The Stages of a Strategy's Life

A trading strategy is not a static thing you either have or don't. It is born as an idea, gets tested, gets trusted with a little money, gets trusted with more, eventually stops working as well as it used to, and is either refreshed or retired. Every strategy that has ever traded, from a single-signal quant book to a discretionary macro fund, has moved through some version of the same stages. Knowing which stage a strategy is in matters because the right question to ask is different at each one — "is this idea worth building" is not the same question as "is this live book decaying," and applying the wrong stage's mindset is one of the most common ways teams waste capital or research time.

The stages

Stage	Core question	Typical duration
Idea and hypothesis	Is there a plausible economic or behavioural reason this should work?	Days to weeks
Backtest and research	Does the idea survive a rigorous historical test, honestly done?	Weeks to months
Incubation / paper trading	Does the signal still behave the way the backtest promised in live, un-traded conditions?	Weeks to a year, strategy-dependent
Small live allocation	Does real execution — costs, slippage, capacity — match what research assumed?	Months
Ramp and growth	How much capital can this absorb before its edge degrades?	Months to years
Mature and monitored	Is performance holding up against its expected envelope?	Years, if healthy
Decay	Is the edge fading from crowding, cost inflation, or a genuine break in the underlying effect?	Months, once it starts
Retirement or refresh	Should this be shut down, resized, or rebuilt with a new version?	A decision point, not a duration

These stages are not always strictly sequential. A strategy can be pulled straight from incubation back to research after a paper-trading failure, or jump from small live allocation directly toward retirement if an unexpected loss reveals a flaw. But the stages themselves, and the questions attached to each, recur across almost every strategy a desk ever runs.

Why the boundaries matter

Each transition between stages is a decision point where a team commits more — or less — resource to a strategy, and each one benefits from being made deliberately rather than by drift. Moving from backtest to incubation without agreeing what "the incubation period succeeded" means invites a team to simply keep waiting for good performance rather than making an honest call. Moving from a small live allocation to a full ramp without a capacity study risks discovering the strategy's real capacity constraint only after it's been breached. And staying in the "mature and monitored" stage without a live definition of what decay looks like is how strategies quietly bleed capital for months before anyone acts.

Worked example: a signal moving through the stages

A researcher notices that a fundamental accounting ratio predicts one-month-ahead returns in a historical sample. The idea stage asks whether there's a plausible reason — perhaps the ratio proxies for a risk the market underprices, or captures information that's slow to be reflected in price. The backtest stage tests this rigorously across time periods and universes, controlling for known factors, and survives. Incubation runs the signal live on paper for six months against a pre-agreed bar: does the live IC roughly match the backtested one. It does, so the strategy moves to a small live allocation, where the team discovers actual transaction costs eat a third of the paper edge — a real, useful finding that reshapes position sizing before any further capital is committed. After that adjustment, the strategy ramps over eighteen months to its target size, runs cleanly for three years, and then begins showing a slow decline in IC that a rolling significance test flags as real rather than noise. The team faces the retirement-or-refresh decision, and chooses to rebuild the signal with an updated universe rather than shut it down outright — a new version enters the idea stage, and the cycle restarts.

In practice

• Name the stage a strategy is in during every review. A five-minute disagreement about whether a book is still "ramping" or already "mature" often reveals a deeper disagreement about what success looks like.
• Write down the exit criteria for each stage before entering it, not after results start coming in — see Defining Kill Criteria Before Launch for the specific discipline of setting these in advance.
• Incubation length should be a deliberate choice, not a default, covered in How Long to Incubate a Strategy.
• Decay detection deserves its own discipline, not a vague sense that recent performance feels worse; see Detecting Decay in a Live Strategy.
• Have a crisis plan ready before you need it. The stage nobody plans for in advance is the one covered in The First 24 Hours of a Blowup, and that is exactly why it deserves planning.