CONVEX
Last updated
▍ STATISTICAL PROJECTION · YEAR-END 2026
Central Estimate
1.66%
-8.8% vs current 1.82%
68% Range (±1σ)
-126.81% to 130.13%
95% Range (±1.96σ)
-250.14% to 253.46%
Central estimate uses the unconditional 25-year historical average because current regime buckets had insufficient observations to produce a reliable blend.
METHOD: CENTRAL = SAMPLE-WEIGHTED MEAN OF PER-ANCHOR CURRENT-REGIME 1Y AVERAGES, SCALED TO 210-DAY HORIZON. BAND = ±σ√T USING 7732.4% ANNUALIZED REALIZED VOL.
EXPECTED TO BE 1.66% BY 2026-12-31 (LOWER FROM 1.82% ON 2026-03-01). NOT INVESTMENT ADVICE.
▍ MODEL · STATISTICAL FORECAST · 2026

US Recession Probability (Smoothed) Forecast 2026

Quantitative analysis from 298 observations of US Recession Probability (Smoothed) history, joined to four universal macro regime classifications. Numbers are computed, not narrated.

ByConvex Research Desk·Edited byBen Bleier·
RECPROUSM156N · LAST
1.82%
AS OF 2026-03-01
Percentile · 25Y History
86.6th

Performance by Window[02]

WINDOWNANN RETANN VOLRET/VOLHIT %TOTAL
1Y13659.39%347.78%1.9058.3%658.33%
3Y3662.73%253.93%0.2557.1%313.64%
5Y61146.53%749.73%0.2061.7%9000.00%
10Y1212.81%12184.74%0.0050.4%31.88%
All298-10.56%7732.41%-0.0046.3%-93.68%

Annualized total return = (1 + total)^(1/years) - 1. Ret/Vol is the annualized return divided by annualized volatility (Sharpe-equivalent without risk-free subtraction). Hit % = pct of single periods that were positive.

Where We Are Now[03]

Percentile Rank
86.6th
0.00median 0.34100.00
Current value 1.8200 on a 298-observation history going back to Aug 1, 2020.
Volatility Regime
low
207.08%REALIZED 30D ANN
Sits at the 23.5th percentile vs full history. Median 385.91%.

Forward Returns by Macro Regime[04]

How US Recession Probability (Smoothed) has performed historically conditional on the prevailing macro regime. The current bucket is highlighted; +1Y averages drive the headline signal above.

VIX
Volatility regime: Low (<15), Normal (15-25), Elevated (25-40), Extreme (>40)
CURRENT: 16.89 Normal (15-25)
REGIME BUCKETN+30D+90D+1Y AVG+1Y MEDHIT %
Low (<15)6540.38%247.43%1315.25%30.43%60.0%
Normal (15-25)9032.51%105.10%702.38%6.46%51.8%
Elevated (25-40)323.63%2.09%137.38%-48.33%31.3%
Extreme (>40)3n/an/an/an/an/a
10Y-2Y Yield Curve
Yield curve regime: Inverted (<0bps), Flat (0-100bps), Steep (>100bps)
CURRENT: 0.51 Flat (0-100bps)
REGIME BUCKETN+30D+90D+1Y AVG+1Y MEDHIT %
Inverted (<0bps)2727.91%87.66%2922.09%10.00%55.6%
Flat (0-100bps)6234.45%311.32%957.38%44.44%62.5%
Steep (>100bps)10026.44%40.07%153.76%-37.50%42.0%
HY OAS Spread
Credit regime: Tight (<350bps), Normal (350-500bps), Stressed (>500bps)
CURRENT: 2.83 Tight (<350bps)
REGIME BUCKETN+30D+90D+1Y AVG+1Y MEDHIT %
Tight (<350bps)2461.23%49.09%183.63%114.29%73.7%
Normal (350-500bps)4539.20%311.27%499.90%-10.00%48.8%
Stressed (>500bps)1822.23%-0.86%-15.98%-20.00%27.8%
Trade-Weighted Dollar
Dollar regime: bottom/middle/top tercile of trailing 5Y rolling distribution
CURRENT: 118.73 Weak (bottom tercile)
REGIME BUCKETN+30D+90D+1Y AVG+1Y MEDHIT %
Weak (bottom tercile)3415.82%221.28%3096.13%109.09%64.7%
Neutral (middle)3875.78%46.17%198.19%50.00%55.9%
Strong (top tercile)7714.46%192.46%293.48%-14.29%44.0%

Forward returns are forward-looking from each historical observation in the bucket; +252d corresponds to one trading year. Buckets with fewer than 5 forward-return observations are reported as n/a. These are conditional historical averages, not forecasts.

