Introduction to the Efficient Market Hypothesis

The concept of market efficiency serves as the foundational bedrock of modern financial economics, dictating the theoretical frameworks through which asset pricing, portfolio management, and corporate finance are understood. At its core, the Efficient Market Hypothesis (EMH) asserts that financial asset prices fully and instantaneously reflect all available information. Consequently, it posits that investors cannot consistently achieve risk-adjusted excess returns, or “alpha,” because any new information regarding an asset’s fundamental value is immediately incorporated into its market price 1.
The intellectual lineage of this proposition traces back to the early twentieth century with the work of French mathematician Louis Bachelier, who in 1900 first modeled stochastic processes in financial markets, recognizing that past, present, and discounted future events are reflected in market prices while showing no apparent relation to price changes 1. This framework was subsequently advanced by Benoit Mandelbrot and Paul Samuelson in the 1960s, with Samuelson mathematically demonstrating that if a market is informationally efficient, price changes must exhibit unforecastable, random-walk behavior 1.
However, it was Eugene Fama’s seminal 1970 review, Efficient Capital Markets: A Review of Theory and Empirical Work, that formally categorized the hypothesis and solidified its place in neoclassical economics 1. Fama’s framework linked the concept of efficient markets to the random walk hypothesis, suggesting that because prices react only to unpredictable future news, subsequent price changes must be inherently unforecastable 1. Formal mathematical predictions regarding stock prices, assuming an absence of arbitrage and a constant stochastic discount factor, demonstrate that the logarithm of stock prices follows a martingale or random walk with a drift 1.
Despite its mathematical elegance, the Efficient Market Hypothesis has been the subject of relentless academic and practitioner debate for over half a century. The tension lies in the observable reality of financial markets: while markets are highly competitive and notoriously difficult to consistently beat, they are also prone to spectacular asset bubbles, crashes, and persistent anomalies that defy standard rational pricing models 5. Critics like Robert Shiller argue that markets are frequently irrational, pointing to the predictive power of metrics like the Cyclically Adjusted Price-to-Earnings (CAPE) ratio, which demonstrates that long-term investors generally earn superior returns when committing capital during periods of low valuations relative to earnings—a direct contradiction to the strict unpredictability assumed by the random walk hypothesis over long horizons 1.
The evolution of behavioral finance, the recognition of strict limits to arbitrage, and the modern proliferation of high-frequency algorithmic trading have continuously challenged the rigid assumptions of the EMH. The ensuing report provides an exhaustive examination of market efficiency, exploring its theoretical underpinnings, the insurmountable joint hypothesis problem, the paradoxes of information costs, structural limits to arbitrage, and the behavioral biases that generate enduring market anomalies.

The Three Forms of Market Efficiency

Eugene Fama’s 1970 classification organized market efficiency into three distinct forms, each predicated on the specific information set assumed to be fully reflected in asset prices 1. This taxonomy provided a structured methodology for empiricists to test the validity of the hypothesis by defining the boundaries of what constitutes “available information.”

Efficiency Form Information Set Reflected in Prices Implications for Investors Empirical Testing Methodology
Weak Form All historical price, volume, and return data. Technical analysis (charting, trend-following) cannot yield consistent excess returns. Autocorrelation tests, runs tests, and variance ratio tests on historical price series to detect predictable patterns.
Semi-Strong Form All publicly available information (earnings, macroeconomic data, news, financial statements). Fundamental analysis cannot yield consistent excess returns. Active management is generally futile. Event studies measuring price reaction speeds to earnings announcements, stock splits, or macroeconomic data releases.
Strong Form All public and private information, including proprietary insider knowledge. No participant, not even corporate insiders, can achieve excess returns. Prices are perfect reflections of intrinsic value. Analysis of insider trading profitability and performance of institutional fund managers with access to proprietary research.

The weak form of the hypothesis is largely supported by decades of empirical research. Studies overwhelmingly confirm that stock prices approximate a random walk, meaning past price movements offer no statistically reliable predictive power for future movements 1. For instance, exhaustive tests evaluating thousands of technical analysis trading rules frequently demonstrate that they fail to generate abnormally high returns once transaction costs and liquidity constraints are incorporated 11. Under weak-form efficiency, attempting to discern geometric patterns in stock charts is theoretically bankrupt 11.
The semi-strong form serves as the primary battleground for the active versus passive investment management debate. If the semi-strong form holds, all publicly distributed information is instantaneously digested by the competitive market, rendering active stock picking an exercise in futility 11. The evidence here is robust but highly nuanced. Early event studies, such as those analyzing stock splits, generally show that prices adjust with remarkable speed—often within milliseconds of a public data release, leaving no residual drift for investors to exploit 1.
The strong form of the hypothesis is universally regarded as an idealized theoretical benchmark rather than an empirical reality 2. Because trading on material non-public information is legally restricted in most jurisdictions, and because corporate insiders routinely demonstrate the ability to earn abnormal profits when they trade, the strong form is generally rejected 10. Fama himself regarded the strong form version primarily as a baseline against which the other forms of market efficiencies are to be judged, recognizing that perfectly costless information is an unrealistic economic assumption 2.

Empirical Support: The Case for Passive Investing

The most compelling practical evidence supporting the Efficient Market Hypothesis lies in the persistent failure of active portfolio managers to outperform passive market benchmarks. If markets were broadly inefficient, highly compensated, highly educated professionals utilizing advanced fundamental analysis should theoretically be able to identify mispriced securities and generate consistent, risk-adjusted alpha 11.
The Standard & Poor’s Indices Versus Active (SPIVA) scorecards provide devastating empirical evidence to the contrary. Over a two-decade observation period, the data reveals that outperforming the market is not only exceedingly rare but also lacks any meaningful persistence 13. What may appear as managerial skill in a single year frequently reverts to the mean, demonstrating that consistent outperformance is largely attributable to random chance 11.

