Course Outline

o Day 1: Return Predictability

o Day 2: Venture Capital I
o Day 3: Venture Capital II
o Day 4: IPO and LBO
o Day 5: LBO
o Day 6: Real Estate
o Exam
o The exam covers the lecture notes and
homework from this part of the course.
o You will not be asked to solve problems that
require computer simulations in the exam.
Constantin Charles FM230: Alternative Investments 1
 Quantitative Strategies
and
Return Predictability

Constantin Charles
c.charles@lse.ac.uk

Constantin Charles FM230: Alternative Investments 2

 Roadmap
o Event Studies
o Implementation
o Application and interpretation
o Return Anomalies
o Momentum
o Implementation, explanation w/ fund flow
o Post Earnings Announcement Drift
o Frictions and profitability of strategies
o Margin requirement, trading costs
o Some Other Anomalies
o Value premium, index inclusion, carry trade
Constantin Charles FM230: Alternative Investments 3
 Roadmap
o Event Studies

Constantin Charles FM230: Alternative Investments 4

 What’s an Event
o Firms often announce events, like:
o Financing/capital structure decisions:
o Issuance of debt, equity
o Payout policy:
o Dividend announcement, repurchase
program
o Mergers and acquisitions
o Question: Should I invest in firms with
certain events? If I do so, how much
extra return can I get?
Constantin Charles FM230: Alternative Investments 5
 Event Time I
o Let’s consider debt issuance as an
example.
o Suppose historical data is relevant.
o Challenge: firms do not all issue debt at
the same time. It is a firm-specific
decision.
o Some firms don’t issue debt at all.
o Some firms issue multiple times.
o We should study this decision in event
time.
Constantin Charles FM230: Alternative Investments 6
 Event Time II
o Identify all firms that issued debt in
your dataset
o Let t=0 be the date (in event time)
when debt is issued
o If some firms issued debt more than
once, these are two separate events
o Calculate the average return around
event time: t=-10, -9, …, -1, 0, 1, …9,
…, 10
o Question: suppose the average return
is positive. What does it mean?
Constantin Charles FM230: Alternative Investments 7
 Abnormal Returns
o A positive return alone does not mean
anything, this positive return may be
expected compensation for risk
o In most cases, we should use abnormal
returns (AR), rather than raw returns
o Abnormal return is the return over and above
what would have otherwise been expected
o It is the return associated with this event that
cannot be explained by our current
knowledge of risk.
o If abnormal returns are non-zero, investors
can use the event as a signal
Constantin Charles FM230: Alternative Investments 8
 Ways to Compute
o
Abnormal
Factor Models
Returns
o Estimate factor model (i.e., Capital Asset Pricing Model
(CAPM) or Fama-French 3 Factor Model) on pre-event
data by regressing the return on factors. This gives us
coefficients A and B:
o Rit-Rft=A+B(Rmt-Rft)+et
o Use A and B (beta) to calculate an expected return
around the event: ERit=Rft+A+B(Rmt-Rft)
o The abnormal return is ARit=Rit-ERit
o Control Group
o Identify a group of stocks which did not undergo the
event simultaneously to our security in real time
o Preferably these stocks should have similar risk
characteristics as the treatment group
o If Microsoft issued debt in May 2022, identify stock(s)
in the tech industry that did not issue debt in 2021-
2024, for example, Apple
o The abnormal return is RMSFT-RAAPL
Constantin Charles FM230: Alternative Investments 9
 Discussion of AR Calculations
o We have two approaches, factor models and
control groups. Which one is better?
o Factor models:
o Theoretically disciplined: based on statistics
o But what if the model itself (CAPM for
example) is wrong? E.g.: missing factor
o Control Group:
o Simple to perform
o Market disciplined: directly comparing returns
o But is subject to the choices of comparing
firms.

