No strategy earns all the time. Markets alternate between states in which a given edge is paid and states in which the same edge is quietly taxed: trending and mean-reverting phases, calm and stressed liquidity, orderly sessions and event-driven chaos. Averaged over a long backtest, these states blur into a single number. Managed in real time, they are the difference between compounding and bleeding.
The Conditional Nature of Edge
We treat every validated strategy as a conditional claim, not a universal one. The claim is not that the strategy earns, but that it earns when specific conditions hold: a volatility range, a liquidity profile, a market state the strategy was built to exploit. Validation therefore includes mapping where the edge lives, and just as importantly, where it demonstrably does not.
This mapping changes behavior. A strategy that earns in quiet, range-bound conditions should shrink or stand aside when volatility regimes shift, rather than average through the storm on the strength of a long-run backtest. The long-run number was earned by a mixture of states; there is no obligation to accept the losing states in live trading when they can be identified.
Volatility as an Input, Not a Surprise
Volatility forecasting is the workhorse of this discipline. Realized volatility clusters and moves in persistent regimes, which makes it one of the few quantities in markets that is genuinely forecastable. We use those forecasts primarily for position sizing: scaling exposure down as expected turbulence rises, so that a fixed risk budget buys a variable number of contracts rather than a fixed bet with variable risk.
Calendar and session structure supply another layer. Liquidity, spreads, and the character of price movement differ across trading sessions, around scheduled announcements, and near expiries and rebalancing dates. Strategies that ignore this structure pay for it in execution and in exposure to moves they were never designed to hold through. Ours carry explicit session logic and event calendars as part of their specification.
The Standing Down Option
Guards and vetoes are therefore first-class components of every deployment, engineered and validated with the same rigor as the entry signal. A volatility guard, an event blackout, a liquidity floor: each is a rule that can override the strategy’s desire to trade. In our accounting, a veto that avoids a bad state contributes to performance just as surely as a signal that catches a good one.
Being flat is a position, and choosing it is a decision that deserves the same evidence as any other. The strategies that survive longest in our inventory are rarely those with the most aggressive signals; they are those that best understand when their own edge is absent. Knowing when not to trade is not the absence of a strategy. It is often the largest part of one.