U.S. Equity Fund Category Benchmark Index 1-Year Underperformance 5-Year Underperformance 10-Year Underperformance 15-Year Underperformance
All Large-Cap S\&P 500 78.78% 88.96% 85.59% 89.93%
All Mid-Cap S\&P MidCap 400 55.41% 72.32% 81.14% 84.49%
All Small-Cap S\&P SmallCap 600 40.65% 62.67% 75.95% 89.90%
Large-Cap Growth S\&P 500 Growth 95.51% 95.26% 91.67% 97.82%
Large-Cap Value S\&P 500 Value 41.30% 83.03% 88.45% 93.29%

The data illustrates a structural reality of highly liquid, highly competitive markets 15. In the U.S. large-cap space, where thousands of institutional analysts evaluate the same public data sets simultaneously, the market approaches near-perfect semi-strong efficiency. Any marginal informational advantage is rapidly arbitraged away. The most striking revelation in recent SPIVA data is that over a 15-year period, zero out of 22 tracked U.S. equity fund categories had a majority of active managers outperform their respective benchmarks 15.
This underperformance remains stark even when accounting for survivorship bias. A significant methodological strength of the SPIVA scorecard is its inclusion of funds that merged or liquidated during the observation period, capturing the true, unvarnished reality of active management failure 15. When an active manager underperforms, the fund is often quietly shuttered or absorbed, erasing its poor track record from standard databases. Adjusting for this bias reveals that the probability of an active fund beating the market shrinks drastically as the time horizon extends, dropping from roughly 35% over a single year to merely 5% over a 20-year stretch 16.
The fixed-income market provides a slightly different, yet ultimately parallel, narrative. In specific segments, such as general investment-grade funds, active managers occasionally exhibit short-term success, with only 30.4% underperforming at the one-year horizon in recent data 15. However, this outperformance is often driven by directional bets on duration or credit quality during favorable macroeconomic shifts. When measured over a 10-year horizon, the arithmetic of compounded management fees becomes punishing, and nearly every bond category shows underperformance rates approaching or exceeding 80% 15. The compounding drag of an active manager’s expense ratio guarantees that the average actively managed dollar must underperform the average passively managed dollar, driving the monumental structural shift toward index funds and passive investing 15.

The Joint Hypothesis Problem and the Factor Zoo

While the SPIVA data strongly supports the practical application of the EMH for investors, academic efforts to strictly prove or disprove market efficiency face an insurmountable epistemological barrier known as the Joint Hypothesis Problem. Formalized by Eugene Fama in 1970, this problem dictates that market efficiency per se is inherently untestable 1.
To determine whether a security is mispriced, an observer must compare its actual real-world return against its expected return. However, calculating the expected return requires a model of market equilibrium, otherwise known as an asset pricing model 18. Therefore, whenever an empiricist discovers an “abnormal” return or anomaly, it is logically impossible to definitively determine whether the market is genuinely inefficient, or if the asset pricing model used to calculate the expected return is simply inaccurate 1. The tests are inextricably joined at the hip: one cannot test efficiency without assuming a perfectly correct pricing model, and one cannot test a pricing model without assuming the market is efficient 19.

The Evolution of Asset Pricing Models

The joint hypothesis problem catalyzed a decades-long evolution in asset pricing models as researchers attempted to define the “correct” required rate of return. Initially, the Capital Asset Pricing Model (CAPM) served as the standard benchmark, asserting that the only risk requiring compensation was a security’s sensitivity to the overall market, measured by its beta 9. However, empirical anomalies quickly surfaced. Researchers discovered that small-capitalization stocks (the size effect) and stocks with high book-to-market ratios (value stocks) consistently generated returns far in excess of what their CAPM betas could explain 1.
Faced with the joint hypothesis problem, efficient market advocates argued that these excess returns were not evidence of mispricing, but rather evidence of unmeasured risk. This logic led to the Fama-French Three-Factor Model in 1993, which explicitly added size and value factors to the aggregate market factor, effectively absorbing the anomalies into a new definition of risk 1. When momentum was subsequently discovered by Jegadeesh and Titman, Mark Carhart proposed a Four-Factor Model in 1997 21. Eventually, Fama and French expanded their framework to a Five-Factor Model, incorporating corporate profitability and investment behavior as additional risk proxies 20.

The Factor Zoo and Publication Bias

The academic arms race to solve the joint hypothesis problem inadvertently created what financial economists now term the “factor zoo.” Over the last three decades, academics have identified over 300 different variables or characteristics that supposedly generate abnormal returns unaccounted for by standard equilibrium models 24.
This unchecked proliferation led to a severe critique by Campbell Harvey, Yan Liu, and Heqing Zhu (2016), who argued that the vast majority of these discovered anomalies are statistical mirages resulting from data snooping, p-hacking, and publication bias 27. Because researchers are incentivized to publish statistically significant findings, they run countless regressions on the same historical datasets until a seemingly profitable anomaly emerges by pure chance. Harvey, Liu, and Zhu adapted multiple-testing adjustments, such as the False Discovery Rate, to establish that a t-statistic threshold of at least 3.0 (rather than the traditional 1.96) is required to consider a factor statistically robust 27.
Furthermore, research by McLean and Pontiff (2016) demonstrated that even when anomalies are genuine, the act of publishing academic research paradoxically destroys their predictability 31. Once an anomaly is documented in a premier journal, hedge funds and institutional investors quickly deploy capital to exploit it. This influx of arbitrage capital compresses the mispricing, leading to an average post-publication decay in returns of nearly 58% 30. This phenomenon beautifully illustrates market efficiency in action: the market adapts to new information (academic publications) and dynamically neutralizes the inefficiency.

