Published: Journal of Banking and Finance, Vol. 82, Sept 2017: 180-190
Optimal Delta Hedging for Options
John Hull and *
. of Management University of Toronto
Copyright By PowCoder代写 加微信 powcoder
First version: September 9, 2015 This version: May 24, 2017
As has been pointed out by a number of researchers, the normally calculated delta does not minimize the variance of changes in the value of a trader’s position. This is because there is a non-zero correlation between movements in the price of the underlying asset and movements in the asset’s volatility. The minimum variance delta takes account of both price changes and the expected change in volatility conditional on a price change. This paper determines empirically a model for the minimum variance delta. We test the model using data on options on the S&P 500 and show that it is an improvement over stochastic volatility models, even when the latter are calibrated afresh each day for each option maturity. We also present results for options on the S&P 100, the Dow Jones, individual stocks, and commodity and interest-rate ETFs.
JEL Classification: G13
Key words: Options, delta, vega, stochastic volatility, minimum variance
*We thank , , , , , , Andrei Lyashenko, Curdy, , , , Managing Editor , and two anonymous reviewers, as well as seminar participants at University of Toronto, the Fields Institute, the 2015 RiskMinds International conference, a Bloomberg seminar in 2016, a Global Risk Institute seminar in 2016, and the Derivatives and Volatility conference at NYU Stern in 2017 for helpful comments. We are grateful to the Global Risk Institute in Financial Services for funding. Earlier versions of the paper were circulated with titles “Optimal Delta Hedging” and “Optimal Delta Hedging for Equity Options”
Optimal Delta Hedging for Options I. Introduction
The textbook approach to managing the risk in a portfolio of options involves specifying a valuation model and then calculating partial derivatives of the option prices with respect to the underlying stochastic variables. The most popular valuation models are those based on the assumptions made by Black and Scholes (1973) and Merton (1973). When hedge parameters are calculated from these models, the usual market practice is to set each option’s volatility parameter equal to its implied volatility. This is sometimes referred to as using the “practitioner Black-Scholes model.” The “practitioner Black-Scholes delta” for example is the partial derivative of the option price with respect to the underlying asset price with other variables, including the implied volatility, kept constant.
Delta is by far the most important hedge parameter and fortunately it is the one that can be most easily adjusted as it only requires a trade in the underlying asset. Ever since the birth of exchange-traded options markets in 1973, delta hedging has played a major role in the management of portfolios of options. Option traders adjust delta frequently, making it close to zero, by trading the underlying asset.
Even though the Black-Scholes-Merton model assumes volatility is constant, market participants usually calculate a “practitioner Black-Scholes vega” to measure and manage their volatility exposure. This vega is the partial derivative of the option price with respect to implied volatility with all other variables, including the asset price, kept constant.1 This approach, although not based on an internally consistent model, has the advantage of simplicity. The price of an option at any given time is, to a good approximation, a deterministic function of the underlying asset price and the implied volatility.2 A Taylor series expansion shows that the risks being taken can be assessed by monitoring the impact of changes in these two variables.
1 In a portfolio of options dependent on a particular asset, the options typically have different implied volatilities. The usual practice when vega is calculated is to calculate the portfolio vega as the sum of vegas of the individual options. This is equivalent to considering the impact of a parallel shift in the volatility surface.
2 This is exactly true if we ignore uncertainties relating to interest rates and dividends.
As is well known, there is a negative relationship between an equity price and its volatility. This was first shown by Black (1976) and Christie (1982) who used physical volatility estimates. Other authors have shown that it is true when implied volatility estimates are used. One explanation for the negative relation is leverage. As the equity price moves up (down), leverage decreases (increases) and as a result volatility decreases (increases). In an alternative hypothesis, known as the volatility feedback effect, the causality is the other way round. When there is an increase (decrease) in volatility, the required rate of return increases (decreases) causing the stock price to decline (increase). The two competing explanations have been explored by a number of authors including French et al (1987), Campbell and Hentschel (1992), Bekaert and Wu (2000), Bollerslev et al (2006), Hens and Steude (2009), and Hasanhodzic and Lo (2013). On balance, the empirical evidence appears to favor the volatility feedback effect.
