Can technical indicators based on underlying assets help to predict implied volatility index

• Published date: 01 January 2024
• Author: Shi Yafeng,Yanlong Shi,Ying Tingting
• Date: 01 January 2024
• DOI: http://doi.org/10.1002/fut.22464

Received: 12 April 2023

|

Accepted: 13 September 2023

DOI: 10.1002/fut.22464

RESEARCH ARTICLE

Can technical indicators based on underlying assets help

to predict implied volatility index

Shi Yafeng

1

|Yanlong Shi

2

|Ying Tingting

3

1

Department of Financial Technology, College of Finance and Information, Ningbo University of Finance and Economics, Ningbo,

Zhejiang Province, China

2

Department of Basic Courses, Zhejiang Pharmaceutical University, Ningbo, Zhejiang Province, China

3

Department of Finance, Accounting and Economics, Nottingham University Business School, University of Nottingham Ningbo China, Ningbo,

Zhejiang Province, China

Correspondence

Ying Tingting, Department of Finance,

Accounting and Economics, Nottingham

University Business School, University of

Nottingham Ningbo China, Ningbo,

Zhejiang Province, China.

Email: tingting.ying@nottingham.edu.cn

Funding information

MOE (Ministry of Education in China)

Liberal arts and Social Sciences

Foundation, Grant/Award Number:

20YJC790111

Abstract

Given the widespread use of technical analysis and the tight relationship

between derivatives and the underlying assets, we employ the copula

approach to investigate whether the technical indicators based on underlying

assets convey extra information about the future movements of implied

volatility (IV) indexes. The empirical results, based on long samples of five

well‐known IV indexes, suggest that although the technical indicators are not

informative for forecasting the future prices of IV indexes, they can provide

extra information about the size of forecasting errors of the IV indexes. The

findings are also robust to the impact of COVID‐19. The technical indicators

are then used to extend Threshold ARCH and Exponencial GARCH models

for improving the estimation of Value at Risks (VaRs). The out‐of‐sample

forecast results show that the proposed model outperforms the benchmark in

estimating the VaRs. These findings have implications for pricing options of

IV indexes and managing the risks of IV‐related portfolios.

KEYWORDS

copula, derivatives, implied volatility, tail dependence, technical indicator, VaR

1|INTRODUCTION

Technical indicators play a significant role in technical analysis, and become the focus of proponents. Therefore, there

are extensive empirical studies on the issue that whether these indicators indeed are helpful to predict future

movements of the stock market after the study of Fama and Blume (1966). This study addresses a similar question in

derivatives: Does the fluctuation of technical indicators based on underlying assets drive the movements of future

prices of corresponding derivatives?

A impetus for this study is the fact that technical analysis is ubiquitous among practitioners, although whether it

indeed results in significant profit is the subject of an ongoing debate. As trading mechanisms of these strategies

depend on a variety of technical indicators to generate buy and sell signals, the changes in technical indicators probably

can trigger new trades. The new transactions usually change the position of investors' portfolios and their exposures to

individual market variables, which are most likely to create a new demand for corresponding derivatives to hedge the

J Futures Markets. 2024;44:57–74. wileyonlinelibrary.com/journal/fut © 2023 Wiley Periodicals LLC.

|

57

risks. This implies that the technical indicators based on underlying assets convey potential information about future

movements in prices of the corresponding derivatives. The existing studies, however, are silent on how intensely these

indicators influence the prices of the derivatives, which are the focus of the vast literature on pricing the derivatives

(see Hull, 2015). Therefore, we attempt to fill in this gap.

The technical trading rules based on the moving average (MA) are the simplest and seemingly most popular in

practice (Zhu & Zhou, 2009), so MA‐based rules have been extensively explored in a considerable amount of literature,

see, Han et al. (2013) and Avramov et al. (2021). In a recent and remarkable work, Han et al. (2016) employed

normalized MA as the trend signal to construct trend factor, and provide three reasons for employing MA to construct

the trading signal. Although it is essentially same as the normalized MA defined by Han et al. (2016), the technical

indicator, price deviation from MA (PDMA), that was proposed by Shi et al. (2020a) to measure the extent to which the

price deviates from MA, can shed more intuitive light on reversal effects than the latter. Considering these, we focus

our attention on the PDMAs in this paper.

Options, as we know, are one of the important derivatives and play an important role on risk management. The

implied volatility (IV) index, calculated from a panel of option prices, is a measure of expected average variance for the

next 30 days, and a profile of movements in option market. An important IV index, volatility index (VIX) index, has

become the benchmark for volatility of stock market and is used as the “investor fear gauge”for financial market

practitioners. A recent empirical study by Shi et al. (2020a) found that the PDMAs can provide information for

forecasting volatilities of stock returns. Based on the B–S formula in Black and Scholes (1973), the volatility of

underlying assets is a key factor of the option premiums. This implies that there is a potential dependence structure

between the PDMAs based on underlying assets and future movements of IV indexes. Therefore, we investigate this

problem based on data from five well‐known IV indexes, including VIX, DJIA volatility index (VXD), NASDAQ

volatility index (VXN), HSI volatility index (VHSI), and KOSPI volatility index (KIX).

This paper contributes to accounting and finance literature in several ways. First, this study is the first academic

study to examine the effect of technical indicators derived from underlying market to the corresponding derivatives

market. It provides a new perspective to analyze the interrelationship between stock market and the corresponding

derivatives market, and opens the avenue for future research. Second, the evidence from the impact of the PDMAs on

estimated errors of IV indexes at extreme has implications for pricing third‐generation volatility products such as VIX

option, which is the focus of the vast literature (Liu et al., 2022; Mencia & Sentana, 2013; Wang & Daigler, 2011, etc.).

Finally, we propose a flexible model framework to comprise the information of technical indicators to improve the

estimation of the key risk measure: Value at Risk (VaR). This is beneficial for the investors to monitor the risk of the

portfolio involving IV indexes.

This paper is organized as follows. Section 2outlines the framework to investigate the average correlation and tail

dependence structure. The data analysis and the empirical results are presented in Section 3. We extend Exponencial

GARCH (EGARCH) and Threshold ARCH (TARCH) models to include the information of PDMAs for improving the

estimation of the VaRs in Section 4. The last section concludes the paper.

2|THE METHODOLOGY

2.1 |Technical indicator and implied volatility index

Suppose for generality we observe the underlying asset price

Pt T,=1,2,…,

.

t

The PDMA of the underlying asset with

given period of

n

trading days, Dn

t

,,

1

is defined as

DPMA=log( )−log( )

,

nt tnt,,

(1)

where



M

AP=

nt ni

nt

i

,1

=0

−1−is denoted as the moving average price with given period of

n

trading days. Similar to MA,

roles of PDMAs playing among investors vary with different time periods (

n

). Following the approach employed by Shi

et al. (2020a), the PDMAs can be classified as short‐term PDMAs (when n=5or10

), medium‐term PDMAs (when

1

In Han et al. (2016), the normalized MA is defined as MAMA P=

nt nt

t

,,

∕

∼

. Then it holds that DMA=−log(

)

nt nt,,

∼

, consequently demonstrates that the

PDMA is essentially same as the the normalized MA.

58

|

YAFENG ET AL.