Bypassing the Problem: Expectations Based Returns (EBR)

Recent scholarship has attempted to sidestep the joint hypothesis problem entirely by utilizing measured analyst expectations of earnings growth rather than relying on assumed asset pricing models. By constructing a firm-level measure of Expectations Based Returns (EBRs) derived from analyst forecast errors and revisions, researchers can shut down cross-sectional differences in required returns 22.
This methodology reveals that a significant portion of the cross-sectional return spreads historically attributed to value, investment, size, and momentum factors are actually driven by predictable errors in expectations rather than compensation for systematic risk 22. When analysts extrapolate recent high growth too far into the future (overreaction), the subsequent disappointment when earnings fail to meet these lofty expectations drives prices down. Conversely, analysts tend to be overly pessimistic about distressed value stocks, leading to positive surprises 22. This evidence suggests that standard characteristics reflect behavioral expectation errors rather than risk, indicating that “correcting” for such factors using multi-factor models may inadvertently obscure genuine market inefficiencies 22.

Theoretical Limitations: The Grossman-Stiglitz Paradox

If the empirical challenges to the EMH are dominated by the joint hypothesis problem, the theoretical foundation of perfect efficiency is entirely undone by the Grossman-Stiglitz Paradox. Introduced by economists Sanford Grossman and Joseph Stiglitz in their landmark 1980 paper, On the Impossibility of Informationally Efficient Markets, the paradox exposes a fatal logical contradiction in the strong and semi-strong forms of the EMH 37.
The paradox centers entirely on the cost of information. Acquiring, processing, and acting upon fundamental data requires significant expenditures of capital, time, and computational resources 38. Institutional investors spend millions of dollars deploying proprietary algorithms, hiring analysts, and securing alternative datasets—such as satellite imagery of shipping containers—to find mispriced assets 38.
If financial markets were perfectly efficient, prices would instantaneously reflect all available information. If prices always perfectly reflected intrinsic value, active researchers and arbitrageurs would never be able to earn a return that compensates them for the costs they incurred to acquire that information 38.
Consequently, if there is no economic incentive to gather information, rational actors will cease to do so, choosing instead to become passive indexers. However, if everyone becomes a passive indexer and no one gathers information, market prices will cease to incorporate new fundamental data. The market then becomes entirely uninformed, creating massive mispricing and structural inefficiencies 38. This severe mispricing instantly revives the profit incentive for active researchers to re-enter the market, collect data, and trade 38.
The inescapable conclusion of the Grossman-Stiglitz paradox is that perfect informational efficiency is a mathematical impossibility 37. Instead, the market must maintain an “equilibrium degree of disequilibrium.” Prices must remain just inefficient enough to allow informed traders to earn excess returns that exactly compensate them for their information-gathering costs, but efficient enough to prevent uninformed traders from extracting easy profits 37. This dynamic tension explains the necessary, symbiotic coexistence of passive index funds and highly sophisticated active managers, the latter of whom serve as the costly but essential engine of price discovery 38.

Behavioral Finance and the Limits to Arbitrage

Neoclassical finance traditionally assumed that even if some investors act irrationally, rational arbitrageurs will aggressively trade against them, driving prices back to fundamental values and ensuring market efficiency. However, the sub-discipline of behavioral finance proves that arbitrage is neither costless nor risk-free, meaning mispricing can persist far longer than theoretical models suggest.

Noise Trader Risk

The model of Noise Trader Risk, formalized by J. Bradford De Long, Andrei Shleifer, Lawrence Summers, and Robert Waldmann (DSSW) in 1990, provides the theoretical framework for why prices can deviate significantly from intrinsic value 42. A “noise trader” is an investor whose decisions are driven by sentiment, technical trends, or behavioral biases rather than rigorous fundamental analysis 42.
According to the DSSW model, rational arbitrageurs face two distinct risks when attempting to correct a mispricing caused by noise traders:

  1. Fundamental Risk: The risk that the intrinsic value of the asset changes adversely while the arbitrageur holds the position. If an arbitrageur short-sells an overvalued technology stock, there is always a probability that the company will announce a revolutionary new product, justifying the high price and causing a devastating loss for the short-seller 43.
  2. Noise Trader Risk (Resale Price Risk): The risk that noise traders become even more irrational before they revert to the mean. Because arbitrageurs (such as hedge fund managers) have finite investment horizons and are subject to capital constraints, margin requirements, and periodic performance reviews, they cannot maintain losing positions indefinitely 42.

If a stock is overvalued by 20%, a rational arbitrageur might short it. But if noise trader sentiment unpredictably drives the stock to a 100% overvaluation, the arbitrageur may face severe margin calls and be forced to liquidate at a massive loss 43. Knowing this risk in advance, rational arbitrageurs will artificially limit the size of their positions against noise traders, preventing prices from fully returning to fundamental values. In some scenarios, rational arbitrageurs may even anticipate noise trader demand and buy into the overvalued asset to ride the bubble, actively destabilizing the market rather than correcting it 47.

Prospect Theory and the Disposition Effect

The irrationality of noise traders is largely mapped by Prospect Theory, developed by psychologists Daniel Kahneman and Amos Tversky in 1979 49. Prospect Theory challenged standard Expected Utility Theory by demonstrating that humans evaluate financial outcomes not based on final absolute wealth, but relative to a reference point—usually the purchase price of an asset 50.

Theoretical Model Perception of Gains and Losses Risk Preference Implications for Market Prices
Expected Utility Theory Evaluated symmetrically based on aggregate wealth changes. Consistent risk aversion across all scenarios. Investors act as perfectly rational profit maximizers, ensuring efficient, instantaneous price discovery.
Prospect Theory Evaluated asymmetrically. Losses induce psychological pain roughly twice as intense as the pleasure of equivalent gains (Loss Aversion). Risk-averse in the domain of gains; Risk-seeking in the domain of losses. Creates severe behavioral frictions, delayed price discovery, and momentum anomalies.

