* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Fin30233_F2016_Hedging and VAR with DeltaGamma
Survey
Document related concepts
Investment fund wikipedia , lookup
Present value wikipedia , lookup
Financialization wikipedia , lookup
Beta (finance) wikipedia , lookup
Modified Dietz method wikipedia , lookup
Business valuation wikipedia , lookup
Short (finance) wikipedia , lookup
Mark-to-market accounting wikipedia , lookup
Investment management wikipedia , lookup
Stock valuation wikipedia , lookup
Stock selection criterion wikipedia , lookup
Financial economics wikipedia , lookup
Modern portfolio theory wikipedia , lookup
Transcript
Hedging and Value-at-Risk (VaR) Run Single asset VaR Delta-VaR for portfolios Delta-Gamma VaR simulated VaR S. Mann November 2016 TCU’s Neeley School of Business Value at Risk (VaR) “VaR measures the worst expected loss over a given time interval under normal market conditions at a given confidence level.” - Jorion (1997) “Value at Risk is an estimate, with a given degree of confidence, of how much one can lose from one’s portfolio over a given time horizon.” - Wilmott (1998) “Value-at-Risk or VAR is a dollar measure of the minimum loss that would be expected over a period of time with a given probability” - Chance(1998) 95% confidence level VaR 5% probability minimum loss (over given horizon) max. loss with 95% confidence min.loss with 5% probability (for given time interval) Asset price standard deviation Assume lognormal returns: Let dS/S ~ lognormal(m,s) where s is annualized return volatility (standard deviation) The standard deviation of the asset price (S) over a period t is: S (s t.5) = S ( s t ) For example, let S = $ 100.00 s = 40% t = 1 week = 1/52 then the weekly standard deviation (s.d.) of the price is weekly s.d. = 100 (0.40) (.192).5 = 40(0.139) = $ 5.55 similarly, daily s.d. = 100(0.40)(1/252).5 = 40(0.063) = $ 2.52 monthly s.d. = 0.40(0.40)(1/12).5 = 40(0.289) = $ 11.55 Confidence levels and the inverse distribution function Let VaR = - S (st) N`(1 - confidence level) (for long position in underlying asset) where N`(x%) = inverse cumulative distribution function for the standard normal N`(x%) = number of standard deviations from the mean such that the probability of obtaining a lower outcome is x% example: desired confidence level is 95%, then N`(1-.95) = N`(5%) = - 1.65 In other words, N(-1.65) = 0.05 (5%) so N`(0.05) = -1.65 Run Standard normal distribution -1.65 -1 0 1 1.65 5% probability of return lower than 1.65 standard deviations below the mean Confidence levels and VaR Normal returns, $100 current price and 40% volatility: time period: sS(t) : C 99% 95% 90% 1-C 1% 5% 10% N`(1-C) -2.33 -1.65 -1.28 daily $2.52 weekly $5.55 monthly yearly $11.55 $40.00 daily VaR $ 5.87 4.14 3.23 weekly VaR $12.93 9.16 7.10 monthly yearly VaR VaR $26.91 $96.66 19.06 14.78 Let VaR = - S (st) N`(1 - C) Run Asset price in one week 5% probability Ignoring drift: assuming return is normally distributed with mean zero. For time frames longer than a week, may need to mean-adjust. $90.84 94.45 100 Weekly VaR = $9.16 105.55 109.16 Call greeks delta (D) = C/S = call sensitivity to asset value vba function: delta(S,K,T,r,s) gamma (G) = 2C/S2 = delta sensitivity to asset value vba function: gamma(S,K,T,r,s) vega = C/s = call sensitivity to volatility vba function: vega(S,K,T,r,s) theta = C/T = call sensitivity to time vba function: calltheta(S,K,T,r,s) rho = C/r = call sensitivity to riskless rate vba function: callrho(S,K,T,r,s) Put Delta and Gamma Put-Call parity: S + P = C + KB(0,T) take derivative of equality with respect to asset price, S. S/S + P/S = C/S + [KB(0,T)]/S 1 + P/S = C/S + 0 P/S = C/S - 1 Put Delta: Dp = Dc - 1 (put delta = call delta - 1) Take second derivative with respect to asset price, S. [P/S]/S = [C/S]/S + (1)/S 2P/S2 Put Gamma: Gp = = 2C/S2 + 0 Gc (put gamma = call gamma) Call VaR: Delta-VaR VaR for long position in underlying asset: VAR = -sS(t ) N`(1-C) Since D represents call exposure to underlying, (for small moves, if asset increases $1, call increases $D) define e.g. = -sDS(t ) N`(1-C), Delta-VaR Delta-VaR = DVaR example: $100 stock, 40% volatility, 95% daily VAR = $4.14 Examine 95% daily D-VaR for the following 110-day calls: Strike Value D 95% D-VaR (D-VaR)/Cost 95 100 105 110 11.96 9.36 7.21 5.47 0.66 0.57 0.48 0.40 2.72 2.35 1.99 1.64 22.8% 25.1% 27.6% 30.0% Delta-VaR for portfolios including options VaR for long position in underlying asset: VaR = - S (st) N`(1-C) and define Delta-VaR = -DS( st) N`(1-C), then define Portfolio D-VaR = - S ni DiS (s t) N`(1-C) for positions ni , {i = 1…N}, all written on the same underlying. This definition holds for puts, calls, and units of underlying. example: $100 stock, 40% volatility, 95% daily VaR = $4.14 Examine D-VaR for straddle: long put (np= 1) and long call (nc=1): call put total Strike Value D 100 100 9.36 8.04 17.40 0.57 -0.43 0.14 95% D-VaR 2.35 -1.79 0.56 (D-VaR)/Value 25.1% -22.3% 3.23% Short Call position Delta-hedged Portfolio D-VaR = (-S ni Di ) S(s t) N`(1-C) sell call: 115 strike, price = $14.00, implied volatility = 49% hedge: buy D shares, priced at $116.625 95% weekly DVaR: a) compute S(s t) N`(5%) = $13.04/share b) find position delta (SnD) (~11%) 95% weekly Position short call long shares total n -100 60 cost $-1,400 6,998 $ 5,609 D 0.593 1.000 95% weekly DVaR/cost = $9/5609 = 0.16% nD DVaR -59.32 $ -773.18 60.0 782.10 0.68 $ 8.92 Short Call Delta-hedged 95% weekly Position short call long shares total n -100 60 cost $-1,400 6,998 $ 5,609 D 0.593 1.000 nD DVaR -59.32 $ -773.18 60.0 782.10 0.68 $ 8.92 Stock 95% weekly VaR = $13.04. What if stock rises 1.65 st = $13.04 to new price of 129.67? If no time elapses, the position changes to: Position short call long shares total n -100 60 loss in value = $89.00 ! value $-2,260 7,780 $ 5,520 D 0.736 1.000 95% weekly nD DVaR -73.60 $ -1067.00 60.0 870.00 -13.60 $ -197.00 Adjusting D-VaR for nonlinearity Use Taylor-series expansion of function: given value at x0, use derivatives of function to approximate value of function at x: f(x) = f(x0) + f ' (x0) (x-x0) + (1/2) f '' (x0)(x-x0)2 + ... Incorporate curvature via Second-order Taylor series expansion of option price around the stock price (quadratic approximation): C(S + dS) = C(S) + (C/S) dS + (0.5) 2C/S2 (dS)2 = C(S) + D dS + 0.5 G (dS)2 so that C(S + dS) - C(S) = D dS + 0.5 G (dS)2 + … or dC = D dS + 0.5 G (dS)2 + … quadratic approximation of call price variance we want an expression for the standard deviation of an option’s price Quadratic approximation of change in call value for given change in stock value: dC = D dS + 0.5 G (dS)2 + … then the variance of the call’s price, V(dC) for given time, t , is: V(dC) = D2V(dS) + 0.5 [ G V(dS) ]2 note that V(dS) = S2s2 t = variance of stock price over period t so that V(dC) = D2 S2s2 t + 0.5 [ G S2s2 t ]2 thus the standard deviation of the option price is: s.d. (dC) = {D 2 S2s2 t + 0.5 [ G S2s2 t ]2 } (what if G = 0?) if G = 0 then s.d.