Constantin Charles FM230: Alternative Investments 10

 Application: Leverage Events
o Let’s now use an event study to understand
the effects of a firm’s capital structure
decision.
o Capital Structure: (use leverage as a proxy)
o Leverage=Debt/Assets=Debt/(Debt+Equity)
o Higher Debt or Lower Equity => Higher
Leverage
o Higher leverage means more volatile cash
flows and more risk, and potentially tax
shields.
o Finding: Prices jump up if leverage increases
and drop if leverage decreases.
Constantin Charles FM230: Alternative Investments 11
 Application: Stylized Facts

Constantin Charles FM230: Alternative Investments 12

 Trading Opportunities
o A jump at t=0 says nothing about market inefficiency
because the market reacts to newly released information
o Meaningless for trading, unless you have private info
o Non-zero abnormal returns post event (t>0) may
indicate market inefficiency since prices did not fully
adjust to new information
o Buy (sell) after event if abnormal returns positive
(negative)
o However, it may be that event changes risk
characteristics of firm and previously estimated beta
no longer valid so abnormal returns not really different
from zero
o Non-zero abnormal returns prior to event (t<0)
suggests possible anticipation of event, limited early
release of information, or insider trading
o Observing run up à Possibility to anticipate the event
and trade on it ahead of time
o An alternative (less sinister) explanation is that non-
zero abnormal returns at t<0 make event more likely
Constantin Charles FM230: Alternative Investments 13
 Event Time Analysis

Constantin Charles FM230: Alternative Investments 14

 Over- and Underreaction
o Why do we have post-event abnormal returns?

o Simple explanation: Psychology!

o Overreaction: people get excited about positive
news, stocks that did well are overpriced, will
do poorly in the future
o Underreaction: people do not fully react to
positive information, stocks that did well are
underpriced, will continue doing well

o A criticism of psychology-based stories (a.k.a.

behavioral stories): lack of discipline.

Constantin Charles FM230: Alternative Investments 15

 Underreaction

Constantin Charles FM230: Alternative Investments 16

 Overreaction

Constantin Charles FM230: Alternative Investments 17

 Roadmap
o Event Studies
o Execution
o Application and interpretation
o Return Anomalies
o Momentum

Constantin Charles FM230: Alternative Investments 18

 Momentum Overview
o Return anomalies: observed return patterns that
cannot be explained by known risk stories.

o Momentum: Stocks that did well continue to do

well over the next 12 months, then a reversal in
following 24 months
o Jegadeesh and Titman (1993)
o Consistent with initial underreaction, long term
overreaction
o Momentum effect seems to be fairly robust and
additional studies have not yet found evidence of
it having gone away
o Many funds are still trading Momentum.
o They refine the strategy to minimize stock
turnover and transaction costs.
Constantin Charles FM230: Alternative Investments 19
 Strategy Implementation
o Each July 1st compute the return for each
security over the previous J months
o J=3, 6, 9, 12
o Sort along the return and assign each
security into one of 10 portfolios based on its
return over the previous J months.
o Portfolio 10 will have the best 10%
performers, portfolio 1 will have the worst
10%
o Buy portfolio 10 and short portfolio 1 and
maintain this position for K months
o K= 3, 6, 9, 12
o This results in 16 possible strategies, usually
called “J-month / K-month” strategy.
Constantin Charles FM230: Alternative Investments 20
 Portfolio Returns (Monthly)

o Table 1 from Jegadeesh and Titman (1993)(B uses 1-week lagged returns)
Constantin Charles FM230: Alternative Investments 21
 Can Momentum Be
Explained by Known Risk Stories?

o It is possible that P10 is more risky than P1, (higher

beta or smaller size), and therefore earns higher
returns
o Double sort portfolios based on past performance
and historical beta (β3 is the highest beta, captures
correlation with market risks)
o If momentum is coming from beta risk, high returns
should be concentrated in high beta portfolios only
o Double sort portfolios based on past performance
and size (S3 is the largest)
o If momentum is coming from small firm risk, high
returns should be concentrated in small portfolios
only
o Neither size nor market risk seem to explain
momentum
Constantin Charles FM230: Alternative Investments 22
 Double Sort on Size and Beta