Prospect theory outlines a two-phase decision process: an editing phase where outcomes are heuristically ordered relative to a reference point, and an evaluation phase relying on an asymmetrical, S-shaped value function 50. The value function is concave for gains and convex for losses, meaning investors exhibit a four-fold pattern of risk: they are risk-averse when facing high-probability gains, but become aggressively risk-seeking when facing high-probability losses in a desperate attempt to break even 50.
A direct, observable market manifestation of Prospect Theory is the Disposition Effect, originally identified by Hersh Shefrin and Meir Statman in 1985 54. The disposition effect dictates that investors have a profound psychological tendency to prematurely sell their winning assets (to lock in the pride of a gain and avoid risk) and hold onto their losing assets for far too long (seeking risk to avoid the deep regret of realizing a loss) 54. Empirical studies suggest that these behavioral biases are measurable predictors of trading behavior, with specific demographics—such as female and mature traders—sometimes exhibiting stronger disposition effects depending on the trading environment 54.
This behavior severely distorts market efficiency. When positive fundamental news breaks about a company, its stock price should immediately jump to a new equilibrium level. However, as the price begins to rise, disposition-prone investors prematurely sell their shares to secure gains, creating artificial downward selling pressure that heavily slows the price’s ascent. Conversely, when bad news breaks, investors refuse to sell their losers, hoping the price will revert. This stubborn reluctance to sell slows the downward price adjustment 58. The disposition effect thereby creates systemic price underreaction to fundamental news, which serves as a primary driver of the persistent momentum anomaly observed in global equity markets 56.

Extreme Mispricing: Can the Market Add and Subtract?

While statistical anomalies like value or momentum are fiercely debated via the joint hypothesis problem, there are rare historical instances of mispricing so severe that they represent an outright failure of basic market arithmetic. The most famous case study is the equity carve-out of Palm by 3Com in March 2000, meticulously documented by Owen Lamont and Richard Thaler in their 2003 paper, Can the Market Add and Subtract? 60.
During the height of the Dot-Com bubble, 3Com, a profitable technology company, decided to spin off its subsidiary, Palm. 3Com retained a massive 95% stake in Palm and announced that within months, it would distribute its remaining Palm shares directly to 3Com shareholders. The explicit, legally binding distribution ratio dictated that the owner of one 3Com share would ultimately receive 1.5 shares of Palm 60.
By the fundamental laws of arithmetic and the Law of One Price, the market price of a 3Com share had to be worth at least 1.5 times the market price of a Palm share, plus the intrinsic value of 3Com’s legacy networking business 60.
However, on the day of Palm’s Initial Public Offering, irrational investor frenzy drove Palm’s stock price to unprecedented heights. Consequently, Palm’s implied market capitalization vastly exceeded 3Com’s total market capitalization 61. The implied “stub value” of 3Com’s legacy business was calculated at a staggering negative $22 billion 61. An investor could buy one share of 3Com, logically securing 1.5 shares of Palm plus 3Com’s actual business, for considerably less than the cost of simply buying 1.5 shares of Palm directly 60.
This glaring violation of market efficiency persisted for months, raising a critical question: why did rational arbitrageurs not instantly correct the mispricing by buying 3Com and shorting Palm? Lamont and Thaler proved that arbitrage failed due to severe institutional frictions—specifically, insurmountable short-sale constraints 62.
Because Palm shares were heavily demanded by retail noise traders and tightly held, the pool of available shares to borrow for shorting was virtually non-existent. When shares could occasionally be located, the borrowing costs (rebate rates) were astronomically high, completely wiping out the potential arbitrage profit 62. Options market data confirmed that the synthetic shorting costs implicitly embedded in put and call prices were so extreme that they formed a physical barrier to arbitrage 60. The 3Com/Palm anomaly proves that without the mechanism of frictionless short selling, over-optimistic investors can drive prices far away from fundamental reality, rendering the market locally inefficient 60. Similar pricing pressure dynamics are observed when a stock is added to the S\&P 500 index; index funds are forced to buy the stock regardless of fundamentals, leading to increased trading volume, heightened return volatility, and evidence of downward-sloping demand curves for equities, challenging the assumption that substitute assets are perfectly elastic 67.

Market Microstructure and Regional Dynamics: The Taiwan Stock Exchange

Market efficiency is not a monolithic global constant; it is heavily dictated by localized market microstructure, regulatory frameworks, and participant demographics. The Taiwan Stock Exchange (TWSE) provides a fascinating real-world laboratory to observe how participant composition affects efficiency and price discovery 69.
Historically, the TWSE has been characterized by exceptionally high retail investor participation. In many mature Western markets, institutional investors dominate daily trading volume, providing continuous liquidity and enforcing strict arbitrage bounds. In contrast, Taiwanese retail investors have traditionally accounted for a vast percentage of market turnover 69. Because retail investors are generally more susceptible to behavioral biases—such as the disposition effect, sentiment trading, and momentum chasing—the TWSE exhibits unique volatility profiles and instances of mean-reversion that deviate from random walk expectations 74.
In response to structural shifts and a desire to align with global capital markets, the TWSE has implemented significant regulatory and structural upgrades to enhance efficiency. For instance, to increase market accessibility and liquidity for younger, capital-constrained retail investors, the TWSE launched an intraday odd-lot trading system in October 2020 75. Previously, odd lots (trades of fewer than 1,000 shares) were restricted to after-hours trading, which fragmented liquidity and hampered price discovery for smaller participants 75. The integration of intraday odd-lot trading smoothed the liquidity curve and allowed broader participation in high-priced securities without forcing investors to assume excessive portfolio concentration risk. The exchange also launched the Taiwan Innovation Board (TIB) for capital-intensive startups and the ESG InfoHub to standardize sustainability reporting, further enhancing informational transparency 78.
Furthermore, the introduction of foreign institutional investors (FIIs) to the TWSE has fundamentally altered its efficiency dynamics. Research indicates that stocks with higher concentrations of foreign institutional ownership exhibit lower price crash risks and faster incorporation of global macroeconomic data into local equity prices 72. FIIs act as a stabilizing counterweight to local retail sentiment, effectively acting as the rational arbitrageurs modeled by DSSW, importing global pricing efficiency into the domestic Taiwanese market. Advanced machine learning models utilizing Radius Neighbors Regressors have demonstrated the ability to extract significant predictive power from the TAIEX by analyzing FII activity and volume-price relationships, yielding massive simulated returns during backtesting phases and proving that unique microstructural features can harbor exploitable, transient inefficiencies 70. Despite periods of intense global volatility—such as inflation concerns and changing U.S. tariff policies in 2024 and 2025—the TWSE has demonstrated robust resilience, frequently hitting record index highs supported by corporate profitability in the semiconductor and technology sectors 80.

