This article explains the idea of regime-switching as a general approach to markets. It is educational, not personalised investment advice, and nothing here is a recommendation to buy or sell any security or to adopt any particular strategy.
Every strategy has a market it was born for. Trend-following loves a sustained advance and bleeds in a chop. Mean-reversion prints in a range and gets run over by a trend. Breakout buying works when participation is broad and fails when it’s narrow. None of these is “good” or “bad”, each is simply matched to a market state, and the market does not stay in one state. That is the whole case for regime-switching: instead of riding a single approach through conditions it was never suited to, you identify which state the market is in and deploy the approach that fits it.
What a “regime” is
A market regime is a persistent condition that changes the odds of what works. The common ones are intuitive: a broad uptrend (most names advancing together), a narrow or deteriorating tape (a few leaders masking weakness underneath), a range-bound chop (no net direction), and a decline (broad selling). These are not forecasts. A regime is a statement about the present, what the tape is doing now, not a prediction of what comes next.
The reason regimes matter is that the same setup has different odds in different ones. A clean breakout follows through far more often when participation is broad than when it is narrow. A pullback buy is constructive in an uptrend and a falling knife in a decline. If you only measure the setup and ignore the regime, you are reading half the picture, and the half you’re ignoring often decides the outcome.
Why a single strategy can’t span every market
The temptation is to find one strategy that works “always.” It does not exist, and the reason is structural, not a failure of effort. The edges that make a strategy work in one regime are the same edges that make it lose in another:
- A trend strategy makes its money by holding through volatility and adding to strength. In a chop, that exact behaviour buys every false start and sells every shakeout, the holding discipline becomes a liability.
- A mean-reversion strategy makes its money fading extremes. In a strong trend, the extreme keeps extending, and fading it is a sequence of small wins ended by one large loss.
- A breakout strategy needs follow-through, which needs participation. In a narrow tape, breakouts fail at a far higher rate regardless of how clean the individual chart looks, the point of how breakout setups work.
Each strategy is not flawed; it is specialised. Asking one of them to perform across all regimes is asking it to be something it isn’t. The realistic options are to accept that a single strategy will have long stretches where it is the wrong tool, or to switch tools as the regime switches.
The two hard parts
Regime-switching sounds obvious stated this way. It is hard for two specific reasons, and most attempts fail on one of them.
First, identifying the regime without using the future. It is trivial to label regimes in hindsight and trivially useless. A regime signal has to be computable from data available on the day, using a rule that doesn’t secretly peek ahead. Breadth, how many names across the universe are in healthy trends, is one of the few reads that is both forward-honest and slow-moving enough to be actionable. How Shishin turns universe-wide breadth into a daily regime read is the subject of the macro regime classifier.
Second, switching without whipsawing. A regime signal that flips every few days will have you changing strategies at exactly the wrong moments, paying the transition cost repeatedly and capturing none of the benefit. A usable regime model has to be deliberately sticky, slow to change its mind, which means accepting that it will be a little late to every turn in exchange for not being fooled by noise. The lateness is the price of not whipsawing, and it is worth paying.
Switching strategies vs. switching exposure
There are two ways to act on a regime read, and they are often confused. The crude version is a portfolio-level switch: go fully in during good regimes, fully to cash in bad ones. This is the kind of all-or-nothing timing that sounds prudent and usually underperforms, because the cost of being wrong on the switch, out during a sharp recovery, in for a sharp drop, is enormous, and the signal is never precise enough to earn it back. The same expected-value problem sinks portfolio-level circuit breakers, which is why Shishin doesn’t use a kill switch.
The version that holds up is switching which strategy is allowed to act, not whether to be in the market at all. The regime read decides which specialised approach fits today’s conditions; that approach then makes its own per-name decisions with its own risk controls. The market exposure follows from the opportunities the active strategy finds, rather than from a top-down bet on direction. It is a rotation of method, not a timing of the index.
How Shishin applies it
Shishin runs four scoring engines, each specialised for a different market state, and a macro classifier decides which one is allowed to fire on any given day. A breakout-oriented engine runs when participation supports it; a more defensive engine takes over when the tape deteriorates; the others cover the states in between. The full architecture, the four engines and the regimes they map to, is described in four engines for four regimes, and the reasoning behind giving regimes names rather than factor labels is in the four guardians framework.
The point of the architecture is not cleverness. It is that no single engine has to pretend to be good at everything. Each one is allowed to be narrow, excellent in its regime, silent outside it, and the classifier handles the handoff. The system’s job is to keep matching the tool to the state, day after day, without whipsawing on the transitions.
What to ask of any regime-switching claim
If a strategy claims to switch regimes, three questions separate the real ones from the hindsight ones:
- Is the regime computed from same-day data? Or is it labelled after the fact, using information that wasn’t available when the decision had to be made?
- Does it switch method, or just time the market?All-or-nothing cash calls are the fragile version; rotating which approach acts is the durable one.
- Is it sticky enough to survive noise? A model that changes its mind weekly will whipsaw away whatever edge the switching was supposed to capture.
Regime-switching is not a way to win in every market. It is a way to stop losing in the markets a single strategy was never built for, by being honest that no one approach fits them all, and matching the tool to the state instead.
Sources & further reading
- Hamilton, J. D. (1989). “A New Approach to the Economic Analysis of Nonstationary Time Series and the Business Cycle.” Econometrica, 57(2), 357 to 384.
- Ang, A. & Bekaert, G. (2002). “International Asset Allocation with Regime Shifts.” Review of Financial Studies, 15(4), 1137 to 1187.