03 · Market Anomalies: Evidence That Markets Are Less Efficient
If markets truly worked as the efficient market hypothesis describes, there would be no "regular" windows of excess return. Yet over decades, academia has dug up a batch of persistently observed phenomena — calendar effects, momentum, the small-firm effect, the index effect… Some later decayed; some survived in a new shape. But together they prove one thing: prices contain more than information — they contain human nature.
I. What Is a Market Anomaly
1.1 Definition
A market anomaly (Anomaly) is an observable regularity that "shouldn't exist" under the EMH framework — a statistically significant relationship between some characteristic (time, size, valuation, event) and future returns.
1.2 Why Anomalies Matter
| Position | Interpretation of Anomalies |
|---|---|
| EMH camp | Statistical artifacts: data mining, survivorship bias, underestimated risk compensation |
| Behavioral camp | Systematic bias: anomalies are human irrationality priced into markets |
| Middle ground | Both: some anomalies are risk premia, others echoes of human error |
The controversy itself is valuable: whether an anomaly is "risk compensation" or an "echo of human nature," it means systematic forces operate in price formation — which is exactly why behavioral finance exists.
II. Calendar Effects
2.1 The January Effect (January Effect)
- [Phenomenon] Small-cap stock returns in January have historically been statistically higher than other months.
- [Explanatory hypotheses] Tax-loss harvesting (selling losers at year-end for tax purposes and rebuying in January), year-end bonuses entering the market, investors reallocating at the New Year.
- [Controversy] This was once behavioral finance's most celebrated piece of evidence, but later research suggests its significance weakened as it became widely known — anomalies that are "discovered" and published tend to be arbitraged flat. The strength and even existence of the January effect differ across markets (US, A-shares, H-shares).
2.2 The Monday Effect (Monday Effect)
- [Phenomenon] Historically, US equities showed significantly negative average Monday returns (the weekend effect), with Thursday/Friday relatively strong.
- [Explanatory hypotheses] Bad news tends to be released after Friday's close; traders' weekend gloom vents at Monday's open; "decision fatigue" accumulates midweek.
- [Controversy] The effect is unstable across decades of samples and has clearly weakened since the 2000s; "red/black Monday" statistics in A-shares also shift with regulatory rhythm and event drivers. The biggest caution with calendar effects: they are the easiest kind of "pattern" to overfit.
2.3 The Pre-Holiday Effect (Pre-holiday Effect)
- [Phenomenon] In the final trading days before long holidays, many markets show a statistical tendency to rise.
- [Explanatory hypotheses] Pre-holiday optimism, institutions positioning early, expectations of ample liquidity; A-shares add the folk narrative of a "red-envelope rally."
- [Controversy] Findings depend heavily on how "holidays" are selected; studies both supporting and refuting the significance of "Spring Festival effects" and "National Day effects" coexist in A-share history. The folk consensus that "markets always rise before holidays" is itself soil for self-fulfilling expectations — and once an expectation is widely believed, exploitation isn't far behind.
2.4 Practical Notes on Calendar Effects
Calendar effect ≠ risk-free calendar signal
Facts: statistical significance ≠ works every time;
published patterns get arbitraged away; small-sample overfitting risk is extremely highIII. Momentum and Reversal Effects
3.1 The Momentum Effect (Momentum Effect)
- [Phenomenon] Assets that performed well over the past 3–12 months (winners) statistically tend to keep outperforming poor performers (losers) over the following period.
- [Discovery] A textbook behavioral-finance anomaly, with replicated evidence across many markets and asset classes (equities, commodities, FX).
- [Explanatory hypotheses] Investor underreaction (slow digestion of good news; herding extending trends) and the gradual diffusion of information.
- [Controversy] Momentum returns can be surrendered all at once during "momentum crashes" (e.g., when markets turn abruptly and crowded momentum positions flee together); turnover costs and shorting constraints also erode returns.
3.2 The Reversal Effect (Reversal Effect)
- [Phenomenon] Over 3–5 year horizons, past-worst portfolios (losers) tend to outperform past winners — short-term momentum, long-term reversal.
- [Explanatory hypotheses] Overreaction and mean reversion: investors over-extrapolate long trends (the institutional version of representativeness), pushing prices past fair value before they revert.
- [Controversy] The time boundary between "short-term momentum, long-term reversal" differs across markets and eras; in A-share history, small-cap long-term reversal and short-term momentum coexist with highly unstable parameters.
3.3 Implications for Traders
| Time Scale | Statistical Tendency | Corresponding Human Nature |
|---|---|---|
| Weeks–months | Momentum (strong stays strong) | Underreaction, herd chasing |
| Months–quarters | Momentum continues | Confirmation bias, extrapolation |
| Years–decades | Reversal (mean reversion) | Correction after overreaction |
💡 Core Insight: Different Projections of the Same Biases
Core insight: momentum and reversal don't contradict each other — they are the same biases projected onto different time scales. Short term, people underreact and follow the crowd; long term, they overreact and overshoot — so prices ride biases in the short run and get dragged back by them in the long run.