o Table 3a from Jegadeesh and Titman, 6 months / 6 months strategy

Constantin Charles FM230: Alternative Investments 23
 Regression Approach to
Account for Known Risk Factors
o Regress momentum portfolio on market
return
o The intercept is the CAPM alpha (abnormal
return)
o If this is positive for P10-P1 then momentum
returns is not accounted for by CAPM risk
o Non-zero alphas remain
o An alternative is to regress momentum
portfolio on FF’s 3 factors and check if there
is an alpha
o The intercept is the FF 3 factor alpha
o If this is positive for P10-P1 then momentum
returns are not accounted for by FF 3F risk
o FF’s model was not yet popularized when
Jegadeesh and Titman wrote about
momentum in 1993
Constantin Charles FM230: Alternative Investments 24
 Momentum

o Table 3b from Jegadeesh and Titman

Constantin Charles FM230: Alternative Investments 25

 Event Time Analysis
o It is useful to consider performance of a
strategy in event time (relative to purchase)
to know when to get out
o Cumulative return (not abnormal) is 9.5%
over first 12 months
o This is likely big enough to withstand
transaction costs
o Cumulative return is 5.5% after 24 months
o Thus, reversal in months 12-24
o Cumulative return 4% and not significant
after 36 months

Constantin Charles FM230: Alternative Investments 26

 Event Time Stylized Facts

o Table 7 from Jegadeesh and Titman

Constantin Charles FM230: Alternative Investments 27
 Fund Flows
and Momentum
o Momentum is partially explained by
mutual fund flows
o “A flow-based explanation for return
predictability” Lou (2012)
o Fund A: 75% in stock 1, 25% in stock 2
o Fund B: 25% in stock 1, 75% in stock 2
o Suppose there is an unanticipated
positive shock to stock 1
o Similarly, an unanticipated negative
shock to stock 2
o Fund A outperforms fund B (just luck)
Constantin Charles FM230: Alternative Investments 28
 Fund Flows
and Momentum II
o Investors (empirically) invest more into
outperforming funds
o Inflows for A, outflows for B
o But if A got lucky, why do this?
o Funds (empirically) tend to not change
allocation
o A invests 75% of new money in stock 1, 25%
in stock 2
o B divests 25% of lost money from stock 1,
75% from stock 2
o But if shock unanticipated, why do this?
o Excess demand pushes price of stock 1 up, and
the price of stock 2 down
o Eventually, prices revert to their true value
Constantin Charles FM230: Alternative Investments 29
 Roadmap
o Event Studies
o Implementation
o Application and interpretation
o Return Anomalies
o Momentum
o Implementation, explanation w/ fund flow
o Post Earnings Announcement Drift

Constantin Charles FM230: Alternative Investments 30

 What’s Post Earnings
Announcement Drift (PEAD)
o PEAD: Ball and Brown (1968); Foster, Olsen, Shevlin
(1984); Bernard and Thomas (1989); Chan,
Jegadeesh, and Lakonishok (1996); several more
recent studies
o Most of below follows BT and CJL
o Firms regularly announce earnings
o Usually once per quarter
o Studies find that when earnings are above (below)
expectations, subsequent abnormal returns are
positive (negative) for 60-180 days
o Note that what matters is a positive (negative)
surprise to earnings, not positive (negative) earnings
o Consistent with underreaction
o Pre-Event abnormal returns are also positive
o Note the relationship with momentum

Constantin Charles FM230: Alternative Investments 31

 Expected Earnings I
o In order to identify an earnings surprise, we
must first identify expected earnings
o Note that for many firms, earnings are highly
seasonal.
o For example, people buy more iPhones in the
fall, after the new iPhone is announced (and
just before the holidays/Xmas).
o Common factor in earnings:
o Regress a firm’s earnings on some factors
(like market earnings) using lagged data:
Et=A+B*EMt+et
o Calculate abnormal earnings as the difference
between announced earnings, and earnings
predicted by regression: AEt=Et-(A+B*EMt)
Constantin Charles FM230: Alternative Investments 32
 Expected Earnings II
o SUE: Standardized Unexpected Earnings
E i
- E i
SUEti = t t -4
StDev Eti - Eti- 4[ ]
o A simple model where past earnings are the
best forecast of future earnings
o Foster, Olsen, Shevlin show that this does
just as well as more complicated models
o To control for seasonality, the surprise is the
difference between today’s earnings and
earnings 4 quarters ago
o The surprise is normalized by the standard
deviation of the past 8 surprises

