Regime Card
← Market Regime
Historical Regime
The market is Mildly Off: the September 25 reading is more risk-off than 77% of days in the last five years.
Line: the daily reading · Diamond: latest close · Shading: the five regimes, Risk-On at the top · Orange dot: today so far, no close yet
What happened next
What the S&P 500 did in the 5 and 20 trading days after each reading, grouped by regime. Updated every day after the market closes.
| Regime | Readings | Times entered | Typical return | How often it rose | Bad month |
|---|---|---|---|---|---|
| Risk-Off | 338 | 33 | +3.6% | 72% | −6.6% |
| Readings 338 · Times entered 33 · Average return +2.3% | |||||
| Mildly Off Today | 951 | 117 | +1.0% | 59% | −6.6% |
| Readings 951 · Times entered 117 · Average return 0.0% | |||||
| Neutral | 1833 | 206 | +1.6% | 68% | −3.9% |
| Readings 1833 · Times entered 206 · Average return +1.0% | |||||
| Mildly On | 909 | 189 | +2.1% | 75% | −2.6% |
| Readings 909 · Times entered 189 · Average return +1.5% | |||||
| Risk-On | 535 | 79 | +1.5% | 71% | −2.7% |
| Readings 535 · Times entered 79 · Average return +1.5% | |||||
4,586 daily readings since July 2008. The newest 20 have no 20-day result yet, so the columns add up to a little less.
What the columns mean
- Regime
- The regime the dial showed on the day of the reading.
- Readings
- Number of daily readings in this regime.
- Times entered
- Times the market entered this regime and stayed at least a day.
- Average return
- Average return across all readings in this regime, shown when you tap a row.
- Typical return
- The middle result. Line up every outcome from worst to best and this is the one in the center, so it is not thrown off by a few extreme months.
- How often it rose
- Share of readings after which the S&P 500 rose.
- Bad week / Bad month
- One reading in ten was followed by a fall this big or bigger.
What it means
Mildly Off is where sell-offs tend to start. Its typical month looks like Neutral, but its bad month runs about twice as deep. Risk-Off is the widest swing: by then the market has usually already fallen, so its typical month is the strongest of the five, while its bad month is just as deep as Mildly Off's.
Each reading is computed with today's formula, using only the data that existed on that day.
Every reading this site publishes is written into a record that can't be quietly changed later. Check it yourself 8259dc66…6948
How to read this
What the line shows. Regime Card is a daily read of US market risk, built from eight cross-asset signals. Each is ranked against its own history and blended into one percentile from 0 to 100: higher is more risk-on, lower more risk-off. The percentile maps to one of five regimes: Risk-Off, Mildly Off, Neutral, Mildly On, Risk-On.
The signals
Reading the chart. Every day is ranked against the selected period, so the same day can read differently when you switch it. The diamond is the latest close and may still be provisional. Hover any point for its date, reading and regime; drag across the chart to zoom, then Reset zoom to restore.
Why five years is the default
The default ranking period isn't arbitrary: the length of the comparison period decides what the score means. Too short, say a single year, and the regimes recalibrate too fast: after an extended stretch of stress the recent distribution shifts, and a given percentile stops carrying a consistent meaning over time. Too long, and the comparison drags in regimes that no longer reflect the current market structure.
Five years is the balance: wide enough to span a full market cycle, so the score holds a stable, comparable meaning across regimes rather than drifting with the last few months, and rolling forward so it stays anchored to the prevailing environment rather than a crash from a decade ago. It's consistent with the length of a typical market cycle: post-war U.S. business cycles have averaged roughly six years trough-to-trough, with wide variation (NBER).
Honest limits
Descriptive, not predictive. It describes where market risk sits today relative to history, not a forecast, not a trade signal. Percentiles are relative to the selected period, so a high or low reading reflects the chosen comparison period, not an absolute level. Informational only, not investment advice.