Five kinds of macro weather, drawn with no recession labels.
Here is 62 years of the S&P 500 on a log scale — the only honest way to compare percentage moves across decades — with every month colored by which of five recurring weather states the economy was in.
The five states were found by a model that does just one thing: it groups together months whose whole economic picture — growth, inflation, jobs, rates, credit stress, the market’s trend — looks alike. It was never told what a recession is. It saw no crisis dates, no NBER calendar, no labels at all. Yet the colored bands it drew land almost exactly on the eras every analyst already knows: the 1970–85 Great Inflation and Volcker squeeze, the 1987 / 2001 / 2008 / 2020 crisis slivers, and the long Goldilocks runs of 2014–20 and 2023–today.
That the map redraws known history so closely — with no answer key — is the first evidence the states are real structure, not noise. The five: Goldilocks, Steady Expansion, Inflation / Tight Money, Slowdown, and Crisis / Bear.
regime_ribbon
Each state has a fingerprint — and a typical lifespan.
A “weather state” is just a recipe: a typical mix of growth, inflation, jobs, rates, credit stress, and the market’s trend. The grid below is that recipe for each state. Bright blue means an ingredient ran far above its normal level; slate means far below. You can read each state’s character straight off the row.
Goldilocks runs low unemployment and a strong market with calm prices. Inflation / Tight Money lights up on inflation and rates — the 1970s–80s and 2021–22. Crisis / Bear lights up on credit stress and a falling market. The states differ on what they’re made of, not on a label someone assigned.
state_centroids
state_durations
The state you're in shifts what usually comes next.
A map is only useful if it changes your odds. So we measured what happened aftereach kind of month — the next 12 months for stocks, gold, and Treasury yields — using only returns that came after the state was assigned (no peeking). The spreads are material and they read the way intuition says they should.
Stocks do best coming out of a Steady Expansion and worst out of a Crisis. Gold does the opposite: it earns its keep in slowdowns and inflation and goes quiet in calm expansions. Same months, different assets — the state sorts them.
forward_by_state
transition_matrix
Every month, mapped — and the dot that is today.
The map below places every month since 1964 so that months with a similar overall economy sit close together, tinted by year. It’s the same idea as the ribbon, flattened onto a page: nearby dots rhymed; distant dots felt nothing alike.
Find the most recent month and look at its neighbours. Today’s closest historical company — setting aside the last couple of years, which are near-copies of now — is the late-2017 to mid-2018 expansion: low unemployment, equities near highs, a firm dollar.
analog_map
current_readout
The honest map.
This is not advice to buy or sell anything. It is a map of the recurring states the economy cycles through, what each one usually preceded, and which one we’re in today. The map is unusually good at saying where — and, by design, silent on when it changes. Observation, not advice.
Five recurring kinds of macro weather — Goldilocks, Steady Expansion, Inflation / Tight Money, Slowdown, Crisis / Bear — found with no recession labels, yet landing almost exactly on the eras everyone knows.
Each state has a stable fingerprint (a recipe of growth, inflation, jobs, rates, stress, market trend) and a typical lifespan — calm states last years, stress states pass in quarters.
The state measurably shifts the next 12 months: stocks +12.5% out of Steady Expansion vs +5.1% out of Crisis; gold +17.8% out of Slowdowns vs +2.3% in calm expansions. A record of what followed, not a forecast.
States are 93–99% sticky month-to-month. That makes nowcasting the current state trivial — and makes calling the flip, the only forecast anyone wants, exactly what this map cannot do.
June 2026: Goldilocks, ~38 months into a typical ~50-month run — late-cycle, not overdue. Closest historical rhymes: late-2017 to mid-2018, but running warmer on inflation.
- —Descriptive, not predictive. The states are fit on the whole 1964–2026 sample — every month 'knows' the global picture. It is a taxonomy of the past, never asked to forecast out-of-sample.
- —Forward-return tables read 'what historically followed this kind of month,' not 'what will happen.' They rest on a handful of independent episodes per state, not a validated signal.
- —The exact month-by-month boundaries near transitions are only moderately stable; the number and gross character of the states are robust, the fine edges are a judgment.
- —Five states is a deliberate choice for legibility. A different count gives a different, also-valid map — the map is one honest lens, not the only one.
- —The persistence that makes the current-state read easy is the same persistence that makes flip-timing impossible. This map says where you are; it does not warn before the weather turns.
- —The current inflation reading sits on a knife-edge: today is Goldilocks in character but warmer on prices than the textbook version. If inflation re-accelerates, the nearest neighbour migrates.
Descriptive research, not financial advice. The economy can leave a calm state abruptly — as it has more than once in this dataset — and this map gives no warning before it does.
Appendix — every chart & model inventory →