Hedge Funds 360

Marco Avellaneda
New York University &
Finance Concepts LLC

Summary of the course

Hedge Funds 360 is meant as an introduction to Hedge Funds, the new
and not-so-new investment product which has flourished and gained media
attention over the last 20 years
HF AUM is approximately USD 2.5 Tn. In comparison, US debt is 14 Tn and
investable assets are approximately USD 50 Tn.
We will discuss
1. The point of view of investors
2. The point of view of managers
3. Main strategies & themes
4.

A systemic view of HFs

Hedge Funds from the point of view of

investors

Review: Mutual Funds

In the US, mutual funds are regulated by the Investment Company Act
of 1940
Law designed to protect the public against unscrupulous managers
Fees, diversification, concentration, transparency, are all regulated by IA 1940
Begins the modern era of equity fund investment in the US.
Mutual Funds allow for cheap access to stock portfolio trading by the
public, encouraging saving in the stock market
Defined by benchmarking in relation to a particular index or sector, e.g.
S&P 500, Russell 2000.

Main Characteristics of MF
Long only (no short selling, no leverage)
Tracking benchmarks or indexes
Daily liquidity
Low fees (<10 basis points for passive funds)
Tradeoff: in order to advertise MFs and sell them to the public,
the company has to be regulated and satisfy considerable legal requirement
What is a good MF? A fund that can track well an index with low cost OR
a fund that can outperform the index but that has similar characteristics.

Example: Vanguard S&P500 Investor

Shares (VFINX)

How Mutual Funds Work

NAV= Net Asset Value
Redemption of fund shares at
NAV, with T+1 liquidity

Investor

Loads and fees apply for fast trading

$$

Mostly passive investment, long only

Fund Shares

$$
Market

Mutual Fund

Stock basket

Hedge Funds
Much less structured and regulated that MFs.
Usually marketed as a Limited Partnerships or Trusts to Qualified Investors
Fund does not track a benchmark: absolute returns versus indexing
Said to be invented by A.W. Jones in the later 1940s, as a fund that
would both buy AND sell shares
Known as alternative investments, as in ``alternative to stocks and
bonds.
In theory, HFs are uncorrelated with the stock market, because they
are not benchmarked to equity indexes, but rather to cash.

HF Management, Liquidity & Fees

HFs can be seen as active management vehicles.
Manager has considerably more discretion about the kinds of
investments he or she makes
Liquidity is limited. Can be monthly, quarterly or even annually.
Fees are not standardized and can vary according to the investor.
(Usually a management fee of the order of 1% to 2% and an
incentive fee which is a percentage of profits (e.g. 10% or 20%)
A hedge fund is an institutional form of an Investment Advisor (a
figure also contemplated by the IA 1940.) Investment Advisors
typically work with Managed Accounts

Investment Advisor with several Managed

Accounts
Advisor
Investor A

Acc #1
Acc #2

Investor B
Acc #4
Investor C

Market
trading

Acc #20
Investor Z

Investment advisors manage accounts on behalf of their clients.

Omnibus account

IAs must register if they have more than a certain number of clients. Restrictions on
fees apply, according to ICA 1940.

Commodity Trading Advisors

Commodities is (was?) an asset class perceived by investors to deliver uncorrelated
returns to the stock market.
Commodity trading is done via futures markets.
Natural framework for managed accounts.
Commodity Trading Advisors (CTAs) provide futures trading strategies
which seek to benefit from trends (periods of increase, periods of decrease)
of commodities and currencies. (MAN is the biggest CTA in the world)
I believe that CTAs and ``commodity pool operators// originated in the 1970s with
the rise of inflation in the US following the Vietnam War and bear market
CTAs are offered by intermediaries (brokers) to qualified investors.

Hedge Funds: ``commingled active

management
$$ invested
Fees
Investors
Shares in LLC

Marco Avellaneda
Superior Returns
Fund, LLC*

$$

Market

PB, EB,
Banks**

Advice, investment strategy.

Fees

orders

MA Advisory, Inc.
(manager)

* Single advisee

**Prime broker, Execution Broker, Custodian, Administrator

How is the HF market accessed?

Banks acting as brokers (traditionally for MFs)
Wealth-management advisors (private banking)
Family offices
Fund of Hedge Funds
Direct contact via Internet site

Managed-account platforms (e.g. Lyxor)

Performances and NAVs are published in Bloomberg pages for

Listed Funds. Often listing in Dublin Stock Exchange, or Malta Stock
Exchange , or other exchanges.

HF information is restricted (particularly in the US) due to the Investment Company

Act of 1940, which regulates advertisement of alternative investments and funds

1. Private Banks and Wealth Managers

Provide due diligence and keep track of a list of managers that the recommend
to clients (e.g. Lombard Odier, GSAM, MSAM)
Due diligence is very different from Mutual Funds and ETFs due to the differences
in regulation. Much more delicate.
Manager may not reveal the details of the strategy; only the general investment
theme and/or asset class.
Hedge fund analysts select the best managers and charge an advisory fee.
These agents may or may not be remunerated by the HF manager.

2. Family Offices and Fund of Hedge Funds

Family Offices typically manage the assets of on or more wealthy families.
Fos have traditionally invested in alternatives via managed accounts or
HFs.
FO s employ investment professionals for fund selection and to construct
a portfolio of investments.
Fund of Hedge Funds are wealth management companies how issue shares
and invest $ in HF. They also perform due diligence on behalf of clients and
risk-manage the portfolio of HFs.
FoHFs can re-allocate among HFs and also leverage.
Caveat emptor.

Largest Hedge Funds (Top 20)

FUND
COUNTRY
Bridgewater Associates
US
MAN
UK
JP Asset Management
US
Brevan Howard
UK
Och-Ziff Capital
US
Paulson & Co.
US
Black Rock Advisors
US
Winton Capital
UK
Highbridge Capital
US
BlueCrest Capital
UK
Baupost Group
US

AUM
77.6
64.5
46.6
36.6
28.5
28
27.7
27
26.1
25
23

FUND
COUNTRY
Cerberus Capital
US
D.E. Shaw & Co.
US
Angelo Gordon Co.
US
AQR Capital Management
US
Farallon Capital Mgt.
US
Goldman Sachs Asset Mgt.
US
Elliott Management
US
King Street Capital
US
Canyon Partners
US
Scenic Stone Capital
US
TOTAL

Total Assets in HF: approximately 2500 billion USD

Top 20 HFs concentrate 25% of all investment assets
Probably the top 100 have 90% of AUM.

