DEEP VALUE
Chapter 1: The Paradox of Dumb Money — In Full Detail
INTRODUCTION: DUMB MONEY VS SMART MONEY
Carlisle begins with a provocative claim:
The so-called "dumb money" often beats the ―smart money‖ — and not by chance, but
systematically.
―Dumb money‖ = individual investors, quantitative value models, or activist
contrarians
―Smart money‖ = institutional investors, hedge funds, mutual fund managers,
analysts
Why is this paradoxical?
Institutions have:
Professional analysts
Access to management
Advanced research tools
Scale and capital
Yet they consistently underperform simple quantitative deep value strategies.
SECTION 1: THE ILLUSION OF SMARTNESS
Carlisle critiques professional money management, showing:
Most mutual funds underperform their benchmarks after fees.
A 2010 Morningstar study found that the best predictor of future returns wasn’t
manager experience, fund size, or Morningstar rating — it was low fees (i.e., passive,
not active investing).
� Institutional investors are often closet indexers, meaning they slightly tweak benchmark
weights to appear active while avoiding career risk.
�Result: Their risk-adjusted alpha is near-zero or negative.
�SECTION 2: SYSTEMATIC UNDERPERFORMANCE OF ―SMART‖
STRATEGIES
Academic Support:
Carlisle cites:
Fama and French (1992) – Value stocks (low P/B) beat growth stocks
Shleifer and Vishny (1997) – Behavioral biases persist because arbitrage is risky
Barber and Odean (2000s) – Individual investors underperform, but so do
institutions
Key Insight: Institutions behave in predictable, rational ways that aren’t profit-maximizing
— they avoid controversial or ugly investments due to short-term pressure.
Institutions can’t afford to look wrong, even if they’re ultimately right.
Traditional CAPM vs. Fama-French 3-Factor Model
�1. Capital Asset Pricing Model (CAPM)
CAPM says a stock’s return is explained by beta — its sensitivity to market movements:
Ri−Rf=βi(Rm−Rf)R_i - R_f = \beta_i (R_m - R_f)Ri−Rf=βi(Rm−Rf)
Where:
RiR_iRi = expected return of the stock
RfR_fRf = risk-free rate
RmR_mRm = expected market return
βi\beta_iβi = sensitivity of stock iii to the market
�2. But Reality: Beta Alone Fails
In practice:
High-beta stocks don’t consistently outperform
Low-beta stocks don’t underperform, especially when small or cheap
�Empirical flaw: CAPM assumes only systematic risk (beta) matters. But studies showed
other anomalies — size and value effects — which beta couldn’t explain.
�3. Fama & French (1992) – The Breakthrough
�What They Did:
They tested U.S. stock returns from 1963–1990, across 2 dimensions:
1. Size – Market Cap (Small vs Big)
2. Value – Book-to-Market (High B/M = value, Low B/M = growth)
�They formed portfolios sorted by both size and value (a 5×5 grid) and examined average
returns.
�What They Found:
Beta was not a good predictor of returns.
Instead:
o Small stocks outperformed large stocks.
o High book-to-market (value) stocks outperformed low book-to-market
(growth) stocks.
�Many low-beta small-cap value stocks had high returns — contradicting CAPM.
� The Fama-French 3-Factor Model
To fix CAPM’s failures, they introduced:
Where:
Ri−RfR_i - R_fRi−Rf = excess return of stock
Rm−RfR_m - R_fRm−Rf = market risk premium
SMBSMBSMB = Small Minus Big → return of small-cap stocks minus big-cap
stocks (captures size premium)
HMLHMLHML = High Minus Low → return of high book-to-market stocks minus
low B/M (captures value premium)
β2,β3\beta_2, \beta_3β2,β3 = sensitivities to these factors
α\alphaα = any abnormal (unexplained) return
� Proof That Value & Size Explain Returns Better
�Empirical Evidence from the Paper:
Factor Average Excess Return T-statistic
Market (Rm – Rf) Yes High
Size (SMB) ~3.6% annually High
Value (HML) ~4.8% annually High
Beta (alone) Weak Insignificant
Beta explained little of the cross-section of returns.
Size and value explained much more.
The 3-factor model had higher explanatory power (R²) and better predictions than
CAPM.