The Future of Efficiency: Adaptive Markets and Artificial Intelligence

As markets transition deeper into the 21st century, static models of efficiency are increasingly inadequate to describe the rapidly shifting technological landscape. A synthesis of biological principles and financial economics is found in Andrew Lo’s Adaptive Market Hypothesis (AMH). First proposed in 2004, the AMH merges the core principles of the Efficient Market Hypothesis with the behavioral realities of human evolution 81.
Under the AMH, market participants are not perfectly rational calculating machines, nor are they hopelessly irrational noise traders. Instead, they are biologically bounded agents who rely on heuristics to make decisions 81. When the market environment is stable, these heuristics work well, and the market appears highly efficient. However, when the market environment undergoes a sudden shock or regime change, these old heuristics fail, leading to irrational behavior, severe volatility, and inefficiency 81. Investors learn from their mistakes, adapt to the new environment, and efficiency is gradually restored. Thus, market efficiency is not a constant state, but an evolutionary process driven by competition, adaptation, and survival 81.

The Algorithmic Market Hypothesis

The most profound evolutionary shock to modern market efficiency is the advent of Artificial Intelligence, Large Language Models (LLMs), and high-frequency Algorithmic Trading (AT). In a 2026 report by the CFA Institute, researcher Joseph Simonian introduced the “Algorithmic Market Hypothesis” (AMH), proposing that asset prices are increasingly reflecting the dominant algorithmic interpretations of information rather than human consensus 85.
Algorithmic trading significantly reduces latency, incorporating new data into prices in milliseconds. On the surface, this appears to drive markets closer to the strong-form efficiency ideal. Machine learning algorithms can digest corporate earnings transcripts, satellite imagery, and social media sentiment at a scale and speed that humans cannot match 85. Limit-order algorithms specifically have been shown to improve overall market welfare by narrowing bid-ask spreads, enhancing liquidity, and driving prices closer to fundamental values, though this surplus is frequently captured entirely by the algorithms at the expense of human traders 89.
However, the Algorithmic Market Hypothesis radically reframes the Grossman-Stiglitz paradox for the AI era. Today, raw data processing capacity is no longer the scarce, expensive resource; model validation is 85. Because AI systems can test millions of correlations at near-zero cost, they generate a massive volume of “false positives”—statistical relationships that look like profitable anomalies but lack any fundamental economic basis 85. The deployment of semantic factors—trading signals derived from the linguistic tone or narrative consistency of corporate management—can offer fleeting alpha, but the value decays rapidly as competing algorithms ingest the same parameters 85.
When multiple institutional algorithms detect the same false signal, they execute trades simultaneously. This creates “algorithmic monoculture,” where homogenous AI models engage in synchronized buying or selling. This can temporarily move prices, creating a self-fulfilling prophecy that validates the false signal in the short term, but ultimately leads to severe market fragility 85. The 2010 Flash Crash, where the Dow Jones plunged nearly 1,000 points in minutes due to algorithmic feedback loops and quote-stuffing, stands as a prime historical example of the dangers of latency-driven, homogeneous trading 86. In the modern era, faster markets are not necessarily more efficient markets; they are simply markets where inefficiencies occur at scales, complexities, and speeds that are imperceptible to human oversight 85.

Conclusion

The debate surrounding the Efficient Market Hypothesis is far from settled, serving instead as a dynamic spectrum upon which all financial activity is measured. The empirical evidence heavily supports the semi-strong form of the EMH for retail investors and traditional mutual fund managers, who consistently fail to outperform passive index strategies over long horizons after accounting for fees and transaction costs. The market is remarkably adept at rapidly processing public information.
However, recognizing that markets are “hard to beat” is fundamentally different from claiming that “prices are always perfectly right.” The insurmountable barrier of the joint hypothesis problem guarantees that academics will continue to debate whether anomalous returns are the result of genuine mispricing, unmeasured systematic risk factors, or simply data-snooping publication bias. Concurrently, the Grossman-Stiglitz paradox ensures that absolute informational efficiency is a logical impossibility; the friction and expense of information acquisition necessitate the existence of active, informed traders who exploit minor inefficiencies to keep prices anchored to reality.
Furthermore, structural frictions such as short-sale constraints and behavioral realities—like noise trader risk, loss aversion, and the disposition effect—prove that arbitrage is limited and human psychology intrinsically distorts price discovery. The blatant mispricing of the 3Com/Palm carve-out serves as a stark reminder that even in highly liquid markets, arithmetic can occasionally fail.
As the financial ecosystem evolves under the frameworks of the Adaptive Market Hypothesis and the Algorithmic Market Hypothesis, it becomes clear that efficiency is an ongoing evolutionary process. Artificial intelligence and machine learning are compressing the time it takes for information to be reflected in prices, but they are simultaneously introducing new structural vulnerabilities through algorithmic monoculture and hyper-correlated false signals. Ultimately, market efficiency should not be viewed as an absolute theoretical destination, but rather as a continuous, self-regulating equilibrium—a ceaseless tug-of-war between the relentless forces of rational arbitrage and the enduring constraints of human and algorithmic behavior.

