{
  "id": "14721182-bbae-5f1b-9714-9ceeb4e25a56",
  "slug": "random-walk",
  "term": "Random Walk",
  "aliases": [],
  "category": "Quantitative Finance",
  "category_slug": "quantitative-finance",
  "difficulty": "basic",
  "definition": "A Random Walk is a mathematical model in which successive changes in a variable — such as an asset price — are independent and identically distributed, meaning past price movements contain no information about future price movements and each step is determined purely by chance. In finance, the random walk hypothesis, associated with Eugene Fama's Efficient Market Hypothesis, asserts that stock prices move in a way that cannot be consistently predicted, implying that active investment management cannot reliably outperform a passive benchmark after costs.",
  "key_takeaways": [
    "The random walk hypothesis is closely linked to the Efficient Market Hypothesis: if all available information is reflected in prices, then future price changes can only be caused by new, unpredictable information.",
    "A random walk with drift adds a constant expected return (the equity risk premium), reflecting that stocks on average appreciate over time even while individual steps are unpredictable.",
    "The Hurst exponent measures the degree to which a time series deviates from pure random walk behavior: H = 0.5 indicates a random walk, H > 0.5 indicates trending (persistence), and H < 0.5 indicates mean-reversion.",
    "Statistical tests for random walk behavior include the Augmented Dickey-Fuller (unit root) test, the variance ratio test, and the runs test.",
    "Active managers and quantitative researchers challenge the random walk hypothesis by identifying persistent anomalies — momentum, value, low volatility — though debate continues about whether these reflect true inefficiencies or compensated risk exposures."
  ],
  "detailed_explanation": "The random walk model was popularized by Louis Bachelier's 1900 thesis 'Théorie de la Spéculation,' which modeled stock prices as Brownian motion decades before Einstein's celebrated derivation in physics. The formal statement in finance is that P_t = P_{t-1} + ε_t, where ε_t is an independently and identically distributed (i.i.d.) random variable with zero mean and constant variance. This structure implies that the best forecast of tomorrow's price is simply today's price — a property known as the martingale property.\n\nThe intellectual connection between the random walk and market efficiency is profound. If markets are efficient in the semi-strong form (prices reflect all publicly available information), then any predictable pattern in returns would immediately be arbitraged away by sophisticated investors. As traders respond to the anomaly, their buying and selling eliminate the pattern, driving prices toward a state where returns are once again unpredictable. The resulting market can therefore be described as a random walk — not because prices move randomly in a trivial sense, but because the systematic extraction of predictable returns has been competed away by rational profit-seeking.\n\nIn practice, empirical evidence presents a more nuanced picture. Short-horizon returns exhibit modest negative autocorrelation (mean-reversion in microstructure) and bid-ask bounce effects. Intermediate-horizon returns (3–12 months) display momentum — a positive autocorrelation inconsistent with a pure random walk — that has been documented across asset classes and geographies. Long-horizon returns (3–5 years) show evidence of mean-reversion, with previously poor-performing stocks outperforming over subsequent years. Whether these patterns represent genuine inefficiencies or risk-factor compensation remains the central debate of empirical asset pricing.\n\nThe Hurst exponent provides a quantitative framework for assessing how far a time series departs from random walk behavior. Estimated using rescaled range (R/S) analysis or detrended fluctuation analysis (DFA), an H value significantly above 0.5 indicates that a trend is more likely to continue than to reverse — a persistence property useful for trend-following strategies. Conversely, H significantly below 0.5 indicates mean-reversion, a characteristic exploited by statistical arbitrage. For equities, estimated Hurst exponents typically cluster near 0.5–0.6 at daily frequency, consistent with modest trend persistence and broadly consistent with near-random-walk behavior.",
  "example": "A quantitative analyst tests whether daily returns of the 10-year U.S. Treasury note exhibit random walk behavior using the variance ratio test of Lo and MacKinlay. The test compares the variance of two-day returns to twice the variance of one-day returns; under a random walk, this ratio should equal one. The analyst finds a variance ratio of 1.08 at a two-day horizon, marginally above one, and 1.14 at a five-day horizon, suggesting modest positive autocorrelation (trend persistence) at short horizons. This finding, statistically significant at the 5% level, motivates a short-term momentum strategy in Treasury futures. However, out-of-sample testing on a subsequent validation period shows the variance ratio falls to 1.02 — marginally statistically significant — illustrating how small departures from random walk behavior can be difficult to exploit reliably after transaction costs.",
  "formula": "P_t = P_{t-1} + ε_t, where ε_t ~ i.i.d.(0, σ²)",
  "formula_latex": null,
  "interactive_type": "chart",
  "calculator_id": null,
  "related_terms": [
    "alternative-data",
    "arbitrage",
    "autocorrelation",
    "brownian-motion",
    "efficient-market-hypothesis",
    "fundamental-law-of-active-management",
    "hurst-exponent",
    "out-of-sample-testing",
    "overfitting",
    "statistical-arbitrage",
    "stock",
    "treasury-note",
    "variance"
  ],
  "backlinks": [
    "arima-model",
    "autocorrelation",
    "autoregressive-model",
    "brownian-motion",
    "cointegration",
    "fundamental-law-of-active-management",
    "hurst-exponent",
    "latin-hypercube-sampling",
    "out-of-sample-testing",
    "serial-correlation"
  ],
  "cross_references": [
    "arbitrage",
    "autocorrelation",
    "brownian-motion",
    "efficient-market-hypothesis",
    "hurst-exponent",
    "out-of-sample-testing",
    "statistical-arbitrage",
    "stock",
    "treasury-note",
    "variance"
  ],
  "tags": [
    "level:basic",
    "cat:quantitative-finance"
  ],
  "asset_classes": [],
  "regulators": [],
  "see_also": [],
  "sources": [],
  "wordcount": 752,
  "checksum": "98f9682255a76f93",
  "version": "2026.05.03",
  "license": "CC-BY-4.0",
  "updated_at": "2026-09-07T02:15:24+00:00",
  "_links": {
    "self": "https://hedgefund.wiki/api/v1/terms/random-walk",
    "jsonld": "https://hedgefund.wiki/api/v1/terms/random-walk?format=jsonld",
    "markdown": "https://hedgefund.wiki/api/v1/terms/random-walk?format=md",
    "graph": "https://hedgefund.wiki/api/v1/graph/random-walk",
    "category": "https://hedgefund.wiki/api/v1/categories/quantitative-finance",
    "schema": "https://hedgefund.wiki/schema/term.schema.json",
    "html": "https://hedgefund.wiki/#/terms/random-walk"
  }
}