AUM
23
23
22
20.5
20
19.5
19
18.5
18.1
12.3
606.5

Some broad HF Categories

Distressed Debt/Restructuring
Merger Arbitrage

Equity Market Neutral

Quantitative Equities (Factor based, stat arb)

CTAs Trend Following

Macro
Emerging Markets
Sectors: Heath Care, Technology

The managers perspective

Starting a Hedge Fund

Raising initial capital

Choosing the business structure

Prime broker
Execution broker
Computer and office equipment
``Soft-dollar funding
HF Incubators

Hiring traders, marketers and back-office

Offering memorandum and marketing

Raising Initial Assets

Initial assets should come from personal funds or first investors
Skin in the game
First investors typically want equity
Later investors want good deals too
Reduced fees

Capacity
Starting with Managed Account?
Track record
Managers pedigree

Business structure
Managed accounts or co-mingled funds?
Onshore managers
Master-feeder structure (onshore/offshore)
Administrator (bank)
MA Capital
Management
LLC (NY)

Prime Broker
MA Capital
(Bermuda*)

MA Gestion, S.A.
(Paris)

Market
* Offshore vis a vis US Tax

UCITS (European Union)

European Union has UCITS legislation, which harmonizes
Mutual Fund, ETFs, FoHFs and Hedge Fund offerings for
all member countries

It is important to know the rules to be able to market the fund

Malta, Luxemburg, Jersey, Guernsey, Isle of Man are the most
common domiciles for HFs due to light taxation of the Master Fund
Legal fees for registration in an offshore venue plus offering memorandum
can run easily above 100K Euros.

Hedge Funds in LATAM

In Brazil, there are three kinds of investment funds
-- Renda Fixa

(fixed income)

-- Renda Variavel

(equity mutual fund)

-- Multi-mercado

(allowed to short, more leeway for manager)

Each country in LATAM has its own legal framework for investment management
and its own investor culture.
It is important to set up the correct business structure, which most likely will have
the management company domiciled offshore and a local entity for investors.
Fiscal matters and qualified investor requirements are VERY important everywhere.
Many HFs in LATAM are sponsored by a bank (e.g. Credit Suisse Hedging Griffo (Brazil))
who is the first investor, provides the infrastructure and takes an equity interest

Prime Broker
Prime brokers provide the connection between the HF and the markets
and perform key functions.
-- Execution
-- Clearing
-- Financing
-- Stock loan
-- Custody
-- Capital Introduction

The prime broker is the way the fund accesses the market and registers
trades. It also acts as a bank, lending money to the fund. Some funds have
multiple PBs for reasons of confidentiality of traders.

Example of a fund executing a trade in

the listed equities market
6. report
HF

1. order

Prime Broker

3. report

Execution Broker

4. give-up*

2. trade

5. clear

Market

* If EB is not PB, a ``give up is made to transfer the trade from the EB to the PB

Financing
PB provides financing for Funds positions
Example: long equities at Fed Funds + 30 bps
short equities at Fed Funds 30 bps (GC)
PB arranges for stock-loan arrangements for short positions
- maintains inventory of stock for lending
- stock loan desk can locate stock for borrowing
PB can also replace stock transactions by swaps (CFDs)
PB gives access to OTC trading in Credit swaps, IR swaps, MBS, bonds, etc.,
and provides financing via repo markets.

Infrastructure, Office Equipment, Fees

Hedge fund ``hotels
Soft-dollar agreements
-- data
-- research
-- office equipment
Conflicts of interest with SD agreements
Fees: Full Service versus Cost Plus
Relationship with PB and other investors plays an important
role in start-up funds and funding operations before fund
is profitable

Back of the envelope ``fee mathematics

Fiduciary obligation of manager vis a vis HF partners/investors

Capital Introduction
PB s offer their clients Capital Introduction events
Typically billed as ``startup fairs
Attended by big multi-manager funds, FoHFs.
Usually held in major HF cities (e.g. London, New York, Geneva)

HF conferences (GAIM, etc).

Meeting investor requirements

Track record or solid backtesting idea

Minimum AUM

Risk-management

Compliance

Key persons
Administrator, Custodian, PB
Well-defined style

Case Study #1: Quantitative Fund

(Europe)
Originally a ``trend following futures fund (CTA) , founded early 1990s
Early to mid-2000s expands from < 1Bn AUM to 5 Bn. AUM
Adds a quantitative, market-neutral equity fund
Adds a quantitative options fund
Completely systematic (no human trading, no fundamental info)
Strict real-time risk-management
Position turnover: few minutes for few hours

Personnel: mostly researchers in equities and futures strategies and

a few administrative staff.

Case study #2: Long-short equity fund

(NY, fundamental)
Created after Internet Bubble
Core funds: Technology, Health Care
Staff: Equity research, trading, back-office

Investment process: trades suggested by research staff that

goes over company reports and gets research from sell-side.
(``Mosaic theory)

Traders (Portfolio Managers) use the ideas to construct active

trading
Hedging with options and ETFs (limited exposure, either to long or
short side).
Turnover: weeks, months

Success of HF business: a few thoughts

for managers
Market for managers and investment/trading talent
Past 10 years: entrance of pensions and endowments; more
attention to administration
Importance of returns vs. risk management
Importance of risk-management versus returns

Liquidity
Transparency
Multi-strategy is necessary to survive in the post 2008 period
Talent for business administration & marketing

Hedge Funds Strategies and Styles

HF
Styles
Fundamental,
research based,
discretionary

Systematic,
``rules-based
Trend-following
(futures,
currencies) CTA

MAN, GLG

Equity market
neutral, stat-arb

CFM, AQR

Market-making
(limited capacity
High-frequency)

JANE ST ?

Event driven,
risk arbitrage

ACKMAN

Convertible
arbitrage

KALLISTA

Global macro

Fundamental GRAHAM
CTA

Tail funds UNIVERSA

Volatility
Relative
value

TUDOR JONES,
SOROS

CAPSTONE

Fixed income,
LTCM ?
credit

Equity longshort

SAC

Trend Following CTA

Trend following is an investment or trading strategy which tries
to take advantage of long, medium or short-term moves that
seem to play out in various markets.
Traders who employ a trend following strategy do not aim to
forecast or predict specific price levels; they simply jump on the
trend (when they perceived that a trend has established with
their own peculiar reasons or rules) and ride it.
Traders normally enter in the market after the trend "properly"
establishes itself, betting that the trend will persist for a long
time, and, for this reason, they ignore the initial turning point
profit.

