PortfolioAI macro research · September 25, 2026

U.S. Recession Risk: Market Odds Swing, Labor Holds

A jump and retreat in year-end recession pricing is a reminder to separate an event-market trade from evidence of an economic contraction.

Executive summary

The U.S. expansion remains the more likely near-term path on the available hard data. August unemployment stood at 4.1% and the August Sahm Rule reading was −0.07 percentage points, far below the historical +0.50 warning threshold. Real GDP rose from $24.180 trillion in the first quarter to $24.270 trillion in the second, in chained dollars at an annual rate. Yet the quarter-on-quarter rise of about 0.37% calculated from those levels offers less room to absorb a new shock than the labor numbers alone might suggest.

Polymarket's 2026 U.S. recession contract told a more volatile story on September 25: its Yes price reached approximately 12.5% around 1 p.m. ET before returning to approximately 9.5% by 7 p.m. ET. These are sampled historical prices, not a change in GDP or unemployment. The contract's end-2026 horizon and its specific resolution rules are not interchangeable with a forecast for the next twelve months.

Next 3 months

Recession is not the base case. Watch whether unemployment begins to rise persistently rather than reacting to a single market move.

Next 6 months

Expansion remains favored; softer output growth makes hiring and consumer demand the decisive tests.

Next 12 months

Uncertainty rises as the window extends into September 2027, beyond the year-end market contract.

Horizon views are qualitative assessments, not calibrated recession probabilities.

Three views of the recession signal

Labor trigger: still well below its warning line

Monthly Sahm Rule, percentage points, through August 2026. Source: FRED / St. Louis Fed. The threshold is historical, not a guarantee.

Output: positive, with a smaller quarterly gain

Quarterly real GDP, trillions of chained dollars at annual rates, through Q2 2026. Source: BEA via FRED. The level change is not the official annualized growth rate.

Year-end odds: an intraday round trip

Selected hourly historical Yes prices, September 25, 2026 (ET); indicative market prices, not measured recession frequencies. Contract and resolution criteria: Polymarket.

Recession risk scorecard

Each reading carries its own observation date. A daily display of a monthly series does not create a new monthly estimate.

IndicatorLatest underlying observationInterpretation
Sahm Rule−0.07 pp · Aug 2026Below the +0.50 historical threshold; labor has not confirmed a recession.
BCIG8.3 · May 8, 2026Recovered from 4.0 on April 3, but too stale to serve as a September signal.
Recession-probability series0.76% · Jul 2026Low model estimate; not a September observation or a directly comparable year-end contract price.
Real GDP$24.270tn · Q2 2026Above Q1's $24.180tn; positive output, smaller sequential gain than in Q1.
Unemployment4.1% · Aug 2026Unchanged from July; sustained increases would materially change the outlook.
Money-market fund assets$8.44tn · Q2 2026Above Q1's $8.29tn; cash allocation is not itself a recession forecast.
“Recession” searches2,124 U.S. searches · Sep 24, 2026Below 2,202 on Sep 23; measures attention, not economic activity.
Polymarket 2026 contract~9.5% Yes · Sep 25, ~7 p.m. ETDown from ~12.5% near 1 p.m. ET; market liquidity and settlement rules matter.

Professional commentary and outlook

September 25 supplied a useful stress test of what a recession dashboard can and cannot say. The year-end contract traded through a roughly three-percentage-point intraday range between its afternoon high and evening level. Meanwhile, the latest underlying unemployment, GDP and model observations remained dated August, Q2 and July, respectively. A price move may reflect new information, positioning or liquidity; without corroboration in fresh economic releases it should not be presented as a deterioration in employment.

The constructive case rests on the employment cushion: the jobless rate has held at 4.1% for two months and the Sahm reading moved from +0.27 in February to −0.07 in August. The weaker part is output momentum. The Q2 real-GDP level rose by $89.2 billion in annual-rate chained dollars from Q1, versus $124.7 billion from Q4 2025 to Q1 2026. That is still growth, not a contraction. It does, however, increase sensitivity to a simultaneous pullback in household spending, hiring and capital investment.

Do not confuse the July 0.76% recession-probability series with Polymarket's approximately 9.5% price. One is a dated model observation; the other is a tradable price for a contract that can resolve on two consecutive negative quarterly GDP growth readings within its specified window or an NBER announcement under its rules. Neither provides a direct three-, six- or twelve-month probability. Similarly, $8.44 trillion in money-market funds may reflect attractive cash yields or defensive positioning, but does not identify which explanation dominates.

What would change the call: repeated increases in the unemployment rate, a Sahm Rule advance toward +0.50 and a negative sequential real-GDP reading together would make a downturn materially more credible. Stable jobs and another positive GDP quarter would support the expansion case. For investors, keep liquidity and refinancing exposure visible in demand-sensitive holdings; a low contract price does not remove tail risk. In capital-intensive cloud and AI projects, including the customer-backed infrastructure thesis discussed in PortfolioAI's recent stock-ideas report, contracted revenue still needs to become cash after power, hardware and financing costs.

As-of September 25, 2026. Series have different release schedules and may be revised. This assessment is for research, not investment advice.