Scribd
Upload
32 views

Volatility Modeling Using Daily Data

The document discusses quantitative finance techniques for modeling volatility using daily data. It begins with an introduction that outlines topics like modeling portfolio return distributions and forecasting dynamic portfolio variance. It then covers simple variance forecasting models like RiskMetrics that use an exponentially weighted average of past squared returns. The document also introduces the GARCH variance model and compares it to RiskMetrics. It discusses how GARCH captures features like volatility persistence and long-run volatility. The document analyzes how GARCH and RiskMetrics make different predictions for volatility, with GARCH allowing for more flexibility in modeling volatility over time.

Uploaded by

Vidaup40
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
32 views

Volatility Modeling Using Daily Data

The document discusses quantitative finance techniques for modeling volatility using daily data. It begins with an introduction that outlines topics like modeling portfolio return distributions and forecasting dynamic portfolio variance. It then covers simple variance forecasting models like RiskMetrics that use an exponentially weighted average of past squared returns. The document also introduces the GARCH variance model and compares it to RiskMetrics. It discusses how GARCH captures features like volatility persistence and long-run volatility. The document analyzes how GARCH and RiskMetrics make different predictions for volatility, with GARCH allowing for more flexibility in modeling volatility over time.

Uploaded by

Vidaup40
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 19

Quantitative Finance

Volatility Modeling using Daily Data

Jesús Ramirez

Universidad del Pacífico


Table of contents

1. Introduction

2. Simple Variance Forecasting

3. The GARCH Variance Model

1
Introduction
General Overview

• First, we will establish a framework for modeling the dynamic


distribution of portfolio returns.
• Second, we will use the model for forecasting dynamic portfolio
variance.
• Later, we will introduce methods for evaluating the performance
of these forecasts.
• Finally, we will consider ways to model non-normal features of
portfolio returns.

2
Overview

• We will discuss the simplest variance model, the RiskMetrics


model.
• We will introduce the GARCH variance model and compare it
with the RiskMetrics model.
• We will estimate the GARCH parameters using quasi-maximum
likelihood method.
• We will consider extensions to capture variance persistence and
leverage effects.
• We will talk about methods for evaluating the volatility
forecasting models.

3
Some facts about volatility

Although volatility is not directly observable, it has some common


characteristics:

• There exist volatility clusters, i.e. volatility is high in certain time


periods and low for other periods.
• Volatility evolves over time in a continuous fashion, i.e. volatility
jumps are rare.
• Volatility does not diverge to infinity, that is, volatility is often
stationary.
• Volatility seems to react differently to a large price increase or
to a large price drop,referred to as leverage effect.

4
Some facts about volatility

Figure 1: S&P 500 log returns


5
Simple Variance Forecasting
Assumptions

Main assumptions

• Let St be the daily closing price at t, then we will define the daily
asset log return as

Rt+1 = ln(St+1 /St )

• We will assume that


iid
Rt+1 = σt+1 zt+1 , with zt+1 ∼ N (0, 1)

• We need to establish a model for a time-varying variance to


know the entire distribution of the asset returns.

6
Simplest Model

• The variance, measured by the squared returns, shows strong


autocorrelation.
• A simplest way to capture this fact is by letting tomorrow’s
variance be the average of the recent observations.

1 ∑ 2 ∑ 1
m m
2
σt+1 = Rt+1−τ = R2t+1−τ
m τ =1 τ =1
m

• The forecast for tomorrow’s variance is immediately available at


the end of today when the daily return is realized.

7
Disadvantages

• An extreme return today will bump up variance by 1/m times the


squared return for m days after which variance will drop back
down.
• Autocorrelation of squared returns suggest a smoother decline
in the effects of past returns on today’s variance.
• It is unclear how we should choose m, which is a key parameter
in deciding the pattern of σt+1 .

8
RiskMetrics

• RiskMetrics1 model considers weights on past squared returns


that decay exponentially as we move backward in time.

RiskMetrics variance model

• The RiskMetrics variance model or exponential smoother is


written as
∞
∑
2
σt+1 = (1 − λ) λτ −1 R2t+1−τ , with 0 < λ < 1
τ =1

1 RiskMetrics is a system for market risk management developed by JP Morgan

9
RiskMetrics

RiskMetrics variance model

• It can show that the previous expression can be written as


2
σt+1 = λσt2 + (1 − λ)R2t

• The forecast for tomorrow’s volatility can be seen as a weighted


average of today’s volatility and today’s squared return.

10
Advantages

• Recent return matter more for tomorrow’s variance than distant


returns as λ < 1.
• The model only contains one parameter that need to be
estimated, namely λ.
• Relatively little data is needed in order to compute tomorrow’s
variance.

11
The GARCH Variance Model
Volatility Modeling

Simplest GARCH model


The simplest generalized autoregressive conditional heteroske-
dasticity (GARCH) model of dynamic variance can be written as
2
σt+1 = ω + αR2t + βσt2 , with 0 < α, β < 1 and α + β < 1

• What can we say about RiskMetrics and GARCH?

12
Long-run Volatility

Long-run volatility

• The unconditional or long-run average variance, σ 2 , is defined in


the previous GARCH as

σ 2 = ω/(1 − α − β)

• If we substitute the long-run variance into the dynamic variance


equation we have
2
σt+1 = σ 2 + α(R2t − σ 2 ) + β(σt2 − σ 2 )

13
Forecast

Forecasting volatility

• The one-day forecast of variance, Et σt+1


2 2
is given by σt+1 .
• We can prove that the k-days ahead forecast of variance is given
by
2
Et σt+k − σ 2 = (α + β)k−1 (σt+1
2
− σ2 )

• We will refer to α + β as the persistence of the model.

14
GARCH vs RiskMetrics

• High persistence implies that shocks that push variance away


from its long-run average will persist for a long time.
• When α + β is close to 1, GARCH and RiskMetrics might yield
similar predictions for short-horizons.
• However, in the RiskMetrics model a high variance day will
predict all the future days will be high variance.

15

You might also like