A market "trend" is a tendency of a financial market price to

move in a particular direction over time. If there is a turn
contrary to the trend, they exit and wait until the turn
establishes itself as a trend in the opposite direction. In case
their rules signals an exit, the trader exits but re-enters when
the trend re-establishes.

Statistical Arbitrage
StatArb evolved out of the simpler pairs trade strategy, in
which stocks are put into pairs by fundamental or market-based
similarities.
StatArb considers not pairs of stocks but a portfolio of a hundred
or more stockssome long, some shortthat are carefully
matched by sector and region to eliminate exposure to beta and
other risk factors.
Portfolio construction consists of two phases. In the first or
"scoring" phase, each stock in the market is assigned a numeric
score or rank that reflects its desirability.

The details of the scoring formula vary, but, generally (as in pairs
trading), they involve a short term mean reversion principle so
that, e.g., stocks that have done unusually well in the past week
receive low scores and stocks that have underperformed receive
high scores.

HFT, Market-making
High-frequency trading (HFT) is the use of sophisticated technological tools and
computer algorithms to trade securities on a rapid basis.[ HFT usually uses
proprietary trading strategies that are carried out by computers.

Unlike regular investing, an investment position in HFT may be held for only
seconds, or fractions of a second (though sometimes it may extend to longer), with
the computer trading in and out of positions thousands or tens of thousands of
times a day. At the end of a day of HFT, there is no open position in the market.

Firms engaged in HFT rely heavily on the processing speed of their trades, and on
their access to the market. Many high-frequency traders provide liquidity and price
discovery to the markets through market-making and arbitrage trading; and highfrequency traders also take liquidity to manage risk or lock in profits.
As of 2009, studies suggested HFT firms accounted for 60-73% of all US equity
trading volume, with that number falling to approximately 50% in 2012.

Long Volatility Funds

In the aftermath of the (i) Internet bubble and (ii) 2007-2008, funds that
systematically bought options and delta hedged, made a lot of money
Following that development, the concept of long volatility fund, or tail
risk fund was established
Such funds maintain an inventory of long premium (rolling long positions
in options). The thinking is that the cost of the premium will be less than
the gains from vega/gamma if the market crashes.
In general, these vehicles lose money except when there is a large crisis
like 2008
Long volatility funds still elicit interest from investors, to some extent.

Harcourt Quarterly Hedge Fund Presentation

Paris, November 2004

Trading Market Volatility

Marco Avellaneda
Capital Fund Management SA
6 Boulevard Haussmann
75009 Paris
[email protected]

Volatility Trading in a Nutshell

Long volatility: buy call option, sell shares
buy put option, buy shares
Short volatility: sell call option, buy shares
sell put option, sell shares

Long Vol

Short Vol

SP 500 June05 1125 Call

Vol=16%
180
160

Theoretical Value

140
120
100
80
60
40
20
0
950

975

1000

1025

1050

1075

1100

1125

Index value

1150

1175

1200

1225

1250

1275

1-day P/L for Long Call/Short Stock

(Constant volatility=16%)
50

40

30

20

10

0
950

975

1000

1025

1050

1075

1100

1125

1150

1175

1200

1225

-10

P/L n 2 1

daily time - decay , n

percent index change

expected daily volatility

1250

1275

Assuming an implied
volatility drop of 1%
Vol=15%
50

40

30

20

10

0
950

975

1000

1025

1050

1075

1100

1125

1150

1175

1200

-10

3.80 loss if stock does not move

and volatility drops 1%

1225

1250

1275

A closer look at the profit-loss

due to a change in volatility
6
5
4
3
2
1
0
1075

1100

1125

1150

-1
-2
-3
-4
-5

1% move in vol => 8% move in premium for a 6m ATM option

1175

Book-keeping: profit/loss
from a delta-hedged option position

P/L n 2 1 V d
or
2

1 dI
2
P/L 2 dt V d
2 I

Return Characteristics
Implied volatility is a market forecast of the future
volatility of the underlying stock (level-dependent)

Long volatility positions gain when realized volatility (over the

lifetime of the option) is higher than implied volatility at
inception

Changes in implied volatility affect the

MTM of the position

S&P500 Implied and Realized Volatilities

Are current levels of volatility justified?

Market volatility is at a historical low
We believe that these levels of volatility are associated with
the ``investment cycle in equities
The crash of 2000-2002 diminished the publics
appetite for equities
The volatility environment resembles
the early 1990s (somewhat)

0
09/10/2003

12-Dec-

20-Mar-2002

06/18/2001

9/20/2000

12/27/1999

04/01/99

07/07/98

10/08/97

01/14/97

04/19/96

07/26/1995

10/28/1994

02/02/1994

11-May-93

14-Aug-92

19-Nov-91

26-Feb-91

01-Jun-90

06-Sep-89

09-Dec-88

17-Mar-88

23-Jun-87

26-Sep-86

02-Jan-86

VIX: 30-day Implied Volatility index of S&P

100
VIX: Jan 1986- May 2004

1987 Crash: 160%

80

70

60

50

40

30

20

10

1/
2/
20
04
1/
16
/2
00
4
1/
30
/2
00
4
2/
13
/2
00
4
2/
27
/2
00
4
3/
12
/2
00
4
3/
26
/2
00
4
4/
9/
20
04
4/
23
/2
00
4
5/
7/
20
04
5/
21
/2
00
4
6/
4/
20
04
6/
18
/2
00
4
7/
2/
20
04
7/
16
/2
00
4
7/
30
/2
00
4
8/
13
/2
00
4
8/
27
/2
00
4
9/
10
/2
00
4
9/
24
/2
00
4
10
/8
/2
00
10
4
/2
2/
20
04

This years VIX

VIX: Jan-Nov 2004

24

22

Madrid
Iraq handover
Unemployment

20

18

16

14

12

10

Can we forecast volatility?