IV. Size and Value Effects
4.1 The Small-Firm Effect (Small Firm Effect)
- [Phenomenon] Historically, small-cap companies' long-run average returns exceeded large caps'.
- [Explanatory hypotheses] Risk premium (small firms are more fragile, less liquid, more likely to fail), inefficient pricing due to thin analyst coverage, retail preference for high volatility (see "lottery preference" below).
- [Controversy] Small-cap samples suffer severe survivorship bias and delisting-handling problems; performance swings violently across periods, with large caps winning many years; small-cap index performance over recent decades has contradicted the historical research findings — the "small-firm premium" has at times been considered vanished or inverted in recent years.
4.2 The Value Effect (Value Effect)
- [Phenomenon] Low-valuation (low P/E, low P/B) portfolios have statistically outperformed high-valuation portfolios over the long run.
- [Explanatory hypotheses] Behavioral view: investors are overly optimistic on growth stocks and overly pessimistic on value stocks (representativeness). Risk view: value stocks bear more risk; the premium is compensation.
- [Controversy] This is the fiercest battleground between behavioral finance and the EMH camp; the value premium held across several historical stretches of US equities but also saw prolonged "growth beats value" phases (around the dot-com bubble). "Cheap stocks win long-term" is a multi-decade statistical tendency, not next year's trading plan.
4.3 On "Factor Investing"
Small cap, value, momentum, quality and the like are collectively called "factors." Practical implications:
- Factors are long-run statistical tendencies, not short-term signals — holding periods measured in years.
- Factor returns come from "bearing specific risks + exploiting collective biases" — neither is free.
- Factors get crowded: when too many people use the same factor, buying pressure itself erases the excess return.
V. Event-Driven Anomalies
5.1 The Index Effect (Index Effect)
- [Phenomenon] Stocks added to major indices (e.g., CSI 300 or S&P 500 inclusion) tend to rise in the short term.
- [Mechanism] Passive index funds must buy new constituents (passive demand), institutions position ahead of inclusion (anticipatory trading), and liquidity premia shift afterward.
- [Current state] As passive investing scales up, the index effect appears statistically stronger in overseas markets; A-shares likewise saw capital flows around inclusion dates. It is a game around a known event, not a secret pattern.
5.2 The IPO Underpricing Puzzle / Speculation
- [Phenomenon] New listings commonly pop on day one (offer prices set below first-day market prices), with intense retail enthusiasm.
- [Explanatory hypotheses] Information asymmetry (issuers/underwriters conceding margin), market sentiment (scarce early supply + speculative money flooding in), winner's curse theory (rational buyers demand a discount for their informational disadvantage).
- [China-specific flavor] The historical "IPO never loses," "limit-up from day one," and IPO speculation frenzies were directly tied to issuance regimes, price-limit rules, and retail participation structure; once institutions change, the phenomenon changes with them — anomalies are shadows of institutions.
5.3 Lottery-Preference Stocks (Lottery Preference)
- [Phenomenon] Cheap, high-volatility, small-cap stocks with recent explosive swings — "lottery-like stocks" — statistically earn lower long-run returns.
- [Explanatory hypotheses] A direct corollary of prospect theory: people overweight tiny-probability/huge-payoff bets (lottery mentality) and pay a premium for the chance to swing — the frenzy in hot penny stocks is precisely the market price of the reflection effect and the certainty effect.
- [Current state] Penny-stock manias and "demon stocks" persist across many markets and eras — among the anomalies behavioral finance explains best.
VI. Anomalies Specific to Chinese Markets
6.1 The History of Shell Value
- [Phenomenon] In the era of approval-based listings, hard IPOs, and rare delistings, A-share listed companies (especially small-cap shells) carried a built-in "reverse-merger option"; shell value significantly inflated small-cap valuations, and "shell speculation" became a special pricing factor.
- [Evolution] With registration-based IPO reform and normalized delistings, shell value shrank historically — institutional change directly killed an anomaly, a superb case study of the anomaly life cycle.
6.2 High Turnover and Retail Dominance
- [Phenomenon] A-share turnover has historically been far higher than mature markets, with retail dominating trading volume and frequent theme speculation and chase-and-dump behavior.
- [Explanatory hypotheses] In retail-dominated markets, the disposition effect, herding, and loss aversion are amplified into price volatility; high turnover itself correlates mostly negatively with long-run returns — overtrading is a leak in returns.
- [Current state] As institutionalization and quant trading grow, some behavioral signatures have been rewritten (e.g., short-term games shifting from manual chasing to programmatic following), but in retail-heavy markets behavioral patterns typically play out more extravagantly.
