Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: THEend8_
STAT8003 Assignment
Due: Monday Oct 23 2023 11:59pm, Please upload to Moodle You can write your answers by hand (then take a photo) Or type in Word (then convert to pdf). 1. Consider an ARMA(2,2) model: = 0.6 − 0.4−1 + 0.5−2 + − 0.9−1 + 0.8−2, where ~(0, 2) a. Is it stationary? Show your workings. b. Find out the MA representation of this ARMA(2,2) model. c. Is it invertible? Show your workings. d. Find out the AR representation of this ARMA(2,2) model. 2. Let Yt be the daily closing value of a trading pair: At – Bt. In other words, this trading pair consists of longing 1 share of A and shorting 1 share of B at time t. It is known that Yt is a stationary MA(1) process specified as follows: = + + 0.1−1, = 1, 2, … where µ = 5 and the noise term εt is i.i.d. N(0, 102). a. Express Y5 in terms of Y1, Y2, Y3, Y4, and ε5. Assume that ε0 = 0. b. The observed Yt’s for t = 1, 2, 3, 4 are listed in the following table: t 1 2 3 4 yt 7 18 2 -1 Given the data listed, compute E[Y5|Y1 = y1, …, Y4 = y4, ε0 = 0] and var[Y5|Y1 = y1, …, Y4 = y4, ε0 = 0]. 2 c. Suppose you hold the trading pair at the end of day 4 and you plan to unwind it (i.e., close out your positions on A and B) at the end of day 5. i. How many shares of A and B should you buy or sell at the end of day 5? ii. What are the expected profit and the standard deviation of the profit generated by this trading strategy? d. Since the trading strategy in part c. takes a position for only 1 day, it is a strategy with fixed investment horizon. However, if the position at the end of day 5 is still profitable for 1 more day, one can always decide not to unwind. Calculate the probability that you should not unwind your position at the end of day 5. 3. The sample autocorrelation function ̂() and sample partial autocorrelation function ̂() for time series {Zt} with a sample size of 150 are given as follows: k 1 2 3 4 5 6 7 8 ̂() 0.8 0.632 0.505 0.42 0.397 0.332 0.282 0.234 ̂() ? ? ? 0.045 0.133 -0.1 0.027 -0.011 We assume that {Zt} is stationary. a. Calculate the three question marks in the table above. b. Construct approximate 95% confidence intervals for the sample autocorrelation function ̂() and sample partial autocorrelation function ̂() in the table above. c. Based on part b., suggest an appropriate AR model for {Zt} and write down its equation. d. Estimate the parameters of your model in part c. by means of the method of moments (i.e., using sample statistics such as sample mean, sample variance, sample ACF, sample PACF, etc. to estimate model parameters). We have calculated that the sample mean and sample variance for {Zt} were 3.5 and 0.75 respectively. 3 4. Given the following AR(2) model: Zt = 0.3 Zt-1 - 0.5 Zt-2 + at, where {at} ~ WN(0, σa2) a. Show that the roots of {Zt} lie outside the unit circle and hence, {Zt} is stationary. b. Find the ACF (auto-correlation function) of {Zt}. 5. Given the following ARMA(1,1) model: 11 tttt eeYY , where {et} is WN(0, σe 2) a. Show that the variance of Yt is: 2 2 2 0 1 21 e b. Hence, find the ACF (autocorrelation function) 1,for kYtk . c. Suppose we estimated the following sample ACFs: r1 = 0.2, r2 = 0.1. Using these sample ACFs, recover the parameters and . Keep only the stationary and invertible solution. d. Suppose the sample variance s2 for Yt is calculated as 0.86, use parts a. and c. to find an estimate for σe2.