Volatility cannot be predicted over long periods of time
Option markets in stocks and indices provide ``market
forecasts (market consensus) for periods ranging from 1
month to 1 year
Statistical models may be used to forecast volatility over
short periods of time
Importance of taking into account ``exogenous news and
events, macro news

Equity Volatility Markets

3000 optionable stocks in the U.S., 500 names have liquid
options

Index options (SPX, OEX) and ETF options (QQQ)

In Europe, liquid markets in Eurostoxx50
and DAX index options
Moderate liquidity in single-name options in Europe
Asia: Nikkei and Kospi200 are very active

Examples of trading strategies

Think of volatility trading as trading in insurance premiums (against large
moves in the underlying stock)
Analyze market in options as an ``insurance market via
normative models and build positions
Long-only strategies: always buy volatility to capitalize on market
dislocations. Typically proposed as an insurance overlay to other
investments

Long-short strategies: relative value approach. Capitalize on perceived

mispricing of options. Originated from market-making activities in dealing
rooms. Only recently adapted to asset-management
Macro strategies: invest across asset classes (equity, fx, fixed income,
currencies)

Long-Only Strategies

Strategies are natural hedges/overlays

Performed well before 2001

Performed less well after 2001 due to drop in levels
P/L profile: many small losses (time decay), occasional
large windfalls

Long-Short Strategies
Vega-neutrality to hedge changes in overall market volatility
Sell AND buy volatility (like insurance/reinsurance)
Adapted from dealing room (market-making) practice
Use correlation models for the underlying assets and their volatilities
Require strict control of exposure to single-name risk and market crash
risk

Index Option Arbitrage

(Dispersion Trading)
Sell index volatility, buy the volatility of the index components
Takes advantage of differences in implied volatilities of
index options and implied volatilities of individual stock
options

Main source of arbitrage: correlations between asset prices

vary with time due to corporate events, earnings, and
``macro shocks

Dispersion trading (graphical)

Index

Sell index call

Stock 1

Stock 2

Buy calls on different stocks.

Also, buy index/sell stocks

Practical aspects of vol trading

Market structure: specialist system in the US, OTC in Europe
Electronic versus voice trading
Reduction in trading costs and improved access allows hedge
funds to compete with B&Ds
Liquidity in single names and ETFs can be a problem
Technical infrastructure for trade processing, riskmanagement, models, etc

Worst-case scenarios for volatility trading

& remedies
Short volatility positions are such that underlying stock moves
considerably and implied volatility goes up

Long volatility positions are such that stocks move moderately

and volatility collapses
Limit exposure by asset
Limit sector exposure
Limit correlation exposure (dispersion trading)

Links between volatility trading and other

markets
Issuers of structured products (guaranteed principal with equity
upside) are sellers of long term equity volatility and correlation
Convertible bond issuers are sellers of equity volatility

Institutional investors tend to be long converts and banks

are short structured notes
Hedging these exposures creates flows in listed options
(Eurostoxx50, Dow, S&P) and basket trades (dispersion)
which connect long-term volatility markets with listed markets
Estimated exposure by issuers of structured products is EUR 200
MM per correlation point (RISK, May 2004)

Risk Arbitrage
If a company (A) is going to buy another one (B), it has to pay a premium
above the pre-announcement value. The stock of B will go up during
the time of the announcement until when the deal is done.

There is a chance that the deal does not go through, in which case stock B
drops in value.
Risk Arb is, by definition, a strategy which attempts to capitalize from these
effects on share prices caused by mergers.

Convertible arbitrage
The premise of the strategy is that the convertible bonds are is
sometimes priced inefficiently relative to the underlying stock, for
reasons that range from illiquidity to market psychology,( to the fact
that there are no bond buyers otherwise).
In particular, the equity option embedded in the convertible bond
(which is typically a 5 year option) may be a source of
cheap volatility, which convertible arbitrageurs can then exploit by
selling shares or selling listed options with shorter maturities.
The number of shares sold short usually reflects a delta-neutral or
market-neutral ratio. As a result, under normal market conditions,
the arbitrageur expects the combined position to be relatively
hedged.
However, maintaining a market-neutral position may require
rebalancing transactions, a process called dynamic delta hedging.
This rebalancing adds to the return of convertible arbitrage
strategies.

Global Macro
Global macro is the strategy of investing on a large scale around the world using economic
theory to justify the decision making process. Usually with currencies, commodities equity
Indexes or fixed income.
The strategy is typically based on forecasts and analysis about interest rates trends, the
general flow of funds, political changes, government policies, inter-government relations,
and other broad systemic factors.
Macro trader Yra Harris claims that "global macro" is really a new term, which used to be
called geopolitics. George Soros famously employed a global macro strategy when he
sold pound sterling in 1992 at the time of the European Rate Mechanism debacle.
In an discussion on global macro, hedge fund manager John Burbank discussed the
increasing importance and shift of private and institutional investors toward more global
macro strategies. Burbank defined global macro as "having a reason to be long or short
something that is bigger than a fundamental or quantitative stock view."

CTA, fundamental or systematic

A Commodity trading advisor (CTA) is an individual or organization who is retained
by a fund or individual client to provide advice and services related to trading
in futures contracts commodity options and/or swaps.

They are usually responsible for the trading within managed futures accounts. The
definition of CTA may also apply to investment advisors for hedge funds and
private funds including mutual funds and exchange-traded funds in certain cases]

A CTA generally acts as an asset manager, following a set of investment strategies

utilizing futures contracts and options on futures contracts on a wide variety of
physical goods such as agricultural products, forest products, metals, and energy,
plus derivative contracts on financial instruments such as indices, bonds, and
currencies.[
The trading programs employed by CTAs can be characterized by their market
strategy, whether trend following or market neutral, and the market segment,
such as financial, agricultural or currency.
There are two major styles of investment employed by CTAs: technical and
fundamental.

Fixed-income/credit arbitrage
Fixed-income arbitrage consists of the discovery and exploitation of inefficiencies in
the pricing of bonds, i.e. debt instruments from either public or private issuers, yielding
a contractually fixed stream of income.
In pursuit of their goal of both steady returns and low volatility, managers can focus
upon interest rate swaps, US non-US government bond arbitrage, see US Treasury
securities, forward yield curves, and/or mortgage-backed securities.
Synthetic instruments such as Credit Default Swaps (CDS) and indexes of CDS, can
be used in fixed-income/credit arbitrage.