6.3 What "Chinese Characteristics" Really Mean
📖 The Essence of "Chinese Characteristics"
The common thread of A-share anomalies is: institutional constraints + investor structure + cultural narratives. Price-limit rules create "consecutive limit-up" games, T+1 and stamp tax reshape trading costs, heavy retail participation amplifies emotional resonance, and "mystical rallies" supply narrative fuel. Studying Chinese market anomalies requires studying institutions simultaneously — otherwise you'll mistake "the shadow of institutions" for "a law of human nature."
VII. Overreaction and Underreaction: A Unified Explanation
Momentum, reversal, value, and event anomalies look wildly varied, but behavioral finance gives them a shared psychological foundation — underreaction and overreaction:
7.1 How Two Reaction Modes Create Price Patterns
| Mechanism | Human Nature Behind It | Resulting Phenomenon |
|---|---|---|
| Underreaction | Confirmation bias, anchoring (slow to process good news) | Prices keep drifting after good news (post-earnings announcement drift) |
| Underreaction | Herding (only chasing once prices rise) | Trend continuation (momentum) |
| Overreaction | Representativeness (extrapolating recent action into the long term) | Overshoot then mean reversion (reversal) |
| Overreaction | Lottery preference (chasing volatility) | High-flying stocks lag long term (value effect) |
Key inference: anomalies aren't isolated statistical coincidences but different projections of the same human nature across time scales — short-term underreaction creates momentum; long-term overreaction creates reversal; both coexist in the same brain.
7.2 What Is PEAD?
- [Phenomenon] After companies announce better-than-expected earnings, prices don't fully adjust immediately but keep drifting in the favorable direction over the following months — good news isn't priced in at once; it's digested slowly.
- [Explanation] Underreaction: investors digest announcements gradually; early movers are few while herds follow step by step.
- [Implication] One of the cleanest proofs that "markets aren't instantly efficient": if prices reflected all information instantly, drift wouldn't exist.
7.3 A Unified Lesson for Ordinary Traders
Short-term momentum (follow trends) + long-term reversal (don't chase extended runs)
= two modes of the same brainOnce you understand these two machines, the next sections (practical use and boundaries of anomalies) make clear why anomalies are both opportunity and trap — because what drives them is you.
⚠️ Counterintuitive: What Drives Them Is You
Anomalies are both opportunity and trap because what drives them is you. Short-term momentum (follow trends) + long-term reversal (don't chase extended runs) = two modes of the same brain — when you think "it's been rising so I'll keep chasing," you've become part of the momentum effect; when you think "it's fallen three years, it should bounce," you've become part of the reversal effect. Only by understanding this can you stand outside your biases rather than inside them.
VIII. Practical Use and Boundaries of Anomalies
8.1 What You Can Do
| Use | Description | Caution |
|---|---|---|
| Build factors | Weight portfolios using momentum/value/low-vol logic | Holding period in years, not a short-term signal |
| Read market states | Use sentiment-type anomalies (e.g., IPO premiums, penny-stock manias) as a market thermometer | A thermometer is not a trade order |
| Reflexivity watch | When an anomaly gets widely reported as "reliably profitable," beware crowding | Published anomalies get arbitraged flat |
8.2 What You Must Not Do
❌ Treat "statistically significant" as "will rise tomorrow"
❌ Extrapolate decades-old samples into the next few years
❌ Ignore institutional and investor-structure shifts behind anomalies
❌ Attribute all anomaly returns to "market error" — some are just risk compensation8.3 Anomalies Decay
| Failure Mode | Example |
|---|---|
| Arbitraged away | After academic publication, an anomaly's returns often systematically decline |
| Institutional change | Registration reform killed shell value; price-limit reforms changed limit-up games |
| Structural change | Rising quant/institutional share rewrites retail behavior patterns |
| Overcrowding | Too many users of one factor: buying itself lifts prices and spends the premium |
💡 One Sentence: Anomalies Are Not an ATM
One sentence: anomalies are evidence that "markets are made of people," not evidence that "markets contain an ATM." The greatest value of understanding anomalies is knowing what you're dealing with when the market is at its craziest.
💀 Iron Law: Anomalies Are Evidence Markets Are Made of People, Not That They Contain an ATM
Anomalies are evidence that "markets are made of people," not evidence that "markets contain an ATM." Their greatest value is knowing what you're dealing with when the market is at its craziest — not using anomalies for "stable arbitrage." Published anomalies get arbitraged flat, institutional change kills them, and overcrowding spends their returns.
⚠️ Risk Warning
This content is for study and research only and does not constitute investment advice. Every anomaly here is a statistical research finding, subject to extensive academic dispute, and may decay or invert as institutions, market structure, and participants change; statistical significance ≠ reliable arbitrage, and historical patterns ≠ promises about the future. Any factor strategy built on anomalies should be rigorously backtested and validated with small capital, combined with the position sizing and risk management of Chapter 07.