The US Market Regime on September 30, 2026, in Numbers

These figures are as stored for September 30, 2026 and do not update. The live regime reading is at /macro/regime.

As of September 30, 2026, Kresmion's cross-asset model reads the US market regime as Strong Risk-Off with HIGH conviction (75%). In plain terms: three of the four factors point risk-off, and the smoothed score is the lowest the model has stored. This page walks through the numbers behind that read.

Kresmion rates the US market regime Strong Risk-Off with HIGH conviction (75%), on a smoothed cross-asset risk score of -0.7267, the lowest of the 154 daily readings stored since April 30, 2026.

What the model said on September 30, 2026

The Kresmion macro regime model combines fifteen cross-asset and macro inputs into four factors. Growth uses the 10-year minus 2-year Treasury spread, an economic surprise index, ISM manufacturing and services PMI, initial jobless claims, and the distance of 10-year breakeven inflation from 2.25 percent. Liquidity uses the 10-year real yield, the dollar index, the US high-yield credit spread, the Chicago Fed financial conditions index (NFCI), and net Fed liquidity, the Fed's balance sheet net of the overnight reverse repo facility and the Treasury's cash account. Risk appetite uses S&P 500 momentum, the high-yield spread, AUD/JPY, and the copper to gold ratio. Volatility uses VIX and MOVE. Most inputs are scored on their change over the last five observations, as a z-score against that input's own history of such changes, clipped at plus or minus 3. The factors are weighted 20 percent Growth, 40 percent Liquidity, 30 percent Risk appetite and 10 percent Volatility; the result is smoothed (70 percent of the day's raw score plus 30 percent of the previous day's smoothed score) and mapped to five labels from Strong Risk-Off to Strong Risk-On at fixed edges, with no hysteresis. The Kresmion methodology page has the full formula.

On the September 30, 2026 row of the Kresmion `macro_regime_history` dataset, the regime label is Strong Risk-Off, the raw risk score is -0.7542, and the smoothed score is -0.7267. Conviction is HIGH at 75.00%: three of the four factors are negative. The Strong Risk-Off label has held for 2 days; it began on September 29, when the score crossed the Strong Risk-Off edge after four days labelled Risk-Off.

The sign convention matters: positive readings tilt risk-on, negative readings tilt risk-off. The labels band the smoothed score at about plus or minus 0.17 (Risk-On or Risk-Off) and plus or minus 0.5 (the two Strong labels), so -0.7267 sits well past the Strong Risk-Off edge. The 75% conviction figure is the share of the factors that agree on direction, not a probability of forward returns.

Key takeaways

The four factors, decomposed

The four contributions sum to the raw score of -0.7542. Each factor is the average of its inputs' z-scores, with the inputs that rise when conditions tighten sign inverted:

• Liquidity (-1.15) is the main drag. The 10-year real yield's five-day rise scored +3.00, the clip limit (FRED series DFII10 went from 2.62 to 2.90 percent between September 21 and 28), and the high-yield credit spread's five-day widening scored +2.31; both are inverted, so they subtract. The dollar (+0.38), NFCI (+0.08) and net Fed liquidity (+0.01) moved little.
• Risk appetite (-0.98) is pulled down by AUD/JPY, whose five-day move scored -2.07, and by the same high-yield spread widening, counted here too. The copper to gold ratio (+0.56) partly offsets them; S&P 500 momentum (-0.12) is close to its usual range.
• Volatility (-0.97) reflects MOVE, the bond-market volatility index, whose five-day move scored +1.72, with VIX's five-day move close to normal (+0.22); both are inverted, so unusually large rises in volatility lower the factor.
• Growth (+0.49) is the one positive factor. The 10-year minus 2-year spread steepened, scoring +2.10 (FRED series T10Y2Y went from 0.25 to 0.37 points between September 22 and 29), and the economic surprise index read +1.01. Breakeven inflation's distance from 2.25 percent (+0.53, inverted), PMI (-0.15) and jobless claims (-0.02, inverted) were small.

What moved the score

Across September 24 to 30, 2026 the smoothed score fell from +0.0941 (Neutral) to -0.7267, and the raw score from +0.0462 to -0.7542. Liquidity did most of the work, falling from -0.0962 to -1.1505; risk appetite fell from +0.3256 to -0.9831 and volatility from -0.4420 to -0.9670, while growth rose from +0.1559 to +0.4881.

The label moved from Neutral to Risk-Off on September 25, held Risk-Off through September 28, and became Strong Risk-Off on September 29, when the raw score dropped from -0.4529 to -0.7644 in one day. Kresmion announces a regime change on its public channels only after the new label has held for five consecutive daily readings, so a two-day-old Strong Risk-Off label is not yet an announced shift.

One caveat for reading day-to-day moves: inputs update on their own cadences. On the September 30 row the real-yield reading is dated September 28, the high-yield spread and the yield curve September 29, net Fed liquidity September 23 (a weekly series) and PMI September 1 (monthly), so part of a day's change can be data-refresh timing rather than fresh market action.

Frequently asked questions

What is the US market regime as of September 30, 2026?

The Kresmion macro regime model labels it Strong Risk-Off with HIGH conviction (75.00%), on a smoothed risk score of -0.7267 and a raw score of -0.7542, from the September 30, 2026 row of the `macro_regime_history` dataset.

Which factor was driving the regime score on September 30, 2026?

Liquidity, at -1.1505, which carries 40 percent of the weight and contributed -0.4602 of the -0.7542 raw score. Inside it, the five-day rise in the 10-year real yield (z +3.00, the clip limit) and the widening high-yield credit spread (z +2.31) did most of the work.

Has the regime changed recently?

Yes. The label was Neutral on September 24, Risk-Off from September 25 to 28, and Strong Risk-Off from September 29. The smoothed score of -0.7267 on September 30 is the lowest of the 154 daily readings stored since April 30, 2026.

Does a high conviction reading predict returns?

No. The 75.00% conviction figure is the share of the four factors that point the same way, not a probability of forward market returns.

Updated October 1, 2026: this page now reports the regime as stored for September 30, 2026. An earlier version reported the reading of June 15, 2026 and misdescribed the model: it placed several inputs in the wrong factors, called the label banding hysteresis, and said the weights were not published.

---

Source: Kresmion Research, `macro_regime_history` dataset (row for 2026-09-30); FRED series DFII10 and T10Y2Y. This is information, not advice.