Some Financial Mathematics The one-period rate of return of an asset at time t. pt pt 1 Rt pt 1 where pt = the asset price at time t. Note: pt pt 1 1 Rt Also if there was continuous compounding during the period t – 1 to t at a rate rt then the value of the asset would be: pt pt 1e rt , thus rt ln 1 Rt rt = ln (1 + Rt) is called the log return. Note: If we have an initial capital of A0, an nominal interest rate of r with continuous compounding then the value of the capital at time T is: AT A0e rT The k-period rate of return of an asset at time t. pt pt k Rt k pt k again pt pt k 1 Rt (k ) Also pt pt 1 1 Rt pt 2 1 Rt 1 1 Rt pt k 1 Rt k 1 1 Rt k 2 1 Rt Thus the k-period rate of return of an asset at time t. 1 Rt (k ) 1 Rt k 1 1 Rt k 2 1 Rt If there was continuous compounding during the period t – k to t at a rate rt (k) then the value of the asset would be at time t: pt pt k e t r k k pt k 1 Rt (k ) pt k 1 Rt k 1 1 Rt k 2 pt k e rt k 1 rt k 2 e e pt k e rt 1 Rt rt k 1 rt k 2 rt Thus the k-period continuous compounding rate of return of an asset at time t is: 1 rt k rt k 1 rt k 2 k rt Taking t = k, let p0 = the value that an asset is purchased at time t = 0. Then the value of the asset of time t is: pt p0er1 r2 rt ln pt ln p0 r1 r2 rt If r1 , r2 , rt are independent identically distributed mean 0, then ln pt is a random walk. Some distributional properties of Rt and rt Ref: Analysis of Financial Time Series Ruey S. Tsay Conditional Heteroscedastic Models Models for asset price volatility • Volatility is an important factor in the trading of options (calls & puts) • A European call option is an option to buy an asset at a fixed price (strike price) on a given date (expiration date) • A European put option is an option to sell an asset at a fixed price (strike price) on a given date (expiration date) • If you can exercise the option prior to the expiration date it is an American option. Black-Scholes pricing formula ct = the cost of the call option, Pt = the current price, K = the strike price, l = time to expiration r = the risk-free interest rate st = the conditional standard deviation of the log return of the specified asset F(x) = the cumulative distribution function for the standard normal distribution Conditional Heteroscedastic Models for log returns {rt} Let Pt-1 denote the information available at time t–1 (i.e. all linear functions of { …, rt-3 , rt-2 , rt-1 }) Let m t = E [rt | Pt-1] and ut = rt – mt. s var rt Pt-1] = E[(rt – mt)2| Pt-1] 2 t Assume an “ARMA(p,q)” model, i.e. p p i 1 i 1 rt mt ut with mt i rt i i ut i The conditional heteroscedastic models are concerned with the evolution of s t2 var rt Pt-1] = E[(rt – mt)2| Pt-1] = var[ ut | Pt-1] The ARCH(m) model Auto-regressive conditional heteroscedastic model ut s t zt , s 0 u 2 t 2 1 t 1 u 2 m t m where {zt} are independent identically distributed (iid) variables with mean zero variance 1. with 0 0 and i 0 if i 0 The GARCH(m,s) model Generalized ARCH model m s i 1 j 1 ut s t zt , s t2 0 iut2i js t2 j where { zt} are independent identically distributed (iid) variables with mean zero variance 1. with 0 0,i 0 , j 0 max m, s and 1 i 1 i i Excel files illustrating ARCH and GARCH models CH models