Joshua Dunnink ยท Trading Systems

Regime cards for strategies that work in some markets

Many trading ideas earn in one kind of market and give it back in another. We do not throw such an idea away, and we do not hide that it is conditional. Every research system gets a regime card, and a card is only filled in when we can back each line with a measurement.

What a regime card holds

  1. Where it earns. The kind of market, described in words and measured on separate periods that do not overlap, not on one lucky stretch.
  2. Where it does not. The same measurement for the periods where it fails, stated plainly. A system with no known bad period has not been tested hard enough.
  3. The state that separates the two. A number computed from price alone, before any trade, that tells the two kinds of market apart. It must be calibrated against the market's own recent history rather than against a fixed level, and it must have a reason that explains the split. For range breakouts that state is how often breaks keep going.
  4. The switch. The state is built into the system, which stands aside when the market is not in its regime and resumes by itself. The system's chart shows whether it is trading or standing aside, and why.
  5. How much evidence the switch rests on. We count the independent swings between the two kinds of market in the data. One or two swings is a guide, not proof, and the card says so.

Rules we keep

Why this matters

Markets rotate. A system that is waiting today can be the one doing the work next year, and the opposite is just as true. What carries over is the discipline: knowing which market a system belongs to, measuring whether that market is here, and acting on the measurement. See the follow-through note for the first measure we run in this way.