�Later Confirmations:
Carhart (1997) added momentum as a 4th factor — further weakening CAPM.
French-Fama Data Library still updates monthly, showing persistent value and size
premia over decades.
90+ countries tested → value and size effects exist globally.
� Summary: Why Value and Size Beat Beta
CAPM Flaws Fama-French Fixes
Assumes only market risk matters Adds size (SMB) and value (HML)
Empirically weak predictions Empirically stronger fit (higher R²)
Cannot explain anomalies like low-beta high returns Explains small-cap and value premium
One-size-fits-all Multi-factor customization
�SECTION 3: HUMAN BEHAVIOR IS PREDICTABLY IRRATIONAL
Behavioral Finance Roots:
Carlisle builds on the work of Kahneman, Tversky, Thaler:
Investors overreact to recent news
People fear losses more than they value gains (loss aversion)
They extrapolate short-term trends too far into the future
�Example: A stock that has underperformed for 3 years gets dumped by institutions,
regardless of fundamentals.
This creates deeply undervalued companies trading far below intrinsic worth — the deep
value playground.
Key Biases That Drive Deep Value Mispricing:
�A. Loss Aversion – (Kahneman & Tversky, 1979)
People feel the pain of a loss 2x more intensely than the pleasure of an equivalent gain.
�Market effect:
Investors sell falling stocks too quickly
Institutions dump underperformers to avoid ―career risk‖
Value stocks become irrationally underpriced
�B. Recency Bias – (Kahneman, Tversky, Barberis)
Investors overweight recent events and extrapolate them into the future.
�Example:
If a company’s earnings have fallen 3 years in a row, the market assumes it will keep
declining forever
But fundamentals often stabilize or mean-revert
� Deep value investors exploit this — they buy when others are extrapolating wrongly
�C. Overreaction & Overconfidence – (De Bondt & Thaler, 1985; Barber &
Odean, 2000)
Investors overreact to negative news, assume it reflects long-term truth, and trade
excessively on that belief.
�What happens:
Stocks fall too far below intrinsic value
Institutions sell to reduce ―volatility‖
Analysts downgrade based on optics, not fundamentals
� This createsshort-term mispricings that deep value investors can buy into before the
eventual rebound
�D. Disposition Effect – (Shefrin & Statman, 1985)
Investors tend to sell winners too early and hold losers too long.
� Irony: They hang onto bad glamour stocks longer than they should, while dumping
unloved value stocks at the worst possible time.
�SECTION 4: MEAN REVERSION VS. MOMENTUM
Carlisle contrasts two opposing views:
Momentum Hypothesis Mean Reversion
Trends persist Performance reverts to average
Buy winners Buy losers
―Growth is eternal‖ ―Bad companies recover‖
Deep value investors bet on mean reversion — that earnings, margins, and stock prices of
distressed companies eventually return to normal.
�Studies show that earnings momentum and analyst forecasts often fail, while value
strategies based on historical price metrics (e.g., P/B, EV/EBIT) succeed.
�SECTION 5: DATA-DRIVEN PROOF
Carlisle draws from his research and backtests:
From 1926–2011, deep value portfolios (lowest decile P/B or EV/EBIT)
consistently outperformed glamour portfolios.
His own model using the Acquirer’s Multiple (EV/EBIT) significantly outperformed
traditional value metrics like P/E.
�Even better performance occurred when these deep value stocks had catalysts, such as:
Shareholder activism
Management changes
Capital allocation shifts (e.g., buybacks)
�SECTION 6: INSTITUTIONAL LIMITATIONS
Carlisle explains why institutions avoid deep value strategies:
Constraint Effect
Career risk Avoid ugly stocks
Short-term mandates Must perform quarterly
Large AUM Can’t buy small caps or illiquid net-nets
Marketing pressure Avoid controversial companies
Herding Follow the crowd; avoid blame
These forces systematically steer them away from where the best long-term returns lie —
deep value.
Ironically, the more professional the money, the less contrarian it becomes.
�SECTION 7: WHO BENEFITS FROM THIS?
Carlisle argues that individual investors and quantitative value managers are uniquely
positioned to exploit this inefficiency.
Why?