Long-short equity funds

Typically, equity long/short investing is based on "bottom up" fundamental analysis of
the individual companies in which investments are made. There may also be "top down"
analysis of the risks and opportunities offered by industries, sectors, countries, and the
macroeconomic situation.

Long/short covers a wide variety of strategies. There are generalists, and managers who
focus on certain industries and sectors or certain regions. Managers may specialize in a
category for example, large cap or small cap, value or growth. There are many trading
styles, with frequent or dynamic traders and some longer-term investors.
A fund manager typically attempts to reduce volatility by either diversifying or hedging
positions across individual regions, industries, sectors and market capitalization bands
and hedging against un-diversifiable risk such as market risk.

There is wide variation in the degree to which managers prioritize seeking high returns,
which may involve concentrated and leveraged portfolios, and seeking low volatility,
which involves more diversification and hedging.

Statistical Arbitrage, Enhanced

Indexing and 130/30

Summary
-- Overview of Statistical Arbitrage
-- Using ETFs as relative-value indicators
-- Mean-reversion & the Ornstein-Uhlenbeck process
-- Examples of trades
-- Portfolio construction: a hypothetical fund (PLATA)
-- Synthetic 130/30 funds constructed from PLATA and Indices
-- Comparison of PLATA+SP with 130/30 Mutual Funds and a 130/30 ETF
-- Conclusions

Statistical Arbitrage
Trading strategy consisting of investing in stocks, both long and short, with
a mean-reversion theme

-- Pairs trading: practiced since the 1990s, invented by Morgan Stanleys prop team,
the precursor of stat-arb
-- Factor neutral trading: generalizes pairs trading using APT-like factors
-- ETF relative-value trading: the ultimate subject of this presentation (ETFs as factors)

Idea: measure statistically a time-series of spread between two or more correlated

stocks or indices and invest in overbought/undersold spreads.

Trading Universe:
Stocks of more than 1BB cap
unit: 1M/usd
Min

ETF

Num of Stocks

Internet

HHH

22

10,350

104,500

1,047

Real Estate

IYR

87

4,789

47,030

1,059

Transportation

IYT

46

4,575

49,910

1,089

Oil Exploration

OIH

42

7,059

71,660

1,010

Regional Banks

RKH

69

23,080

271,500

1,037

Retail

RTH

60

13,290

198,200

1,022

Semiconductors

SMH

55

7,303

117,300

1,033

Utilities

UTH

75

7,320

41,890

1,049

Energy

XLE

75

17,800

432,200

1,035

Financial

XLF

210

9,960

187,600

1,000

Industrial

XLI

141

10,770

391,400

1,034

Technology

XLK

158

12,750

293,500

1,008

Consumer Staples

XLP

61

17,730

204,500

1,016

Healthcare

XLV

109

14,390

192,500

1,025

Consumer discretionary

XLY

207

8,204

104,500

1,007

1417

11,291

432,200

1,000
January, 2007

Total

Average

Market Cap
Max

Sector

Investing: a quant perspective

Separate the systematic components of stock returns from the
company-specific, or idiosyncratic components

Ri i RMkt i
Ri

j 1

ij

Fj i

Project returns on single

Market Factor (CAPM)

Project returns on Multiple

(sector, size) Factors (APT)

In principle, market-neutral portfolios should have no

exposure to market factors (defactoring)

Defactoring: the Correlation Matrix Approach

Rit daily stock returns in panel form
i 1,..., N , t 1,..., T
2

1
Ri Rit ,
T i 1

Rit Ri

T 1 i

Rit Ri R jt R j
1
ij

T 1 i
i j

Principal Component Analysis

1 2 3 ... N

V ( j ) v1( j ) , v2( j ) ,..., vN( j )

Eigenvalues are all non-negative

Orthogonal eigenvectors

Stock market fluctuations can be characterized as moves along

the eigenvector directions. We seek to extract mathematical factors
from the PCA analysis.

PCA: Explained variance from the viewpoint of

eigenvalues
25.0%

20.0%

15.0%
Significant eigenvalues
10.0%

5.0%
Noise eigenvalues
0.0%

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50

Big universe: Jan 2007-Dec 2007

Nasdaq-100
Components of NDX/QQQQ
Data: Jan 30, 2007 to Jan 23, 2009
502 dates, 501 periods
99 Stocks (1 removed) MNST (Monster.com), now listed in NYSE
50%
45%
40%
35%
30%
25%
20%
15%
10%
5%
0%

10

13

16

19

22

25

28

31 34

37

40

43

46

49 52

55

58

61 64

67

70

73

76

79

82

85

88

91 94

97

Density of States (from previous data)

20
F(E)

15
10

E
One large mass at 0.44,
Some masses near 0.025
Nearly continuous density for lower levels

0.97

0.93

0.9

0.86

0.82

0.79

0.75

0.72

0.68

0.64

0.61

0.57

0.54

0.50

0.46

0.43

0.39

0.36

0.32

0.28

0.25

0.21

0.18

0.14

0.10

0.07

0.03

Zoom of the DOS for low eigenvalues

18
16
14
12
10
8
6
4
2
0

Marcenko-Pastur
Random Matrix DOE

Edge of DOS

3.40%

3.20%

3.00%

2.80%

2.60%

2.40%

2.20%

2.00%

1.80%

1.60%

1.40%

1.20%

1.00%

0.80%

0.60%

0.40%

0.20%

0.00%

Significant EVs

Factors & Eigenportfolios

For each eigenvector, build a portfolio which is weighted proportionally
to the coefficient of each stock and inversely proportionally to its volatility

Qi( j )

vi( j )

Portfolio weight of j-th eigenportfolio

( j)

vi
( j)
Ri
F j Qi Ri

i 1
i 1
i
N

J-th factor is the return of the

j-th eigenportfolio

First eigenportfolio returns compared with S&P 500

returns (1/5/2009 to 1/29/2010)

Y= 0.9856 * X + 0.0223
R_squared=0.9716

How many eigenportfolios are significant?