Works cited

  1. Efficient-market hypothesis - Wikipedia, https://en.wikipedia.org/wiki/Efficient-market_hypothesis
  2. Market Efficiency Today∗ - Index of /, https://files.econ.cam.ac.uk/people-files/mhp1/PesaranMEHDec05.pdf
  3. (PDF) HYPOTHESIS (EMH) 1 Efficient Market Hypothesis (EMH) - ResearchGate, https://www.researchgate.net/publication/356541414_HYPOTHESIS_EMH_1_Efficient_Market_Hypothesis_EMH
  4. Efficient Market Hypothesis: What It Is & Why It Matters 2026 Quantt, https://www.quantt.co.uk/resources/efficient-market-hypothesis-explained
  5. View of Efficient Market Hypothesis: A Systematic Literature Review, https://journal.ilmudata.co.id/index.php/RIGGS/article/view/3549/2656
  6. Principles of the Efficient Market Hypothesis Grin, https://www.grin.com/document/1150190
  7. CAPE Fear: Why CAPE Naysayers Are Wrong Research Affiliates, https://www.researchaffiliates.com/insights/publications/articles/645-cape-fear-why-cape-naysayers-are-wrong
  8. Introducing P-CAPE: Improving Our Favorite Returns Estimator Elm Wealth, https://elmwealth.com/p-cape/
  9. The Efficient Markets Hypothesis Dethroned - UNI ScholarWorks, https://scholarworks.uni.edu/cgi/viewcontent.cgi?article=1043\&context=mtie
  10. Efficient Market Hypothesis (EMH) Definition + Examples - Wall Street Prep, https://www.wallstreetprep.com/knowledge/efficient-market-hypothesis-emh/
  11. Efficient Market Hypothesis: Strong, Semi-Strong, and Weak - Oblivious Investor, https://obliviousinvestor.com/efficient-market-hypothesis-strong-semi-strong-and-weak/
  12. Eugene Fama - Econlib, https://www.econlib.org/library/Enc/bios/Fama.html
  13. U.S. Persistence Scorecard Year-End 2025 - SPIVA S\&P Dow Jones Indices, https://www.spglobal.com/spdji/en/spiva/article/us-persistence-scorecard/
  14. SPIVA® U.S. Year-End 2025 - SPIVA S\&P Dow Jones Indices, https://www.spglobal.com/spdji/en/spiva/article/spiva-us/
  15. SPIVA Scorecard: How Active Funds Performed vs. Their Benchmarks, https://icfs.com/specialists-desk/spiva-scorecard-results
  16. FLIP FEATURE, https://campus.kennesaw.edu/colleges-departments/coles/centers/markets-economic-opportunity/docs/flip-feature-bray-spring-2026.pdf
  17. SPIVA S\&P Dow Jones Indices, https://www.spglobal.com/spdji/en/research-insights/spiva/
  18. Joint hypothesis problem - Wikipedia, https://en.wikipedia.org/wiki/Joint_hypothesis_problem
  19. Interview with Eugene Fama Federal Reserve Bank of Minneapolis, https://www.minneapolisfed.org/article/2007/interview-with-eugene-fama
  20. Pricing Without Mispricing - NBER, https://www.nber.org/system/files/working_papers/w29016/revisions/w29016.rev0.pdf
  21. Multifactor Models and Market Efficiency PDF - Scribd, https://www.scribd.com/document/880941155/Lecture-4-BM01FI-Investments-Canvas
  22. Finance without exotic risk - Andrei Shleifer, https://shleifer.scholars.harvard.edu/sites/g/files/omnuum10626/files/2025-10/FINEC104145%20%281%29.pdf
  23. Finance without exotic risk - Università Bocconi, https://didattica.unibocconi.it/mypage/upload/154156_20250715_042155_BGLS3JFEMAY27.PDF
  24. Episode 404: The Finance Paper that Changed Everything - Rational Reminder, https://rationalreminder.ca/podcast/404
  25. Anomaly or Possible Risk Factor? Simple-To-Use Tests∗, http://wp.lancs.ac.uk/fofi2022/files/2022/08/FoFI-2022-021-Benjamin-Holcblat.pdf
  26. Which stock return predictors reflect mispricing?, http://wp.lancs.ac.uk/fofi2024/files/2024/04/FoFI-2024-096-Jonas-Frey.pdf
  27. Market anomaly - Wikipedia, https://en.wikipedia.org/wiki/Market_anomaly
  28. Tree-Based Conditional Portfolio Sorts: The Relation Between Past and Future Stock Returns - cfr-cologne.de, https://www.cfr-cologne.de/download/kolloquium/2016/MoritzZimmermann.pdf
  29. The History of the Cross Section of Stock Returns - Rodney L. White Center for Financial Research, https://rodneywhitecenter.wharton.upenn.edu/wp-content/uploads/2014/04/12-16.Roberts.pdf
  30. Does Peer-Reviewed Research Help Predict Stock Returns?, https://www.cuhk.edu.hk/fin/event/CUHK-RAPS-RCFS-Conference2024/Papers/CUHK-RAPS-166.pdf
  31. POST-MERGER RETURNS IN FRONTIER MARKETS, OR HOW WE LEARNED TO STOP WORRYING AND LOVE THE ACQUIRERS, https://journals.vilniustech.lt/index.php/JBEM/article/download/1584/1243/3228
  32. Liquidity Characteristics of Market Anomalies and Institutional Trading - UConn Finance Department, https://finance.business.uconn.edu/wp-content/uploads/sites/723/2020/09/liqexp_202009.pdf
  33. Does it Pay to Follow Anomalies Research? Machine Learning Approach with International Evidence, http://wp.lancs.ac.uk/fofi2020/files/2020/04/FoFI-2020-056-Martin-Hronec.pdf