Lead-Lag Relationships[05]

For each universally-recognised leading indicator, the lag at which the daily-return correlation peaks. Positive lag means the anchor leads US Recession Probability (Smoothed); negative means it lags.

ANCHORROLEPEAK LAGPEAK CORRZERO-LAGRELATIONSHIP
Initial Jobless ClaimsLabor leader-1d0.998-0.006coincident
HY OAS SpreadCredit risk leader-1d0.5620.182coincident
10Y-2Y Yield SpreadRecession leader-29d-0.4860.028lags target by 29d
Baa-10Y SpreadCredit risk (slow)-1d0.4780.087coincident
VIXVolatility leader+54d0.4120.264leads target by 54d
10Y Treasury YieldDiscount-rate driver-1d-0.261-0.145coincident
U-Mich Consumer SentimentSurvey leader-1d-0.206-0.120coincident
Trade-Weighted DollarFX driver-1d0.1880.062coincident
CopperGlobal growth proxy+39d0.139-0.001weak
NFCIFinancial conditions-3d-0.0150.002weak

Pearson correlation of daily returns over up to 25 years of overlapping history, searched across a ±60-day lag grid. Indicators classified as “weak” don't have meaningful predictive power at daily resolution; many of these (yield curve, NFCI, sentiment) lead at monthly/quarterly horizons instead.

Historical Analogs[06]

Periods where US Recession Probability (Smoothed) sat at a similar percentile rank to today, with what happened over the next 30 / 90 / 252 trading days. Analogs are clustered to avoid double-counting nearby dates.

DATEVALUE+30D+90D+1Y
Jan 1, 20242.8200-82.27%-86.52%-79.43%
Sep 1, 20211.5600-89.74%-35.90%-66.67%
Oct 1, 20192.0400-56.86%-87.25%-98.04%
Apr 1, 20191.6600-33.73%-56.63%5924.10%
Jan 1, 20191.76003.41%-37.50%-75.00%

Worst Historical Drawdown[07]

-100.00%PEAK-TO-TROUGH
Peak Mar 1, 2020 → trough Aug 1, 2020. Has not yet recovered to prior peak.
All-time high: 100.0000 on Mar 1, 2020 · Current DD from ATH: -98.18%

Largest Single-Period Moves[09]

▲ Up
  • Mar 1, 202038361.54%
  • Apr 1, 20211300.00%
  • Aug 1, 2021957.14%
  • Jun 1, 2004666.67%
  • Nov 1, 2020600.00%
▼ Down
  • Aug 1, 2020-100.00%
  • May 1, 2020-99.80%
  • Oct 1, 2005-97.79%
  • Jul 1, 2009-95.92%
  • Feb 1, 2014-90.00%

Calendar-Month Seasonality[10]

Average single-period return aggregated by the calendar month in which the period ended.

MONTHAVG RETURNHIT %N
January33.93%40.0%25
February28.02%48.0%25
March1557.30%40.0%25
April83.14%50.0%24
May0.08%41.7%24
June49.00%50.0%24
July12.63%44.0%25
August46.51%48.0%25
September25.18%50.0%24
October1.51%48.0%25
November21.53%40.0%25
December47.92%56.0%25

N = 298 OBS · GENERATED 2026-05-03 06:01Z

Forecast Approach

regime implied: The current macro regime classification (Goldilocks, Reflation, Stagflation, or Deflation) dictates the expected direction and magnitude of movement, calibrated against historical regime performance.

Key Drivers & Risks

  • Economic growth
  • Yield curve
  • Labor market
  • Credit conditions
  • Leading indicators

Historical Volatility

Low: composite indicators move slowly

How US Recession Probability Forecasts Have Held Up Historically

The Chauvet-Piger smoothed recession probability has a strong historical track record: probabilities above 80% have reliably coincided with NBER-dated recessions, with no false positives since the model's inception in the 1970s. The 2008 recession was flagged at 85%+ probability six months before the NBER call; the 2020 COVID recession was flagged at 95%+ within weeks.

Regime-conditional models on recession probability are tautological because the probability is itself the regime classifier output. The model aggregates payrolls, unemployment, industrial production, and real income into a single Markov-switching estimate.