A number of researchers have recognized that the negative relationship between an equity price and its volatility means that the practitioner Black-Scholes delta does not give the position in the underlying equity that minimizes the variance of the hedger’s position. The minimum variance (MV) delta hedge takes account of the impact of both a change in the underlying equity price and the expected change in volatility conditional on the change in the underlying equity price. Given that delta hedging is relatively straightforward, it is important that traders get as much mileage as possible from it. Switching from the practitioner Black-Scholes delta to the minimum variance delta is therefore a desirable objective. Indeed it has two advantages. First, it lowers the variance of daily changes in the value of the hedged position. Second, it lowers the residual vega exposure because part of vega exposure is handled by the position that is taken in the underlying asset.
A number of stochastic volatility models have been suggested in the literature. These include Hull and White (1987, 1988), Heston (1993), and Hagan et al (2002). A natural assumption might be that using a stochastic volatility model automatically improves delta. In fact, this is not the case if delta is calculated in the usual way, as the partial derivative of the option price with respect to the asset price. To calculate the MV delta, it is necessary to use the model to determine the expected change in the option price arising from both the change in the underlying asset and the associated expected change in its volatility.
A number of researchers have implemented stochastic volatility models and used the models’ assumptions to convert the usual delta to an MV delta. They have found that this produces an improvement in delta hedging performance, particularly for out-of-the-money options. The
researchers include Bakshi et al (1997) who implemented three different stochastic volatility models using data on call options on the S&P 500 between June 1988 and May 19913; Bakshi et al (2000), who looked at short and long-term options on the S&P 500 between September 1993 and August 1995; Alexander and Nogueira (2007), who looked at call options on the S&P 500 during a six month period in 2004; Alexander et al (2009), who consider the hedging performance of six different models using put and call options on the S&P 500 trading in 2007; and Poulsen et al (2009) who looked at data on S&P 500 options, Eurostoxx index options, and options on the U.S. dollar euro exchange rate during the 2004 to 2008 period. Bartlett (2006) shows how a minimum variance hedge can be used in conjunction with the SABR stochastic volatility model proposed by Hagen et al (2002).
This paper is different from the research just mentioned in that it is not based on a stochastic volatility model. It is similar in spirit to papers such as Crépey (2004), Vähämaa (2004) and Alexander et al (2012). These authors note that the minimum variance delta is the practitioner Black-Scholes delta plus the practitioner Black-Scholes vega times the partial derivative of the expected implied volatility with respect to the asset price. Improving delta therefore requires an assumption about the partial derivative of the expected implied volatility with respect to the asset price. Crépey (2004) and Vähämaa (2004) test setting the partial derivative equal to (or close to) the (negative) slope of the volatility smile, as suggested by the local volatility model.4 Alexander et al (2012) build on the research of Derman (1999) and test eight different models for the partial derivative, including a number of regime-switching models.
This paper extends previous research by determining empirically a model for the partial derivative of the expected implied volatility with respect to asset price. We show that, when the underlying asset is the S&P 500, this partial derivative is to a good approximation a quadratic function of the practitioner Black-Scholes delta of the option divided by the product of the asset price and the square root of the time to maturity. This leads to a simple model where the MV delta is calculated from the practitioner Black-Scholes delta, the practitioner Black-Scholes vega, the asset price, and the time to option maturity. We show that the hedging gain from approximating the MV delta in this way is better than that obtained using a stochastic volatility model or a local volatility model. The results have practical relevance to traders, many of whom
3 They also looked at puts on the S&P 500, but did not report the results as they were similar to calls.
4 See for example Derman et al (1995) and Coleman et al (2001). The local volatility model was suggested by Derman and Kani (1994) and Dupire (1994).
still base their decision making on output from the practitioner Black-Scholes model. The hedging gain from using our approach for options on other indices was similar to that for options on the S&P 500. The approach also led to a hedging gain for options on individual stocks and ETFs, but this was not as great as for options on indices.
