The Hurst Exponent and Market Efficiency

The efficient market hypothesis (EMH) in its strong form implies that prices follow random walks—future price changes cannot be predicted from past information. A random walk corresponds to a Hurst exponent of exactly 0.5. When researchers find Hurst values significantly different from 0.5, does this refute market efficiency? The relationship between Hurst and efficiency is more subtle than this binary framing suggests.

The Random Walk Benchmark

A true random walk has specific statistical properties:

• Each step is independent of previous steps
• No serial correlation in returns
• Variance scales linearly with time
• Hurst exponent equals 0.5

If markets were perfectly efficient and prices fully reflected all available information, price changes would be random—driven only by new information that is by definition unpredictable. This would produce H = 0.5.

In practice, most financial time series show H values between 0.5 and 0.7, suggesting some degree of persistence. Does this mean markets are inefficient?

Forms of Market Efficiency

The EMH exists in three forms:

Weak form: Prices reflect all historical price information. Technical analysis cannot earn excess returns.

Semi-strong form: Prices reflect all publicly available information. Neither technical nor fundamental analysis can earn excess returns.

Strong form: Prices reflect all information, including private information. No one can earn excess returns.

A Hurst exponent different from 0.5 most directly challenges the weak form—it suggests past prices contain information about future prices.

Persistence Without Profit

A critical distinction: statistical persistence does not automatically imply exploitable inefficiency. H > 0.5 means price changes are positively correlated, but:

• The persistence may be too small to overcome transaction costs
• The persistence may be unstable (varying over time)
• Acting on persistence may move prices (market impact)
• Risk-adjusted returns may not exceed a risk-free alternative

Market efficiency is about risk-adjusted profits net of costs, not raw statistical patterns. A statistically significant H > 0.5 does not automatically translate to profitable trading.

The Economic Versus Statistical Distinction

There is a difference between statistical significance and economic significance:

Statistical significance: H is reliably different from 0.5 in the data.

Economic significance: The difference generates profits after accounting for costs, risk, and implementation challenges.

Many studies find statistically significant departures from H = 0.5 that do not translate to economic profits. This is consistent with markets being approximately efficient—deviations exist but are difficult or impossible to exploit profitably.

Time-Varying Efficiency

Modern research recognizes that market efficiency varies over time:

• Efficiency may be lower in emerging markets than developed markets
• Efficiency may be lower during crises and higher during calm periods
• Efficiency may improve over time as markets mature
• Different instruments may show different efficiency levels

The Hurst exponent measured over rolling windows can track these efficiency changes. Rising H (toward 0.5) suggests improving efficiency; falling H (toward extremes) suggests deteriorating efficiency.

Adaptive Markets Hypothesis

Andrew Lo's adaptive markets hypothesis (AMH) provides a framework reconciling efficiency with predictability. The AMH proposes that:

• Markets are competitive ecosystems with evolving participants
• Efficiency varies based on market conditions and competition
• Profitable opportunities arise and are arbitraged away
• The degree of efficiency is not constant but adaptive

Under the AMH, a time-varying Hurst exponent is expected. Periods of H significantly different from 0.5 represent temporary inefficiencies that attract arbitrage activity, eventually pushing H back toward 0.5.

What Hurst Values Mean for Efficiency

Different Hurst values suggest different efficiency implications:

These categories are approximate. Context matters significantly.

Practical Interpretation

For practitioners, the efficiency question matters less than practical applicability:

1. Measure current Hurst exponent
2. If H ≈ 0.5, apply approaches that assume unpredictability
3. If H significantly deviates, test whether deviation is exploitable
4. Account for transaction costs, market impact, and risk
5. Monitor whether the Hurst estimate is stable or shifting

Whether markets are "efficient" in some philosophical sense matters less than whether measurable structure can inform better decisions.

The Empirical Evidence

Decades of research produces a nuanced picture:

• Most developed equity markets show H between 0.5 and 0.6
• Emerging markets often show higher persistence
• Commodities and currencies show variable results
• Short-term returns tend toward H = 0.5; longer horizons show more persistence
• Time-varying estimates show efficiency is not constant

The evidence suggests markets are approximately efficient but not perfectly so, with efficiency varying across markets, instruments, and time periods.

Conclusion

The Hurst exponent provides a quantitative measure related to market efficiency, but the relationship is not simple. H ≠ 0.5 suggests statistical dependence in returns, but statistical dependence does not automatically imply profitable opportunity. Markets can show persistent patterns while remaining economically efficient if those patterns cannot be profitably exploited after costs and risk adjustment.

For practical purposes, the Hurst exponent is better viewed as a regime indicator than an efficiency test—a measure of market character that informs strategy selection rather than a verdict on market efficiency.