(dC) = DSst (1/2) Delta-Gamma VaR Taylor expansion leads to quadratic approximation for the standard deviation of the option price: s.d. (dC) = { D2 S2s2 t + 0.5 [ G S2s2 t ]2 (1/2) } Note that if G = 0 then s.d.(dC) = DSst Examine Delta-VaR: D-VaR = D Sst N(1-C) delta-VaR is a linear operator on underlying asset VaR Delta-Gamma VaR = option s.d. N(1-C) (1/2) = [D2 S2s2 t + 0.5 [ G S2s2 t ]2] N(1-C) delta-Gamma VaR is nonlinear function of asset VaR Calculating portfolio Delta-Gamma VaR Delta-Gamma VaR = option s.d. N(1-C) (1/2) 2 2 2 2 2 2 = [D S s t + 0.5 [ G S s t ] ] N(1-C) Since Delta-Gamma VaR is nonlinear function of Delta and Gamma, cannot simply add VaR, as in Delta-VaR. For portfolio of linear and/or nonlinear derivatives on a single underlying: 1) compute portfolio Delta Dp 2) compute portfolio Gamma Gp Solution: 3) compute the quadratic approximation to the portfolio standard deviation [Dp 2 S2s2 t + 0.5 (Gp S2s2 t ] (1/2) 2 ) 4) multiply the portfolio standard deviation by N(1-C) to get D-G VaR Delta-Gamma VaR for delta-hedged short call Position short call long shares total n -100 60 cost $-1,400 6,998 $ 5,609 D 0.593 1.0 nD G nG -59.32 0.012 -1.2 60.0 0 0 0.68 -1.2 Find quadratic approximation to position standard deviation: (1/2) 2 2 2 2 2 2 [Dp S s t + 0.5 (Gp S s t) ] =[ (0.68)2(62.80) + .5 (-1.2 x 62.80)2 ] (1/2) =[ 0.4679 (62.80) + .5 (-78.03)2 ] (1/2) =[ 29.3875 + .5 ( 6088.81) ] (1/2) =[ 29.3875 + 3044.40 ](1/2) =[ 3073.79 ] (1/2) = 55.44 (using S = 116.625, s = 49%, t = 1 week, so that Sst = $7.92; S2s2 t = $62.80) Multiply the portfolio standard deviation by N(1-C) = -1.65 for C=95% Delta-Gamma VaR = 55.44 x (-1.65) = -91.19 Delta-Gamma VaR compared to Delta VaR Position n short call -100 long shares 60 total cost D $-1,400 0.593 6,998 1.000 $ 5,609 nD DVaR G -59.32 60.0 0.68 $ -773.18 -.012 782.10 0 $ 8.92 nG -1.2 0 -1.2 D-G VaR nm nm -91.19 Stock 95% weekly VaR = $13.04. What if stock rises 1.65 (Sst) = $13.04, to new price of 129.67? the position changes to: (assume no time lapse) Position short call long shares total n -100 60 value $-2,260 7,780 $ 5,520 D nD 0.736 1.000 -73.60 60.0 -13.60 DVaR $ -1067.00 870.00 $ -197.00 loss in value = $89.00 - a bit less than D-G VaR = 91.19 If position is unchanged, G moves to -0.9 and the new D-G VaR = $214.90 Alternative VaR approaches - Monte Carlo Monte Carlo simulate distributions of underlying assets; for each simulated outcome use pricing model to evaluate portfolio (underlying and options). Heavy computational requirements. Requires many inputs (e.g. variance, correlations) Assumes specific return-generating process (e.g. normal). Should use expected returns (not risk-neutral drift, as in monte carlo pricing). Alternative VaR approaches - Bootstrapping Bootstrapping (historical) database of return vectors (e.g., rt = r1t, r2t,…. rNt ) randomly sample from historical returns to generate return sequences - potential future scenarios based on historical data. All asset returns on given date are kept together - thus the bootstrap captures historical correlations between assets. Incorporates correlation, but not autocorrelation. Allows for non-normality. Data requirements are large. references Chance, Don, 1998. An Introduction to Derivatives. The Dryden Press. Jorion, Phillipe, 1997. Value at Risk: The New Benchmark for controlling market risk. Irwin Professional Publishing. Stulz, Rene, 1999. Derivatives, Financial Engineering, and Risk Management. South-Western College Publishing (in press). Wilmott, Paul, 1998. Derivatives - The Theory and Practice of Financial Engineering. John Wiley & Sons. (www.wilmott.com)