The structure of the rest of the paper is as follows. We first discuss the nature of the data that we use. Second, we develop the theory that allows us to parameterize the evolution of the implied volatilities of options. The theory is then implemented and tested out-of-sample using options on the S&P 500. The results are compared with those from a stochastic volatility and a local volatility model. Based on the results for the S&P 500 we then carry out tests for options on other indices and for options on individual stocks and ETFs.
We used data from OptionMetrics. This is a convenient data source for our research. It provides daily prices for the underlying asset, closing bid and offer quotes for options, and hedge parameters based on the practitioner Black-Scholes model. We chose to consider options on the S&P 500, S&P 100, the Dow Jones Industrial Average of 30 stocks (DJIA), the individual stocks underlying the DJIA, and five ETFs. The assets underlying three of the ETFs are commodities, gold (GLD), silver (SLV) and oil (USO). The assets underlying the other two ETFs were the Barclays U.S. 20+ year Treasury Bond Index (TLT) and the Barclays U.S. 7-10 year Treasury Bond Index (IEF). The options on the S&P 500 and the DJIA are European. Both European and American options on the S&P 100 are included in our data set. Options on individual stocks and those on ETFs are American. The period covered by the data we used is January 2, 2004 to August 31, 2015 except for the commodity ETFs where data was first available in 2008.5
Only option quotes for which the bid price, offer price, implied volatility, delta, gamma, vega, and theta were available were retained. The option data set was sorted to produce observations for the same option on two successive trading days. For every pair of observations the data was normalized so that the underlying price on the first of the two days was one. Options with remaining lives less than 14 days were removed from the data set. Call options for which the
5 This is a much longer period than that used by other researchers except Alexander et al (2012).
practitioner Black-Scholes delta was less than 0.05 or greater than 0.95, and put options for which the practitioner Black-Scholes delta was less than –0.95 or greater than –0.05 were removed from the data set. For options on individual stocks, in addition to the filters used for options on the indices, days on which stock splits occurred were removed.
After all the filtering there remain more than 1.3 million price quotations for both puts and calls on the S&P 500, about 0.5 million observations for the other indices and ETFs, and about 200,000 observations for options on each individual stock in the Dow Jones Industrial Index. The trading volume for puts on the S&P 500 is much greater than that for calls.6 Puts and calls trade in approximately equal volumes for other indices. Calls trade more actively than puts for the individual stocks. Trading tends to be concentrated in close-to-the-money and out-of-the-money options. One notable feature is that the trading of close-to-the-money call options is particularly popular. The majority of trading is in options with maturities less than 91 days.
III. Background Theory
In the Black-Scholes model the underlying asset price follows a diffusion process with constant volatility. Many alternatives to Black-Scholes have been developed in an attempt to explain the option prices that are observed in practice. These involve stochastic volatility, jumps in the asset price or the volatility, risk aversion, and so on. Departures from Black-Scholes tend to reduce the performance of delta hedging. For example, Sepp (2012) shows that this is so for a mixed- jump diffusion model and some of the papers referenced earlier show that this is so for stochastic volatility models. In this section we provide a theoretical result for determining the minimum variance delta from the practitioner Black-Scholes delta. The result involves the implied volatility and is exactly true in the limit for diffusion processes while being an approximation in the case of other models.
6 The bid-offer spread for puts on the S&P 500 is smaller than that for calls except in the case of deep in-the-money options where the spreads are about the same.
Define S as a small change in an asset price and f as the corresponding change in the price of an option on the asset. The minimum variance delta, MV, is the value that minimizes the variance of 7
f MVS (1) We show in Appendix A that it is approximately true that
fBS fBS Eimp Eimp (2) MV S imp S BS BS S
where fBS is the Black-Scholes-Merton pricing function, imp is the implied volatility, BS is the practitioner Black-Scholes delta, BS is the practitioner Black-Scholes vega, and E(imp) is the expected value of the implied volatility as a function of S.