- Perform PCA on the empirical correlation matrix with 1 year moving

window
- Consider the correlation matrix of the residuals after removing 1, 2 ,3
eigenportfolios
- Compare the DOS of the correlation of the residuals with the spectrum of the
correlation matrix of pure noise (Marcenko-Pastur)
-The number of significant factors corresponds to the first m for which the matrix
of residuals is close to MP (e.g. in the sense of hypothesis testing)

Eigenvalues of the correlation matrix of

residuals (m=1)
1 33.59
1 33.59

6.7%
N
500

Recall that lambda_1 for the original correlation

matrix was ~ 220, so the residuals matrix has
``smaller correlations.
Ratio lambda/N is a proxy for the average
correlation.

Density of States, or Histogram, of Eigenvalues

for m=1 (correlation of returns with first factor removed)

Lambda_1 = 33.59
A lot of detached
evs suggests that
there are additional
factors.

Top 100 eigenvalues of the correlation matrix of

residuals (m=15)

1 8.91
1 8.91

1.8%
N 500

Marcenko Pastur compared to data with m=30

1 7.75
1 7.75

1.6%
N 500

Number of factors explaining 55% variance vs. VIX

(2002-2008)

Number of significant EVs versus VIX (1/2006-2/2010)

subprime

Lehman, AIG, etc.

Number of significant factors vs. first eigenvalue

of correlation matrix

Using ETFs as factors and relative value indicators

( e.g. trade EBAY versus QQQQ)
Idea: rather than decomposing risk into abstract factors, consider a stock and
the sector ETF corresponding to its industry
Pro: the relative analysis of a stock compared to its peers is done comparing
with a tradable index. We deal with one stock-specific risk at a time, not 2 or more.
Pro: this is better than using an abstract factor from PCA because the latter are
more likely to be unstable
Pro: the complexity of the signal generation is significantly reduced and
the origin of excess returns is simple to explain
Con: some large cap stocks track too closely the industry, so no signals are
available. Equivalently, the strategy can be viewed as somewhat biased to small caps
Con: Some stocks do not fit well the ETF identikit. We leave those
to fundamental analysts.

Defactoring with ETFs as factors (19 ETFs)

(Red line=Marcenko-Pastur, universe =S&P 500)

Lambda_max=11.9613
lambda/500= 2.39%

Modeling the Evolution of Stock Residuals

dSi t
dI t
i
i t
Si t
I t

Stock returns a sum of a multiple

of an ETF return and a residual process

i t i dt dX i t

Residual= drift component (expected

excess return above mkt.) + increment
of a stationary process

dX i t i mi X i t dt i dWi t

Ornstein-Ulembeck
AR-1 process

Statistical Estimation Window=3 months (~ 60 business days) to

include at least one earnings announcement

Statistics on the Estimated OU Parameters

ETF
HHH
IYR
IYT
RKH
RTH
SMH
UTH
XLF
XLI
XLK
XLP
XLV
XLY
Total

Abs(Alpha)
0.20%
0.11%
0.18%
0.10%
0.17%
0.19%
0.09%
0.11%
0.15%
0.17%
0.12%
0.14%
0.16%
0.15%

Beta
0.69
0.90
0.97
0.98
1.02
1.01
0.81
0.83
1.15
1.03
1.01
1.05
1.03
0.96

Kappa
38
39
41
39
39
40
42
42
42
42
42
38
39
40

Reversion days
7
6
6
6
6
6
6
6
6
6
6
7
6
6

EquiVol
4%
2%
4%
2%
3%
4%
2%
2%
3%
3%
2%
3%
3%
3%

Average over 2006-2007

Abs(m)
3.3%
1.8%
3.0%
1.7%
2.7%
3.2%
1.4%
1.8%
2.4%
2.7%
2.0%
2.5%
2.5%
2.4%

Trading Signals
We introduce an s-score for each stock:

si (t )

X i t mi

eq ,i

Open long position if

si 1.25

Open short position if si 1.25

Close long position if

si 0.50

Close short position if si 0.50

Schematic view of mean-reversion trading

S_i
1.25

Sell stock here & hedge

Open short

0.50
Buy stock here & hedge

-0.50

-1.25
Open long

Stock-ETF trading example : ACGL/XLF trades in 2009

Arch Capital Group Ltd. Market cap: 4.3 BB
ticker

trade_date

close_date

days

stock_PNL

etf_PNL

Total_PNL

etf_ticker

size

ACGL

20090206

20090212

3,598.79

2,038.35

5,637.14

XLF

75,579.64

ACGL

20090226

20090302

(1,416.04)

4,036.11

2,620.08

XLF

78,906.91

ACGL

20090309

20090319

15,016.04

(11,763.40)

3,252.64

XLF

78,469.98

ACGL

20090420

20090428

4,643.71

(1,046.79)

3,596.92

XLF

79,350.04

ACGL

20090505

20090512

3,989.70

(1,289.76)

2,699.94

XLF

79,815.06

ACGL

20090519

20090602

2,257.45

24.34

2,281.79

XLF

79,201.01

ACGL

20090608

20090618

(1,784.06)

1,563.52

(220.54)

XLF

79,210.95

ACGL

20090806

20090812

955.02

0.00

955.02

XLF

81,552.52

stock_cost

stock_shares

stock_exe_price

etf_cost etf_shares

etf_exe_price

status

open_s_score

close_s_score

60.13

1,257.00

62.99

9.83

(2,811.52)

9.10

1.00

(1.83)

(0.08)

53.94

1,463.00

52.97

8.18

(3,603.67)

7.06

1.00

(1.30)

(0.49)

45.10

1,740.00

53.73

6.28

(4,801.39)

8.73

1.00

(1.77)

(0.33)

54.02

1,469.00

57.18

10.09

(2,754.71)

10.47

1.00

(1.29)

0.94

56.85

1,404.00

59.69

11.66

(2,528.93)

12.17

1.00

(1.96)

(0.37)

58.28

1,359.00

59.94

12.23

(2,433.83)

12.22

1.00

(1.68)

(0.12)

58.03

1,365.00

56.72

12.44

(2,605.86)

11.84

1.00

(1.69)

(0.87)

61.97

1,316.00

62.70

14.01

(2,293.69)

14.01

1.00

(1.25)

(0.01)

Stock-ETF trading example : EBAY/QQQQ trades in 2009

eBay Inc. Market cap= USD 38 BB
ticker

trade_date close_date

days

stock_PNL

Total_PNL

etf_ticker

size

(2,447.11)

(6,331.81)

QQQQ

82,152.96

EBAY

10/9/2009

11/9/2009

EBAY

5/26/2009

6/8/2009

497.31

(5,278.60)

