Refresh It or Retire It?

Once a review has established that a strategy really is decaying — not just riding out an ordinary losing streak, but showing a persistent, explainable erosion of its edge — the question shifts from "is something wrong" to "what do we do about it." Broadly there are two paths: refresh the strategy, re-fitting or reworking it to adapt to whatever has changed, or retire it, winding it down and freeing its capital and risk budget for something else. Neither path is automatically right, and picking between them badly wastes either a still-viable idea or a still-scarce risk budget.

The case for refreshing is strongest when the cause of decay is identifiable and fixable without changing the strategy's core logic: transaction costs have risen because the venue mix shifted, so re-optimizing execution routing restores most of the edge; a signal's optimal lookback window has drifted, so re-fitting the parameter on more recent data recovers performance; a universe has grown and the strategy needs updated liquidity filters. These are maintenance problems, not existential ones, and refreshing is typically cheaper and faster than building something new. The case for retiring is strongest when the cause is structural and outside the strategy's control: the underlying market inefficiency has been arbitraged away by competitors, a regulatory change has closed off the mechanism entirely, or the strategy's edge depended on a market structure that no longer exists. No re-fitting recovers an edge that the world has genuinely closed.

A useful test: after diagnosing the cause of decay, ask whether a knowledgeable competitor building a similar strategy today, with no legacy attachment to this one, would still choose to build it. If the honest answer is "yes, but they'd tune the parameters differently" or "yes, but on updated data," that points to refresh. If the honest answer is "no — the opportunity this exploited just doesn't exist at meaningful scale anymore," that points to retire, however painful that is for whoever built it. A strategy that has been refreshed three times in eighteen months, each time buying only a few more months of marginal performance, is usually a strategy answering that question honestly for the questioner, even if no one wants to hear it.

What this means in practice

The decision is made harder by the fact that the people closest to a strategy — its builders and the desk trading it — have the least incentive to conclude "retire," which is exactly why this decision, like capital allocation itself, usually goes through a process involving people without that attachment. Retiring a strategy well is also a skill in its own right, distinct from deciding to do it, since an abrupt unwind can itself move markets and destroy value on the way out.