Regime Sensitivity for RECPROUSM156N

Recession probability has clean regime sensitivity to labor market and growth indicators. Goldilocks regimes anchor probability below 5%; stagflation regimes typically push it above 30% as growth slows; deflation regimes spike it above 80%.

The April 2026 setup has recession probability at low single-digit levels despite the curve re-steepening signal and the 2024 Sahm-Rule false positive. The regime conditional reads as low-recession-probability with the caveat that the model can shift sharply when payroll revisions or unemployment spikes hit.

What Drives RECPROUSM156N Forecast Errors

Three structural issues drive recession-probability forecast errors. First, the model is purely backward-looking based on confirmed data; it doesn't incorporate forward-looking signals (curve, credit spreads, claims). Recessions can begin in real time before the probability rises.

Second, the 2024 Sahm-Rule false positive reflected immigration-driven labor force expansion rather than demand destruction. The recession probability didn't spike because the underlying employment data didn't deteriorate sharply, but the curve and Sahm signals had flagged risk.

Third, BLS payroll revisions can change historical probability estimates. The February 2025 annual benchmark revision cut prior payroll estimates by 800k jobs, which retroactively raised recession probabilities for 2024 H2 in the smoothed series.

How to Use This Forecast in Practice

For US recession probability, the 80% threshold is the actionable trigger: above this level, NBER recession dating is highly likely. Sub-30% readings are constructive; 30-80% reads as warning-but-not-confirmed.

The cleanest cross-check is the New York Fed yield-curve recession probability (a different methodology using 10Y-3M curve). When both Chauvet-Piger and NY Fed flag elevated probability, the regime read is high-conviction; when they diverge, weight the Chauvet-Piger signal because it incorporates contemporaneous data. The 68% band on recession probability is the tightest of any composite indicator because the smoothing dampens noise.

Frequently Asked Questions

What factors could push US Recession Probability (Smoothed) higher?

The primary drivers that tend to lift US Recession Probability (Smoothed) depend on the current macro regime. Recession indicators distill complex economic dynamics into actionable signals. The Sahm Rule, triggered when the 3-month average unemployment rate rises 0.5 percentage points above its 12-month low, has a perfect track record since 1970. Combined with yield-curve inversions and declining leading indicators, these metrics help traders identify turning points before they become consensus. Convex tracks these drivers live across the Recession Indicators category and flags when multiple forces align in the same direction. See the "Key Drivers & Risks" section on this page for the current list, and check the regime dashboard for how the macro backdrop is currently tilted.

What factors could push US Recession Probability (Smoothed) lower?

The same transmission channels that drive US Recession Probability (Smoothed) higher operate in reverse when conditions flip. The risk drivers listed above map directly to scenarios that, if triggered, would pull this metric in the opposite direction. Convex aggregates these into a scenario-weighted probability distribution rather than a point forecast, so the magnitude depends on which scenarios activate.

Where does consensus see US Recession Probability (Smoothed) heading?

Rather than publish a point target that goes stale the day after release, Convex assembles consensus from the macro regime classification, active scenario probabilities, and historical base rates. Point forecasts from banks and strategists are worth reading for context, but they typically cluster around the consensus and miss the tail events that actually move markets. The scenario-weighted approach here captures that tail risk explicitly.

What is the historical range for US Recession Probability (Smoothed)?

Historical ranges for US Recession Probability (Smoothed) vary dramatically by regime. A level that is extreme in Goldilocks can be routine in Stagflation, and vice versa. The Historical Volatility section on this page describes the typical range and regime-specific behavior. For the full multi-decade history, visit the US Recession Probability (Smoothed) chart page, which includes selectable time ranges up to five years and downloadable data.

How often is the US Recession Probability (Smoothed) forecast updated?

This forecast page recalculates whenever the underlying data or regime classification changes, typically within hours of new data releases. The scenario probabilities refresh daily as the macro state is regenerated. Specific drivers listed on this page reflect the current state of the Convex regime engine, not static historical assumptions.

Is this forecast actionable for trading?

Convex forecasts are informational and educational. They describe probability distributions and regime-conditional paths rather than specific entry and exit levels. Traders and portfolio managers use them alongside other inputs including position sizing rules, risk management, and their own conviction calibration. They are not investment advice.

ShareXRedditLinkedInHN

Get forecast updates for US Recession Probability (Smoothed) and related indicators.

Forecasts are model-based projections derived from current regime classification, scenario probabilities, and historical patterns. They are not investment advice. All investments involve risk.