Other authors, in particular Alexander et al (2012), have explored the effectiveness of various estimates ∂E(imp)/∂S in determining the minimum variance delta. In what follows we estimate this function empirically and then conduct out-of-sample tests of the effectiveness of the estimated function.
When presenting our results, we shall define the effectiveness of a hedge as the percentage reduction in the sum of the squared residuals resulting from the hedge. We denote the Gain from an MV hedge as the percentage increase in the effectiveness of an MV hedge over the effectiveness of the practitioner Black-Scholes hedge. Thus:
Gain 1 SSEf MVS (3) SSEf BSS
where SSE denotes sum of squared errors.8
7 An early application of this type of hedging analysis to futures markets is Ederington (1979)
8 Using standard deviations rather than SSEs would produce a similar measure but the Gain would be numerically smaller
IV. Analysis of S&P 500 Options
In this section we examine the characteristics of the MV delta for options on the S&P 500 with the objective of determining the functional form of the MV delta. Once we have a candidate functional form, we will test it out of sample for both options on the S&P 500 and options on other assets.
We start with an implementation based on equation (1) applied to daily price changes:
f MVS (4)
where is an error term. Because the mean of S and f are both close to zero, minimizing the variance of in this equation, and other similar equations that we will test, is functionally equivalent to minimizing the sum of squared values. Several other variations on the model were tried such as using non-normalized data, replacing f with f – BSt, where BS is the practitioner Black-Scholes theta9 and t is one trading day, or including an intercept. None of the variations had a material effect on the results we present. The results that we report are for the model in (4).
We estimated equation (4) for options with different moneyness and time to maturity. Moneyness was measured by BS. We created nine different moneyness buckets by rounding BS to the nearest tenth and seven different option maturity buckets (14 to 30 days, 31 to 60 days, 61 to 91 days, 92 to 122 days, 123 to 182 days, 183 to 365 days, and more than 365 days). For each delta and each maturity bucket, the value of MV was estimated. In all cases MV – BS was negative. This result is consistent with the results of other researchers. It means that traders of S&P 500 index options should under-hedge call options and over-hedge put options relative to relative to the hedge suggested by the practitioner Black-Scholes model.10
9 The practitioner Black-Scholes theta is the partial derivative with respect to the passage of time with the volatility set equal to the implied volatility) and time is measured in days. If the asset price and its implied volatility do not change, the option price can be expected to decline by about BSt in one day.
10 A call has a positive delta and the MV delta, MV, is less positive than BS; a put has a negative delta and MVis more negative than BS.
The bucketed results show that MV – BS is not heavily dependent on option maturity and is roughly quadratic in BS. It is approximately true that11
BS S TGBS
for some function G where T is the time to the option maturity. From this equation, equation (2),
and the assumption of scale invariance12 we obtain
BS ab c2 (5)
MV BS ST BS BS
where a, b, and c are constants. Applying (5) to equation (2) shows that
ab c2 S Eimp BS BS
In the balance of the paper we examine the effectiveness of the approximation in equation (5) for
V. Out of Sample Tests of S&P 500 Options
To this point our work has been largely descriptive, motivated by a desire to produce a simple model of how the volatility surface for S&P 500 options evolves as a result of stock price changes. Our simple model is that for a particular moneyness and a particular stock price change, the expected size of the change in the implied volatility is inversely proportional to the square- root of the option life. For a particular option maturity and a particular percentage stock price change, the expected size of the change in the implied volatility is a quadratic function of our measure of moneyness, BS. The same model applies across the range of deltas considered.
11 ForEuropeanoptions, S TNd
程序代写 CS代考 加微信: powcoder QQ: 1823890830 Email: powcoder@163.com