Constantin Charles FM230: Alternative Investments 33

 Expected Earnings III
o Many firms are followed by analysts, who
forecast earnings. Any time an analyst revises
her forecast can be considered a surprise
o Analysts’ estimates are not always revised
every month, thus take 6-month average
o REV6: 6-month moving average of revised
earnings forecasts by analysts
f t i- j - f t i- j -1
REV6 it = å
j =0,6 pti- j -1
o f is the consensus (mean) I/B/E/S
(Institutional Brokers’ Estimate System)
estimate; estimates are scaled by price
Constantin Charles FM230: Alternative Investments 34
 Proxy for Expected Earnings
o We ultimately care about market’s perception of the
announced earnings.
o Why not just use Cumulative abnormal stock return at
announcement time
A BR ti = å ( R ti+ j - R tM+ j )
j = -2 , +1

o If earnings were truly a positive (negative) surprise,

then firm’s stock returns should be very positive
(negative) at time of announcement
o Pro: No need for a model of expected earnings
o Con: Need to model expected returns
o The equation above is widely used and assumes CAPM
and beta=1, could estimate beta on historical data or
use FF3 model to get expected return

Constantin Charles FM230: Alternative Investments 35

 Implementation
o Pick some measure of earnings surprises, for example SUE
o Look back 6 months, sort all firms based on SUE, and set
cutoffs to assign firms
o What is SUE of top 10%, top 20%, bottom 10%, etc.?
o Look back because should only use information currently
available to you
o Assign firms to one of the portfolios based on these cutoffs
o Track the abnormal return of these portfolios in event time
o Bernard and Thomas (1989) simply use Rit-Rmt as
abnormal return
o Note that this is not yet a trading strategy, a trading
strategy would need to specify:
o Which of 10 portfolios to long/short
o When to unwind positions
o Looking at performance in event time helps us formulate a
trading strategy

Constantin Charles FM230: Alternative Investments 36

 PEAD

o Table 1 from Bernard and Thomas (1989): CAR

Constantin Charles FM230: Alternative Investments 37

 PEAD
o After a positive (negative) announcement
drift is positive (negative)
o P10-P1 has 60-day CAR of 5.3%, 4.5%,
2.7% for small, medium, large firms!
o Disproportionate amount in first 5 days but
significant through 180 days

o This is potentially very useful to a trader

o Buy portfolios with positive surprises, short
portfolios with negative surprises
o Unwind position 60-180 days after
announcement (delivers max CAR)

Constantin Charles FM230: Alternative Investments 38

 Event Time Analysis

o Figure 1 from Bernard and Thomas (1989), reprinted from Foster, Olsen,
Shevlin (1984)

Constantin Charles FM230: Alternative Investments 39

 Pre- and Post- Announcement

o Figure 2 from Bernard and Thomas (1989)

Constantin Charles FM230: Alternative Investments 40
 Large Firms Only

o Figure 3 from Bernard and Thomas (1989): PEAD large firms only
Constantin Charles FM230: Alternative Investments 41
 Can You Pocket
the Pre-Announcement Returns?
o Prior to a positive (negative) announcement, drift is
positive (negative) and large compared to post-
announcement returns

o This is not necessarily useful information since at t=-60 you

don’t know what the announcement surprise will be

o On the other hand, you can use the drift to predict the
direction of announcement
o Suppose that an announcement is coming up and you
observe a positive pre-announcement driftà Expect
announcement to be positive

o May be indicative of market anticipating surprise or early

release of info
o Note drift speeds up just before announcement

Constantin Charles FM230: Alternative Investments 42

 Can Risk Explain PEAD?
o Could it be that the positive returns are
explained by higher risk?
o P10 indeed has higher beta than P1 exactly
during the period of higher returns
o High betas: (-59,0) and (1,180)
o No difference: (-120,-59) and (181, 240)
o However, beta differences seem too small:
o ∆β*RM=0.2*1.5%=0.3% but return
differences are ≈ 4%
o Furthermore P10-P1 is consistently positive
(46 of 50 quarters, 13 of 13 years)
o Perhaps CAPM is not a good model of risk and
a risk-based explanation exists after all?
Constantin Charles FM230: Alternative Investments 43
 Can Risk Explain PEAD?