They can invest in microcaps or net-nets
They can hold through drawdowns
They don’t face career pressure
He suggests that a simple, rules-based, contrarian strategy beats the experts precisely
because it is emotionless and patient.
�Example from history: Benjamin Graham’s net-net strategy, which found huge success
in the 1930s–60s, continues to outperform even now — when executed in an unemotional,
rule-based way.
THE MISMATCH
Why "great businesses" don’t always generate great stock returns — and why ugly, broken
companies do.
� I. THEORETICAL FOUNDATIONS
Carlisle roots ―The Mismatch‖ in two concepts:
1. Expectations vs. Reality
Borrowing from Lakonishok, Shleifer, and Vishny (1994):
Investors extrapolate past growth into the future
They overpay for glamour stocks because they expect perpetual outperformance
They underprice value stocks assuming recent poor performance continues
�Mismatch arises when expectations deviate from reality
→ Glamour stocks disappoint
→ Value stocks revert to the mean
2. Risk vs. Uncertainty (Knightian Distinction)
Risk: Measurable — e.g., 50% chance a coin lands heads
Uncertainty: Unmeasurable — e.g., what will iPhone demand be in 5 years?
Glamour stocks look safe but are priced under false certainty.
Value stocks look dangerous but are priced under uncertainty — which often corrects.
�The market avoids uncertainty, not because of rational risk aversion, but due to
psychological discomfort.
� II. HOW THE MISMATCH LOOKS IN DATA
Carlisle leverages Fama–French datasets and his own research:
Metric Top 10% (Glamour) Bottom 10% (Value)
P/B Ratio High (3–12x) Low (0.2–0.8x)
Earnings Trend Up Down or flat
Forecasted Growth 15–30%+ Negative or none
Actual Returns (20Y) ~5–8% annually ~14–20% annually
�Despite worse fundamentals, the value stocks beat the glamour stocks.
Because glamour stocks are priced for perfection, and value stocks are priced for disaster.
� III. BEHAVIORAL ANCHORS THAT DRIVE THE
MISMATCH
Bias Effect Result
Representativeness “Recent = typical” Extrapolate recent success/failure
Bias Effect Result
Availability heuristic News drives perception Analysts boost known winners
Herding Institutions copy others Overcrowding in same trades
Loss aversion Avoid ugly stocks Miss rebound potential
Confirmation bias Ignore signs of recovery in value stocks Miss turnaround
�These systematic human errors distort prices.
Carlisle argues that this is not market noise — it’s market pattern.
� IV. HOW TO TRADE THE MISMATCH
A. Glamour Stock Trap
Let’s say a cloud company trades at EV/EBITDA of 70.
Expectations: 30% YoY revenue growth, margin expansion
Reality: Revenue misses by 5%, margin compressed
Stock falls 60% in a week
�Mismatch Outcome: Even great businesses can't meet lofty expectations forever.
B. Value Stock Opportunity
A cyclical steel company trades at EV/EBIT of 2.
Margins compressed, debt high, but no bankruptcy risk.
Expectations: Business will never recover
Reality: Cycle turns, prices rise 10%, volumes improve
Market re-rates it to EV/EBIT 6
➡� Stock triples, even if earnings barely rise
�Mismatch Outcome: Poor companies don't need to become great — just less bad.
� V. MECHANICAL STRATEGY BASED ON THE
MISMATCH
Carlisle quantifies a repeatable system:
1. Screen for deep value:
EV/EBIT < 5
P/B < 0.8
P/E < 10
Avoid red flags (e.g., fraud, delisting risk)
2. Ignore narrative:
Don’t trust stories, media, analysts — these are biased toward glamour
3. Hold long enough:
Mean reversion takes 2–5 years
Avoid selling during noise (e.g., short-term earnings miss)
� VI. REAL-WORLD EXAMPLE: IRFC (India)
�Market perception (2021):
Government-run, slow, bureaucratic, not a ―real‖ business
P/B < 0.6, P/E ~4
Ignored by analysts, fund houses, and momentum traders
�Actual outcome (2023–2024):
Massive infrastructure spending
Steady earnings, strong dividend
Stock rallied over 400%
�Glamour story? No. Just deep value mispricing + stable business + time
