Market regimes

A market does not behave the same way all the time. Volatility clusters, trends persist and then vanish, and correlations that held for months break in a week. The comfortable model — one stationary distribution, parameters estimated once — describes a market that does not exist.

Regime thinking starts from a different premise: the market switches between a small number of distinct states, each with its own behaviour. Bull and bear, quiet and volatile, trending and mean-reverting. The analytical task is to know which state the market is in now — not to predict what it will do next.

That distinction is the whole point. A regime classifier is an input to a decision, not the decision. Tools that blur classification into trading advice get trusted more than they have earned, and they fail exactly when trust matters: at the transition, where the old regime's statistics stop applying.

Returns are not normally distributed. Extreme moves happen far more often than a bell curve allows, which means risk estimates built on normality are systematically optimistic. Mandelbrot made this argument decades ago; markets keep re-proving it. Any regime model that assumes calm Gaussian behaviour inside each state inherits the same blind spot, so heavy-tailed distributions or rank-based measures are the safer default.

A series sitting between two regimes should be reported as ambiguous. Forcing it into the nearest category manufactures confidence the data does not contain. Fuzzy or probabilistic assignment — "60% quiet-bull, 40% transition" — is more honest and more useful, because the reader learns both the classification and how much to trust it.