(4,781.29)

QQQQ

79,388.76

EBAY

5/13/2009

5/14/2009

2,973.75

(1,391.58)

1,582.17

QQQQ

79,023.75

EBAY

4/16/2009

4/20/2009

1 (1,981.80)

3,673.72

1,691.92

QQQQ

79,547.25

EBAY

9/23/2009

10/7/2009

1,304.16

1,409.00

2,713.16

QQQQ

82,127.76

EBAY

9/8/2009

9/10/2009

6,439.61

(2,043.87)

4,395.73

QQQQ

81,613.02

stock_cost

16 (3,884.70)

etf_PNL

stock_sha
stock_exe_price etf_cost
res

etf_shares

etf_exe_price

open_s_score close_s_score

beta

24.32

3378

23.17

42.43

(2,562.42)

43.39

(2.24)

(0.89)

0.756

17.56

4521

17.67

34.69

(2,559.32)

36.75

(1.84)

(1.00)

0.894

16.21

4875

16.82

33.00

(2,676.12)

33.52

(1.53)

(0.41)

0.895

14.45

5505

14.09

33.35

(3,251.08)

32.22

(1.27)

(0.43)

0.734

23.93

3432

24.31

42.58

(2,471.93)

42.01

(2.35)

0.00

0.780

21.81

3742

23.53

40.62

(2,579.02)

41.41

(2.26)

0.13

0.779

Trading Time vs. Actual Time

Statistics on equity returns (residuals) can be done
-- in actual time (% change/day)

-- in trading time ( % change per share/day)

Trading time incorporates volume information.

In trading-time framework, mean-reverting signals (S-Scores) are
-- weaker when volume is heavy

-- stronger when volume is light

Trading Time vs. Actual Time, II

Using the daily trading volume, construct a residual process which measures
the change in price per share

S
I

S
I

(usual residual)

V
,
V

V daily volume

V average V

Yt i
i 1

dY m Y dt dW

Estimate AR-1 / OU process for

the new process Y(t)

This makes deviations on unusually high-volume more likely, so the signal is weaker

Amazon.com Jan 2007-Oct 2008:

Avoiding short-selling on large volume
120

120000000

100

100000000

80

80000000

60

60000000

40

40000000

20

20000000

Building a portfolio from ETF-based signals:

the ``PLATA strategy
-- Large, diversified trading universe of equities (~ 500 names)
-- Select those stocks within the trading universe that have a
trading signal (s-score) and open trades
-- All trades consist of stocks paired with ETFs
-- Monitor for closing trades through s-score

-- Monitor for degradation of statistical parameters, stop-losses, etc.

-- Investment per stock ~ 25 bps (~250K per 100MM notional capital)
-- Typical profile 30 to 50 % long / 30 to 50 % short, dollar-neutral.
-- Portfolio-level risk management used to ``vet trades.

Portfolio Risk-Management
Qi i 1,..., N

holdings in stock or ETF i.

N

V Q1 , Q2 ,..., QN Cij Qi Q j

daily portfolio variance

Q Q1 , Q2 ,..., QN

position at date T

Q' Q1' , Q2' ,..., QN'

target (desired) position date T 1

ij 1

C: covariance
matrix estimated
with 3m window

Risk - control algorithm : let denote a risk threshold (e.g. 25bps of daily stdev)
If V Q' then only execute `closing trades' Q Q'
If V Q' then execute all trades Q Q'

Long and Short positions for unmanaged portfolio

( % of account equity )

Long/short balances with portfolio riskmanagement

Comparison between Long balances:

managed & unmanaged simulations

Basic PLATA : 0.25% per stock position

Leverage: 20/20 to 40/40, target daily vol=25 bps
Money market (Fed Funds)

Trailing annualized volatility without portfolio

risk management

Volatility of PLATA with and without risk

limit

Difference between managed risk and

unmanaged risk in the Fall of 2008

PLATA monthly performance

since September 2002
82 positive months out of 91
75 months above 10 bps

PLATA Monthly Statistics

Mean

0.50%

Standard Error

0.05%

Median

0.46%

Mode

NA

Standard Deviation

0.48%

Sample Variance

0.00%

Kurtosis

3.01

Skewness

0.75

Range

3%

Minimum

-1%

Maximum

2%

Sum
Count

0.46
91

Largest(5)

1.12%

Smallest(5)

-0.25%

Confidence Level(99.0%)

0.13%

Volatility and Stat Arb (Plata)

PLATA works better in the aftermath of volatility spikes and less

well when volatility drops. It is therefore reasonable to blend it
with an index strategy

Statistical Arbitrage and 130/30

Indexers: mutual fund managers and long-only managers
Objective: Track (or beat) returns of the overall market or sector
Underlying theory: CAPM, etc.
130/30 Managers: Long 130%, short 30% with periodic revisions of the portfolio
Essentially, a beta strategy with stock picking

Market-neutral managers: Seek returns that are uncorrelated

with the market (alpha)
Long-Short Equity MN: Fundamental Stock Picking with shorting
Statistical Arbitrage: Quantitative long-short MN

SPY+PLATA: a synthetic 130/30 fund

Based on a notional amount of 100 MM:
-- go long 100 MM SPY and
-- enter into a PLATA strategy based on 100MM notional amount
(30 to 50 mm long/ 30 to 50 mm short)
(parameters for PLATA: big universe, 25bps per stock,
target daily stdev of portfolio=25bps)

Due to market-neutrality of PLATA, this portfolio looks essentially like

a 130/30 to a 150/50 depending on the volatility in the market and the
turnover.
-- Assume fee structure: ~ 1.20% per year (average for quant mutual funds)

Comparing SPY+PLATA with

SPY

Comparing PLATA + SPY to enhanced indexing mutual funds

MainStay 130 30 Core C (MYCCX)

The investment seeks long-term growth of capital, with income as a secondary
consideration. The fund primarily invests in common stocks of well-established U.S.
companies, primarily those with large-capitalizations that are similar to companies
in the Russell 1000 index. It generally holds long positions, either directly or through
derivatives, equal to approximately 130%(within a range of 120% to 140%) of net
assets, and short positions, either directly or through derivatives, equal to
approximately 30% (within a range of 20% to 40%)of net assets.

