Efficient markets hypothesis

Definition

The efficient markets hypothesis (EMH) states that prices reflect available information — in Fama’s taxonomy, weak form (past prices are in the price: charts alone can’t beat it), semi-strong form (public information is in the price: news is priced on arrival), and strong form (even private information is in — contradicted by documented insider profitability). It is this KB’s most important adversary and its best discipline: every entry claiming an edge is implicitly claiming a SPECIFIC, bounded failure of semi-strong efficiency, and owes an account of why that failure survives.

How it works / structure

• The mechanism: competition among informed traders arbitrages predictability away — any reliable pattern attracts capital until it doesn’t pay; efficiency is not an assumption about wisdom but an equilibrium claim about incentives.
• The Grossman-Stiglitz repair (the version the platform holds): PERFECT efficiency is impossible — if prices reflected everything, no one would be paid to gather information, so no one would, so prices couldn’t; markets must be exactly inefficient enough to compensate the marginal information gatherer. Edges therefore EXIST, are COSTLY, and are COMPETED — the equilibrium is near-efficiency with paid frictions, which is this platform’s operating assumption stated formally.
• The evidence ledger (both columns): FOR — active- fund underperformance (style-passive-indexing), the speed of event pricing (earnings moves complete in minutes, event-earnings), failed-pattern graveyards; AGAINST — the documented anomaly families (momentum, post-earnings drift, value, quality — pillar-wide citations), bubble episodes (episode-dotcom-2000), and structural-flow predictability (event-index-rebalance — forced flows are not information).
• The joint-hypothesis problem (Fama’s own caveat): every efficiency test is simultaneously a test of the pricing model used — “anomaly” and “misspecified risk model” are observationally hard to separate; the KB labels this on every factor entry.

When it applies

Every edge claim (the null hypothesis the thesis must displace — “why does this survive competition?” is the platform’s mandatory question); post-publication decay expectations (documented attenuation of published anomalies is EMH working in real time); strategy triage (claims with no answer to “who is the loser and why do they persist?” — structural flows, behavioral bias, risk-bearing, horizon — are rejected before backtesting).

Risk profile & failure modes

• Over-belief: treating EMH as literally true makes documented structural edges (forced flows, liquidity provision) invisible — the Grossman-Stiglitz version keeps the search rational.
• Under-belief: treating every backtest artifact as an inefficiency is how data mining ships to production — EMH is the prior that makes overfitting expensive to believe.
• Form confusion: weak-form evidence against chart-only prediction does not preclude information-based or structural edges — arguments must match forms.
• Regime blindness: efficiency varies by habitat (mega-cap US equities vs micro-caps vs distressed debt) — the null’s strength is asset-class-specific, documented via anomaly concentration in low-competition habitats.

Evidence & limits

Fama (1970) and Grossman-Stiglitz (1980) are the canonical statements; the anomaly and attenuation literatures are cited throughout this KB pillar by pillar. The hypothesis is not settled — it is the standing argument the platform’s falsifiable-thesis architecture is designed to have honestly, one thesis at a time.

Falsifiable-thesis examples

Illustrations only, not signals:

• “Published anomaly X’s post-publication return is less than half its in-sample return (efficiency- in-action check)” — falsified by the split-sample measurement.
• “This strategy’s edge survives after naming its loser and testing that the loser’s constraint persists (Grossman-Stiglitz audit)” — falsified when the constraint’s removal kills the replay edge.

Cross-references

• The practical corollary: style-passive-indexing
• The evolutionary repair: philosophy-adaptive-markets
• The standing counterexamples: strategy-factor-investing (joint-hypothesis labeled), event-index-rebalance (non-informational flows)
• The rival philosophy: philosophy-reflexivity