o Table 2 from Bernard and Thomas (1989): Beta differences

Constantin Charles FM230: Alternative Investments 44
 Robustness of PEAD

o Figure 5 from Bernard and Thomas (1989): consistency of returns

Constantin Charles FM230: Alternative Investments 45

 An Explanation from Psychology
o People care about gains and losses relative to some
starting point, not their total amount of money
o Disposition Effect: value function concave over
gains (risk averse) but convex over losses (risk
loving)
o Negative announcement à in the loss region à risk
loving and do not sell à underreaction
o Positive announcement à in the gain region à risk
averse and sell too early à underreaction
o Frazzini (2005): Finds empirical support for this:
stock prices underreact more to bad news when
more current holders are facing a capital loss,
underreact to good news more when more current
holders are facing a capital gain

Constantin Charles FM230: Alternative Investments 46

 Disposition Effect: Utility Function

Constantin Charles FM230: Alternative Investments 47

 Roadmap
o Event Studies
o Implementation
o Application and interpretation
o Return Anomalies
o Momentum
o Implementation, explanation w/ fund flow
o Post Earnings Announcement Drift
o Frictions and profitability of strategies

Constantin Charles FM230: Alternative Investments 48

 Why Do We Have Anomalies?
o Trading strategies create price pressure.
o For example, consider momentum: long past
winners, short losers.
o Current price of the past winners increases due
to the extra buying pressure.
o If enough people trade momentum, then the
winners’ price will be so high to wipe out any
positive future returns.

o Question: people have been trading strategies

like momentum and PEAD for decades, why do
we still observe return anomalies?
o Perhaps there is something that prevents
traders from taking positions or realizing
returns.
Constantin Charles FM230: Alternative Investments 49
 Margin Requirements I
o What’s a margin requirement?
o Whenever you buy/sell asset, you need to
post some capital as collateral.
o No margin: Short $1 of A, Long $1 of B
o Initial cost of trade is zero à you can
make this trade as large as you want
o $1 Gazillion
o A goes up by 1%, B goes down by 1%
o Profit: -0.02
o Percent return: -∞
o A goes down by 1%, B goes up by 1%
o Profit: 0.02
o Percent return: ∞
Constantin Charles FM230: Alternative Investments 50
 Margin Requirements II
o 10% Margin: Short $1 of A, Long $1 of B
o Deposit $0.10 (10% x Size of short
position) with broker
o Initial cost of trade is 0.10
o A goes up by 1%, B goes down by 1%
o Profit: -0.02, Percent Return: -20%
o A goes down by 1%, B goes up by 1%
o Profit: 0.02, Percent Return: 20%
o If you have $1 million, you can take a
position of $10 million at most
Constantin Charles FM230: Alternative Investments 51
 Frictions: Overview
o Bid/Ask spread: You buy for more than the average price,
and sell for less than the average price
o The difference (“spread”) is how market makers make
money
o Spread is small for large/liquid stocks but large for
smaller/illiquid stocks
o If a strategy relies on heavily on small/illiquid stocks, it is
less likely survive transaction costs
o Short sale constraints: It is impossible to short sell
certain stocks, and even if short selling is possible, it is
usually costly
o Margin requirements limit size of short position
o If a strategy relies heavily on short selling, it is less likely
to work in the real world
o Liquidity: Some stocks may be difficult to sell
o This is especially true during “extreme” times
o Liquidity risk may make strategies much riskier than they
appear (LTCM)

Constantin Charles FM230: Alternative Investments 52

 Bid/Ask Spread
o Bernard and Thomas (1989) also consider
transaction costs associated with a PEAD strategy
o They consider bid/ask spreads of 4% for small and
2% for large firms
o Since strategy involves going long in P10 and
short in P1, we must multiply these numbers by 2
o Their strategy involves 78% turnover each quarter
o 22% of firms that were in either P10 or P1 stayed
in P10 or P1, so no need to rebalance
o Implied cost is .78*2*4%=6% for small firms and
.78*2*2%=3% for large firms
o These are above the 60-day CAR but below 180-
day CAR, so the strategy may still be profitable
after transaction costs
o However, it is far less profitable, and will be
consistently positive far less often