Expense

MYCCX

Category Avg

Total Expense Ratio:

2.35%

1.12%

Max 12b1 Fee:

1.00%

N/A

Max Front End Sales Load:

Max Deferred Sales Load:
3 Yr Expense Projection*:

N/A
1.00%
1,027

5.28%
2.53%
596

5 Yr Expense Projection*:

1,741

894

10 Yr Expense Projection*:

3,631

1,787

AUM: 327 Million USD

Source: Yahoo!Finance

MainStay 130 30 Core C (MYCCX)

+ SPY

Fidelity Advisor Large Cap 130/30

FITOX, FOATX
The investment seeks long-term growth of capital. The fund
normally invests at least 80% of assets in common stocks of
companies with large market capitalizations. It normally
establishes long and short positions in equity securities. The
fund invests either in "growth" stocks or "value" stocks or
both. It may invest in domestic and foreign issuers.

Expense Ratio: 1.12 %., 1.56%

AUM: 26 MM,.

Source: Yahoo!Finance

Fidelity Advisor Large Cap 130/30

(FITOX, FOATX)

BNY Mellon US Core Equity 130/30

MUCIX
The investment seeks capital appreciation. The fund
normally invests at least 80% of net assets in equity
securities. It focuses on growth and value stocks of large cap
companies. Although the fund typically invests in seasoned
issuers, it may purchase securities of companies in initial
public offerings (IPOs) or shortly thereafter.

AUM: USD 222 Million

Fee: 2%

Source: Yahoo!Finance

BNY Mellon US Core Equity 130/30

(MUCIX)

+SPY

CSM: the first 130/30 ETF

-- Proshares launched the first 130/30 ETF in July 2009
-- Based on the 13030 Large Cap Index constructed by Andrew Lo (MIT) and
Panjak Patel (Credit Suisse)
-- Based on ranking stocks in S&P 500 according to 10 quantitative criteria
(Book to Value, Momentum, etc)
-- Monthly rebalancings

-- Data available does not include the credit crunch

The investment seeks investment results, before fees and
expenses, that track the performance of the Credit Suisse
130/30 Large Cap Index. The fund invests in a combination
of equity securities and derivatives that Proshares Advisors
believes should track the performance of the Index. It
invests typically the rest of the assets in money market
instruments. The fund is non-diversified.

AUM: USD 60 Million

Fee: 0.95%

Source: Yahoo!Finance

Comparison of SPY+ PLATA with CSM

since inception (7/2009)

Conclusions
Main properties of Stat Arb:
-- It provides a systematic way of picking stocks (long and short) based on
relative-value criteria
-- It operates in medium frequency with relatively low turnover costs, so it is
highly scalable
-- It can be leveraged an treated as a total-return strategy (benchmark=cash)
-- It can be merged with a long-index fund to create a systematic
enhanced-indexing fund or quantitative 130/30 (benchmark=S&P 500)
-- Stat arb makes money when the market has high cross-sectional volatility, which
is typically when indexing works less well. This is why Stat Arb and indexing can
be combined to make a superior yet simple investment product.
-- BUT: The devil is in the details and the investor community must understand how
the strategy works and its overall sensitivity to market volatility (e.g. VIX)

A systemic view of hedge funds

1988 2004: Steady growth of AUM

to 1000 Billion

LTCM
Long Term Capital Management, created by Solomon Brothers veterans
when Solomon Brothers was acquired by Citigroup (late 1990s)
AUM= 4Bn
Strategy: Fixed-income relative value (hedged)

Wall Street Firms (PBs) were financing LTCM at 25:1 leverage

In 1998, Long Term Capital blows up and is bailed out by the Federal Reserve Bank
of New York (sold the book for $1 to a consortium of Wall Street firms)
Increased public awareness of systemic risk of HFs
Disrepute of so-called ``convergence trades (but see MF Global 2012).
Liquidity problem

HFs take off in the 2000s

Hedge funds provide an alternative to classical investments
Appetite for HFs grows after the 2000 stock bubble
Regulation on HF in the US evolves considerably since 2005
Registration as HF was mandatory for a while (in 2006)
More regulation means higher barriers to entry
Entrance of Pensions (CALPERS) and College Endowments (Yale)
2002-2007 sees massive inflows to alternatives
Bear Stearns- and Soc. Gen-sponsored HFs blow up at the inception
of the Credit Crunch in Fall 2007

1997 2013 Hedge Funds AUM

Source: Barclay Hedge

A Market for Managers

With bad stock markets and low yields, the search for alternatives
is on.
Thesis: smart managers can outperform the market and produce
superior results
Antithesis: we cannot all be ``better than the market

This means that if all $$ went into alternatives, the returns should
look essentially like the market, or worse, due to fees and transaction
costs
Capacity: the natural size of AUM that a strategy can run

Returns

Annual Returns Mizuho HF Index

HF Returns vs. S&P 500 Returns

Equity Index outperforms HFs in Post Lehman period.

Law of diminishing returns

returns

Top managers

Stock mkt. return

Not so good managers

AUM
Thesis: The maximal size of the HF industry should be reached when
the marginal return on a $$ invested in HFs is no greater than the
return on standard returns (adjusting for taxes, etc).
M Avellaneda, P Besson, 2005, unpublished

AUM-weighted Return
t

Rt

a r
k 1
t

k k

a
k 1

ak assets under management in year k

rk returns of HF Index in year k (k 1 1988)

This justifies that if assets were constant then the return would be the average
HF return, but bigger assets have more weight.

Forecasting the size of HF industry

based on returns (in 2004)
Avellaneda and Besson, 2005

(hypothetical)

(forecast from regression)

HF line and stock market returns intercept at around 2000 Billion , which is roughly
the size of the HF industry in 2012!!!

HF in the post-internet bubble

Dodd-Franck and Volker Rule
Migration of sell-side traders to buy side
Supply of traders for HF, but trading a banks book is much different
than trading clients money
Different sources of capital
Different risk profiles
No hedging with other bank activities (retail, IB, flow)

Hedge Funds: Looking Ahead

Need for size. Emergence of giant funds 30+ bb
Size contradicts capacity limits
New strategies are necessary, otherwise the ``law of diminishing
returns caps the size of the industry
It is more and more difficult to find truly uncorrelated strategies
Arguably, commodities are no longer uncorrelated with the market,
since they ARE an alternative asset class (see post 2008)