The forecasting industry sells point estimates. The decisions that actually matter for portfolios sit one level above: which regime the market is in, and which one it is migrating toward. A correct point forecast in the wrong regime is often less useful than a directional view in the right one.
What a regime is, mechanically
A regime is a persistent joint distribution of returns, volatility, and cross-asset correlation. It is defined by the variables that determine investor behaviour at the margin — typically some combination of:
- The level and direction of real rates.
- The trajectory of inflation and its dispersion across components.
- Liquidity conditions, both funding and market.
- The dominant marginal investor and their risk budget.
- The credibility of policy reaction functions.
Regimes are stable because each of these inputs is institutionally slow to change. They shift when one or more of those inputs breaks its prior relationship — for example, when correlation between stocks and bonds inverts, or when central bank reaction functions are perceived to change target.
Why regimes matter more than forecasts
Three reasons:
- Strategy performance is regime-conditional. Risk parity, trend following, carry, value, and short-vol all have different Sharpe ratios in different regimes. The choice of strategy implicitly bets on a regime, whether the manager admits it or not.
- Drawdowns happen in regime transitions, not within regimes. Within a regime, mean-reversion logic works. Across a transition, the same instinct produces compounding losses, because the prior conditional distribution no longer holds.
- Position sizing matters more than direction. In a stable regime, getting direction right is most of the work. In a transition, the entry timing and size determine survival, regardless of whether the long-run call is correct.
Identifying transitions
Three signals tend to precede regime transitions:
- Correlation instability. Joint return distributions begin to break down before realised volatility moves. Watch for sustained shifts in cross-asset correlation, not single days.
- Policy credibility re-pricing. Markets begin to question reaction functions before officials change them. Term premium, breakeven inflation, and forward rate volatility tend to move first.
- Positioning convexity. When the majority of positioning is on one side, even small flow changes produce disproportionate price moves. That fragility is itself a regime signal.
None of these tells you what comes next. They tell you the current regime is losing structural integrity, which is a different and often more actionable question.
How to use this lens
For research framing, the discipline is to ask three questions in order before any positioning question:
- What regime are we in, defined by which variables?
- Which inputs are likely to break first?
- What are the two or three plausible successor regimes, and what would confirm each?
A view sized to a forecast requires the forecast to be correct. A view sized to a regime requires only the regime to persist or transition in a recognisable way. The second is a more honest description of what is actually knowable.