Adaptive markets hypothesis

Definition

Lo’s adaptive markets hypothesis (AMH) reframes the efficiency debate in evolutionary terms: markets are ecologies of competing strategies, efficiency is a STATE that waxes and wanes with the population of participants, and edges behave like species — emerging in new niches, thriving while competition is thin, and dying as capital crowds in. It reconciles this KB’s two stubborn facts: anomalies exist (against strict EMH) and anomalies decay (for it) — both are what an ecology predicts. For an engine that manages a LIBRARY of strategies, AMH is the operating philosophy: every edge has a lifecycle, and monitoring for decay is part of owning one.

How it works / structure

• The core translation: EMH’s equilibrium becomes ecology — arbitrage opportunities are resources; strategies are species consuming them; profitability attracts entry (population growth) until the resource depletes (documented post-publication anomaly decay is the signature prediction); dislocations that remove capital (2008 — episode-gfc-2008) leave surviving strategies richer niches (documented post-crisis premium widenings: converts in 2009, the record this KB carries in instrument-convertible-bond).
• Behavior as adaptation, not error: heuristics (bias-herding and the pillar-13 catalog) are evolved responses fit for ancestral environments, misfiring in novel ones — AMH’s reading of behavioral finance: biases persist because environments change faster than heuristics.
• Risk-premium instability (engine-relevant): under AMH, factor premia (strategy-factor-investing) are not constants but population-dependent — expected to vary with crowding, which the documented factor-timing and factor-crowding literature partially supports (labeled: measurement is hard).
• The strategy-lifecycle discipline it implies: track each live strategy’s realized edge vs its documented baseline (decay monitoring); expect regime-dependence (regime-volatility conditioning); maintain a research pipeline (niches open as others close); retire strategies on evidence, not loyalty.

When it applies

Strategy-library governance (the platform’s exact use case — allocation across strategies with lifecycle states); anomaly-decay expectations (published edges are dated the day they publish); post-dislocation opportunity mapping (capital destruction reopens niches — the documented pattern); crowding surveillance (the strategy’s population, not just its signal, is a monitored variable).

Risk profile & failure modes

• Unfalsifiability risk (the honest critique): AMH explains both persistence and decay — a framework that fits everything predicts weakly; the platform uses it as a DISCIPLINE (lifecycle monitoring), not as a forecasting engine, and labels it so.
• Premature retirement: normal drawdowns (style-value-investing 2007-2020) are indistinguishable from decay for years — the retire-on-evidence rule needs pre-registered decay criteria, or loyalty and panic replace analysis.
• Crowding measurement softness: strategy populations are estimated (13F proxies, flow data, spread compression) — the ecology’s census is always approximate.
• Novelty worship: “old edges die” can rationalize chasing untested strategies — the lifecycle claim cuts both ways: new niches are unproven by construction.

Evidence & limits

Lo (2004) is the framework statement; its supporting evidence is the documented anomaly-decay record, post-crisis premium episodes, and time-varying factor premia (all labeled where cited in this KB). AMH is a philosophy of evidence management more than a tested hypothesis — carried as such.

Falsifiable-thesis examples

Illustrations only, not signals:

• “Strategy X’s rolling 3-year information ratio has fallen below half its documented baseline for 2+ years with rising crowding proxies (retirement criteria — pre-registered)” — graded by the monitored series.
• “Post-dislocation quarters (VIX>40 episodes) are followed by above-baseline returns in capital-intensive arbitrage strategies (niche- reopening check)” — falsified by the episode-cohort measurement.

Cross-references

• The debate it reframes: philosophy-efficient-markets
• The lifecycle subjects: strategy-factor-investing, style-trend-following-school (post-2010 softness)
• The behavioral reading: bias-herding (pillar 13 as adaptations)
• The governance machinery: risk-scenario-analysis, disc-process-vs-outcome