The Strategy Graveyard and Institutional Memory

When a strategy stops working and gets retired, the natural instinct is to move on and forget about it. But without a record of why it was retired — the alpha decayed, a data source got arbitraged away, a regime shift broke the assumption, or it simply never worked as backtested predicted — that knowledge leaves with whoever built it. A new researcher joining a few years later, with no visibility into past failures, can easily re-discover the same idea, re-test it, and re-learn the same lesson at the cost of real research time and possibly real capital.

A strategy graveyard is simply a maintained log of every retired strategy: what it was, why it was shut down, what the live performance actually looked like versus the backtest, and any known reason for the gap. Firms that take this seriously treat it as a searchable resource researchers are expected to check before starting new work in an adjacent area, not a dusty file nobody opens.

The graveyard is also useful in the other direction: a strategy retired for a reason that no longer applies — a data vendor fixed a bug, a market structure change reversed — is a candidate for revival, and that opportunity is invisible if the record doesn't exist.