in global quality equities tilted to the AI capex cycle, joined by a single-name position in TPL, a Permian royalty play where the research weighs a cash-rich balance sheet and steady buybacks against an extreme multiple and no control over drilling volumes, with sentiment reading neutral. The clearest active theme is the build-out of the missing international leg: a first tranche spread across several unhedged country funds, positioned for a weakening-dollar regime and still well short of the sleeve's 12% target.

Track Record

The scoreboard so far is mixed, and it is published in full: every recommendation logged since mid-June is scored when its horizon arrives, with no filtering of the uncomfortable outcomes. Roughly half of the evaluated ideas have finished ahead of the market over the one-day and one-week windows, but the four-week record is weaker, only about a third beat the benchmark, and average alpha sits modestly below zero across all three matured horizons. The first three-month calls begin maturing around mid-September, which will provide the initial read on the longest horizon. With just 72 days of outcomes behind it, the sample remains too short to prove much either way; the point of the exercise is the discipline itself, calls written down before the result is known and graded when the clock runs out, whether or not the grade flatters the system.

Macro forecasts: 85 graded, 442 open — each call is scored against what actually happened when its horizon arrived.