Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
FIN285a: Problem Set 3 Fall 2004 Due Monday, October 11th 1. You have been observing a stock whose return you assume to be normal with mean 0.05 (5 percent), and standard deviation of 0.15. You currently have received two samples of 50 returns each. Your second set of returns has both a higher mean, and a higher median. They have both increased by 3 percent. What is the probability of observing both the mean and the median increase by 3 percent across these two samples? Estimate this with a monte-carlo experiment, drawing normal random numbers from the above distribution (null hypothesis) from the appropriate sample length. Perform this monte-carlo experiment for 100,000 iterations to estimate the probability of an increase in the mean of 3 percent or greater, an increase in the median of 3 percent or greater, and an increase in both of 3 percent or greater. Was it worth your while to use both numbers in your experiment? Program: % Create matrices s1, s2, mean_s and median_s to make program running faster. s1=zeros(1,50); s2=zeros(1,50); mean_s = zeros(100000,2); meadian_s = zeros(100000,2); %Run 100000 iterations and in each iteration draw 2 samples with 50 % elements in each. Calculate mean and median for each sample. for i=1:100000 s1 = normal(50,0.05,0.15^2); s2 = normal(50,0.05,0.15^2); mean_s(i,:) = [mean(s1) mean(s2)]; median_s(i,:) = [median(s1) median(s2)]; end test1 = proportion(mean_s(:,2)-mean_s(:,1)>=0.03) test2 = proportion(median_s(:,2)-median_s(:,1)>=0.03) test3 = proportion(mean_s(:,2)-mean_s(:,1)>=0.03 & median_s(:,2)-median_s(:,1)>=0.03) Results: test1 = 0.1574 test2 = 0.2086 test3 = 0.1149 2. You are monitoring the positions of FX traders at a large commercial bank. You know that the daily percentage changes in the value of each trader's portfolio is a normal random variable with mean zero, and standard deviation of 0.05. Your traders all start with a portfolio value of 100, and you will monitor them over the next 40 days (2 months). 1. Run a monte-carlo experiment of 100,000 interations, and report the mean, median, and percentiles 0.1, and 0.9 of the portfolio value at the end of 40 days. Plot a histogram too using 100 bins. (Note: You will probably find the matlab function prod() useful for this problem. prod gives you the product of a vector. If r is a return, then prod(1+r) is a useful object.) Program: endpos = zeros(100000,1); r = zeros(40,1); for i = 1:100000 r = normal(40, 0, 0.05^2); end_pos(i) = 100*prod(1+r); end m = mean(end_pos) med = median(end_pos) p_10 = percentile(end_pos, .1) p_90 = percentile(end_pos, .9) histogram(end_pos,100) Results: m =100.0329 med = 95.3683 p_10 = 63.3937 p_90 = 142.2701 2 2. Now you are continuously monitoring traders. You will stop all traders when their portfolios go below 80. Report the same stastics and histogram for this case. If a portfolio doesn't go below 80, you use the final value. If it drops below 80, you can assume that the trader was stopped at exactly 80. (Note: Here you will find the matlab function cumprod() useful. This gives a running product. We'll talk about this in class.) Program: endpos = zeros(100000,1); r = zeros(40,1); pos = zeros(40,1); for i = 1:100000 r = normal(40, 0, 0.05^2); pos = 100*cumprod(1+r); if any(pos<=80) end_pos(i) = 80; else end_pos(i) = pos(end); end end m = mean(end_pos) 3 med = median(end_pos) p_10 = percentile(end_pos, .1) p_90 = percentile(end_pos, .9) histogram(end_pos,100) Results: m = 101.2995 med = 85.7331 p_10 = 80 p_90 = 142.3144 4