  34. From theory to practice: Polish equity risk factors and their implementation costs - Biblioteka Nauki, https://bibliotekanauki.pl/articles/63760112.pdf
  35. Anomalies Never Disappeared: The Case of Stubborn Retail Investors*, http://wp.lancs.ac.uk/fofi2024/files/2024/04/FoFI-2024-188-Xi-Dong.pdf
  36. Does peer-reviewed research help predict stock returns? - EconStor, https://www.econstor.eu/bitstream/10419/294837/1/1888120150.pdf
  37. What Does the Grossman Stiglitz Paradox Tell Us About the Efficiency of the Markets?, https://optionstradingiq.com/grossman-stiglitz-paradox/
  38. The Grossman-Stiglitz Paradox - Super Business Manager, https://www.superbusinessmanager.com/the-grossman-stiglitz-paradox/
  39. The Grossman-Stiglitz Paradox - LARS P. SYLL - WordPress.com, https://larspsyll.wordpress.com/2025/10/09/the-grossman-stiglitz-paradox-5/
  40. GROSSman Stiglitz paradox - Medium, https://mysticwealth.medium.com/grossman-stiglitz-paradox-39b9685733d8
  41. On the Impossibility of Informationally Efficient Markets - Semantic Scholar, https://www.semanticscholar.org/paper/On-the-Impossibility-of-Informationally-Efficient-Grossman-Grossman/fa8882edbf4d13ce575c818e6b423952177882ac
  42. Noise-Trading, Costly Arbitrage, and Asset Prices: Evidence from Closed End Funds - University of Warwick, https://warwick.ac.uk/fac/soc/wbs/subjects/finance/research/wpaperseries/wp02-13.pdf
  43. Noise Trader Risk in Financial Markets - IDEAS/RePEc, https://ideas.repec.org/p/wop/calbec/_124.html
  44. Noise Trader Risk in Financial Markets - McMaster University, https://ms.mcmaster.ca/~grasselli/DeLongShleiferSummersWaldmann90.pdf
  45. The Noise Trader Approach to Finance Andrei Shleifer; Lawrence H. Summers The Journal of Economic Perspectives, Vol. 4, No. 2. (, https://www.rose-hulman.edu/~bremmer/EMGT/paper/shleifer\&summers.pdf
  46. (PDF) WHO ARE THE NOISE TRADERS? - ResearchGate, https://www.researchgate.net/publication/46538959_WHO_ARE_THE_NOISE_TRADERS
  47. Noise Trader Risk and Wealth Effect: A Theoretical Framework - MDPI, https://www.mdpi.com/2227-7390/10/20/3873
  48. Does Noise Trading Affect Securities Market Efficiency? - Columbia University, http://www.columbia.edu/~pt2238/papers/Tetlock_Noise_and_Efficiency_09_06.pdf
  49. Prospect theory Business and Management Research Starters - EBSCO, https://www.ebsco.com/research-starters/business-and-management/prospect-theory
  50. Prospect theory - Wikipedia, https://en.wikipedia.org/wiki/Prospect_theory
  51. Prospect Theory - The Decision Lab, https://thedecisionlab.com/reference-guide/economics/prospect-theory
  52. How we view risk and uncertainty – prospect theory Sveriges Riksbank, https://www.riksbank.se/en-gb/press-and-published/publications/economic-commentaries/monetary-policy-and-behavioural-economics/some-key-features-of-behavioural-economics-research/how-we-view-risk-and-uncertainty–prospect-theory/
  53. (PDF) The Prospect Theory and The Stock Market - ResearchGate, https://www.researchgate.net/publication/370696299_The_Prospect_Theory_and_The_Stock_Market
  54. Impact of disposition effect on financial decisions among women investors, https://www.abacademies.org/articles/impact-of-disposition-effect-on-financial-decisions-among-women-investors-12756.html
  55. The Disposition Effect, Individual Differences, Stability, and Learning: An Experimental Investigation - University of Warwick, https://warwick.ac.uk/fac/soc/wbs/subjects/finance/events/recentevents/pastevents/behaviouralfinance1day/weber_06_version_1.pdf
  56. Disposition effect - Wikipedia, https://en.wikipedia.org/wiki/Disposition_effect
  57. Are Short Sellers Subject to the Disposition Effect? Evidence from Short Sale Disclosures in the United Kingdom - Erasmus University Thesis Repository, https://thesis.eur.nl/pub/66970/Final_Master_Thesis_Riemer_ter_Burg.pdf
  58. The Disposition to Sell Winners too Early and Ride Losers too Long - ResearchGate, https://www.researchgate.net/publication/313248569_The_Disposition_to_Sell_Winners_too_Early_and_Ride_Losers_too_Long
  59. Disposition Effect and its outcome on endogenous price fluctuations, https://mpra.ub.uni-muenchen.de/113904/1/MPRA_paper_113904.pdf
  60. Can the Market Add and Subtract? Mispricing in Tech Stock Carve-Outs - ResearchGate, https://www.researchgate.net/publication/278064689_Can_the_Market_Add_and_Subtract_Mispricing_in_Tech_Stock_Carve-Outs
  61. Speculative Trading and Stock Prices: Evidence from Chinese A-B Share Premia *, http://aeconf.com/articles/nov2009/aef100201.pdf
  62. Can the Market Add and Subtract? Mispricing in Tech Stock Carve-outs - IDEAS/RePEc, https://ideas.repec.org/a/ucp/jpolec/v111y2003i2p227-268.html