Constantin Charles FM230: Alternative Investments 53

 Liquidity
o Chordia, Goyal, Sadka, Sadka, and
Shivakumar (2006) investigate the role of
liquidity in PEAD
o A PEAD long/short strategy provides a return
of 0.24%/month in the most liquid stocks but
1.79%/month in the most illiquid stocks
o Before subtracting out costs

o Illiquid stocks have high trading costs and

market impact costs
o They find that transaction costs account for
66% to 100% of the apparent PEAD long/short
strategy

Constantin Charles FM230: Alternative Investments 54

 Frictions

o Table 2 from Chordia, Goyal, Sadka, Sadka, and Shivakumar (2006)

Constantin Charles FM230: Alternative Investments 55

 Frictions
o Everyone knows about momentum,
many people trade momentum
o The best funds don’t just trade
momentum, but minimize the
associated costs
o Higher frequency strategy à more
costs

Constantin Charles FM230: Alternative Investments 56

 Roadmap
o Event Studies
o Implementation
o Application and interpretation
o Return Anomalies
o Momentum
o Implementation, explanation w/ fund flow
o Post Earnings Announcement Drift
o Frictions and profitability of strategies
o Margin requirement, trading costs
o Some Other Anomalies

Constantin Charles FM230: Alternative Investments 57

 Value Premium
o Define ”book-to-market ratio”:
o B/M=Book value of Equity/Market
Value of Equity
o High B/M stocks are called “value stocks”,
low B/M ones are called “growth stocks”.

o Value Premium: a strategy that goes

long value stocks and short growth
stocks generates positive abnormal
returns.

Constantin Charles FM230: Alternative Investments 58

 Value and Momentum
Everywhere?
o Asness, Moskowitz, and Pedersen
(2009) do a comprehensive study of
value and momentum
o Individual stock momentum in 4
markets (US, UK, Europe, Japan)
o Country Equity Indexes (18 countries)
o Gov’t Bonds (10 countries)
o Currency (exchange rates among 10
currencies)
o Commodity (27 commodity futures)

Constantin Charles FM230: Alternative Investments 59

 Value and Momentum
Everywhere?
o Construct momentum in standard way:
past 12 months performance
o Construct value in standard way for
equity, for others akin to measuring
current price to old price
o Commodities: 5-year-old spot price
relative to spot price today (like B/M)
o Note that value is negatively related to
momentum by construction, though the
horizon is different

Constantin Charles FM230: Alternative Investments 60

 Value and Momentum
Evidence

Alphas (Table 2) from Asness, Moskowitz, and Pedersen (2009)

Constantin Charles FM230: Alternative Investments 61
 S&P 500
Inclusion Effect
o S&P 500 regularly includes/deletes firms from the index
o 1993-2001: 246 firms added/deleted
o Changes made because of mergers, or changes in size,
volume, financial soundness, industry, etc. of underlying
firms
o Changes are announced publicly approximately 5 days
before the change is made
o Stocks included in the S&P 500 index increase in value
after inclusion
o Shleifer (1986) finds a 3% increase in value between
1976 and 1983
o Chen, Noronha, Singal (2004) find a 5.4% abnormal
return (1989-2000)
o Index tracking funds must increase holdings of this
security once it is in the index
o Other institutions likely to increase holdings too
o Increased demand drives up prices
o Being in S&P 500 also increases visibility and liquidity
Constantin Charles FM230: Alternative Investments 62
 Carry Trade
o Borrow in a currency with a low interest rate
(e.g., Japanese Yen, Swiss Franc, US dollar)
o Invest in currency with a high interest rate
o This strategy has historically earned high
returns
o However high returns may simply be
compensation for high risk
o Carry trade strategy is highly skewed: it does
extra bad in bad times
o Correlated with measures of aggregate risk

Constantin Charles FM230: Alternative Investments 63

 Summary
o There are multiple quantitative strategies that
deliver high returns by trading on:
o Leverage changes
o Momentum
o Post-Earnings Announcement Drift
o Inclusion Effects
o Carry Trade

o Important! It is not clear whether these

strategies…
o Are simply fair compensation for risk
o Survive transaction costs
o Work out-of-sample (i.e., in the future)

Constantin Charles FM230: Alternative Investments 64