  63. Can the Market Add and Subtract? Mispricing in Tech Stock Carve‐outs Journal of Political Economy: Vol 111, No 2, https://www.journals.uchicago.edu/doi/abs/10.1086/367683?mobileUi=0&
  64. Can the Market Add and Subtract? Mispricing in Tech Stock Carve-Outs - ResearchGate, https://www.researchgate.net/publication/228316767_Can_the_Market_Add_and_Subtract_Mispricing_in_Tech_Stock_Carve-Outs
  65. Can the Market Add and Subtract? Mispricing in Tech Stock Carve‐outs Journal of Political Economy: Vol 111, No 2, https://www.journals.uchicago.edu/doi/10.1086/367683
  66. Short Sale Constraints and Overpricing - NBER, https://www.nber.org/reporter/spring05/short-sale-constraints-and-overpricing
  67. Changes in trading volume and return volatility associated with S\&P 500 index additions and deletions - ResearchGate, https://www.researchgate.net/publication/235282608_Changes_in_trading_volume_and_return_volatility_associated_with_SP_500_index_additions_and_deletions
  68. Exchange-Traded Funds, Market Structure and the Flash Crash - ResearchGate, https://www.researchgate.net/publication/228261387_Exchange-Traded_Funds_Market_Structure_and_the_Flash_Crash
  69. January 31, 2024 Annual report is available at Taiwan Stock Exchange Market Observation Post System： https, https://www.yuanta.com/Files/b4eaaecf-9e63-4e7a-8360-7a5514f1b983/YFH2023_E.pdf
  70. Stock Market Forecasting in Taiwan: A Radius Neighbors Regressor Approach - MDPI, https://www.mdpi.com/2504-2289/10/4/109
  71. 2024 Investment Climate Statements: Taiwan - State Department, https://2021-2025.state.gov/reports/2024-investment-climate-statements/taiwan/
  72. Do Foreign Investors Curb Stock Price Crash Risk? Evidence from Ownership Concentration in Taiwan - Scirp.org., https://www.scirp.org/journal/paperinformation?paperid=124740
  73. Investors with Trading Accounts by Age - Taiwan Stock Exchange Corporation, https://www.twse.com.tw/en/trading/statistics/list02-264.html
  74. 以乖離率為基礎的交易策略之績效：套利有限性之啟發__臺灣人文及, https://tci.ncl.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dnclresource\&s=id=%22A17028096%22.\&searchmode=basic\&tcihsspage=tcisearch_opt1_search
  75. Trading Mechanism Introduction - Taiwan Stock Exchange Corporation, https://www.twse.com.tw/en/products/system/trading.html
  76. 2025 Guide to Investing in Taiwan - Taiwan Stock Exchange Corporation, https://www.twse.com.tw/en/about/company/guide.html
  77. Intraday odd lot trading - Taipei Exchange, https://www.tpex.org.tw/en-us/mainboard/trading/rules/odd-lot.html
  78. Taiwan Stock Exchange Corporation: Home, https://www.twse.com.tw/en/
  79. Trading Value of Foreign & Other Investors - Taiwan Stock Exchange Corporation, https://www.twse.com.tw/en/trading/foreign/bfi82u.html
  80. Central Bank of the Republic of China (Taiwan), https://www.cbc.gov.tw/dl-221034-7e280a7510ec4bbf899d2a177760326f.html
  81. Adaptive Market Hypothesis (AMH) - Scribd, https://www.scribd.com/document/1035343437/Adaptive-Market-Hypothesis
  82. Adaptive Market Hypothesis: Reshape Your Marketing - Fractional, https://business901.com/blog1/adaptive-market-hypothesis-reshape-your-marketing/
  83. Machines and Markets : Assessing the Impact of Algorithmic Trading on Financial Market Efficiency - IDEAS/RePEc, https://ideas.repec.org/p/wrk/wrkesp/11.html
  84. Adaptive Markets Hypothesis: A Systematic Review of Its Testing and Application in Financial Market Efficiensy Studies - ResearchGate, https://www.researchgate.net/publication/394590569_Adaptive_Markets_Hypothesis_A_Systematic_Review_of_Its_Testing_and_Application_in_Financial_Market_Efficiensy_Studies
  85. The Algorithmic Market Hypothesis RPC - CFA Institute Research and Policy Center, https://rpc.cfainstitute.org/research/reports/2026/algorithmic-market-hypothesis
  86. The Role of Algorithmic Trading in Shaping Futures Market Efficiency: - IGI Global, https://www.igi-global.com/viewtitle.aspx?TitleId=367901\&isxn=9798369363867
  87. AI’s Dual Role in Financial Markets: Efficiency or Volatility? - Van Hessen, https://www.vanhessen.com/first-dry/AIs-Dual-Role-in-Financial-Markets-Efficiency-or-Volatility-42-19434
  88. Artificial intelligence in financial market prediction: advancements in machine learning for stock price forecasting - Frontiers, https://www.frontiersin.org/journals/artificial-intelligence/articles/10.3389/frai.2025.1696423/full
  89. Algorithmic Trading, Price Efficiency and Welfare: An Experimental Approach - EconStor, https://www.econstor.eu/bitstream/10419/302411/1/vfs-2024-pid-107790.pdf
  90. Algorithmic Trading Efficiency and its Impact on Market-Quality - ResearchGate, https://www.researchgate.net/publication/355469189_Algorithmic_Trading_Efficiency_and_its_Impact_on_Market-Quality