How to code conducting tests to different time periods?

Hi!

Im trying to conduct a ttest and ftest on a dataset using a movingsum as an signal on the dataset, and I want the months used in the sum to go from 2-15 and test for each one of these periods. I started to duplicate the code for n=2 to fit n=3, but thought that there might be a more clever way to do this. Does anyone know of one? If you look at the code below, you should get the idea. NOTE; I also want to gather all the results from the tests in 2 matrices, so I can easily access these.

Any help would be greatly greatly appreciated as it would save me alot of time! Thanks in advance!

load austrial.txt % load the data matrix from the file

mkt = austrial(:,2); % market return in the second column

w = log(1 + mkt/100)
numElementsToSum = 2;
mSum = conv(w, ones(numElementsToSum, 1), 'valid');
movsum = mSum(1:end-1);
returns = mkt(numElementsToSum+1:end)
Buytwo=returns(movsum>0);
Selltwo=returns(movsum<0);

ExRetBuytwo = mean(Buytwo)
ExRetSelltwo = mean(Selltwo)
StdBuytwo = std(Buytwo)
StdSelltwo = std(Selltwo)

[h,ptwomean] = ttest2(Buytwo,Selltwo,0.05,'right','unequal');
[h,ptwovar] =  vartest2(Buytwo,Selltwo,0.05,'left');

numElementsToSum = 3;
mSum = conv(w, ones(numElementsToSum, 1), 'valid');
movsumthree = mSum(1:end-1);
returns = mkt(numElementsToSum+1:end)
Buythree=returns(movsumthree>0);
Sellthree=returns(movsumthree<0);

ExRetBuythree = mean(Buythree)
ExRetSellthree = mean(Sellthree)
StdBuythree = std(Buythree)
StdSellthree = std(Sellthree)

[h,pthreemean] = ttest2(Buytthree,Sellthree,0.05,'right','unequal');
[h,pthreevar] =  vartest2(Buythree,Sellthree,0.05,'left');

pmean= [ptwomean,pthreemean]
pvar= [ptwovar,pthreevar]

Answer 1

[Walter Roberson]

Guessing here about which values you wish to save:

nvals = 2:15;

for nidx = 1 : length(nvals)

    numElementsToSum = nvals(nidx);

    mSum = conv(w, ones(numElementsToSum, 1), 'valid');
    movsum = mSum(1:end-1);
    returns = mkt(numElementsToSum+1:end)
    Buytwo=returns(movsum>0);
    Selltwo=returns(movsum<0);

    ExRetBuytwo = mean(Buytwo)
    ExRetSelltwo = mean(Selltwo)
    StdBuytwo = std(Buytwo)
    StdSelltwo = std(Selltwo)

    [h, ptwomean(nidx)] = ttest2(Buytwo,Selltwo,0.05,'right','unequal');
    [h, ptwovar(nidx)] =  vartest2(Buytwo,Selltwo,0.05,'left');

end

plot(nvals, ptwomean, 'b', nvals, ptwovar, 'g:')

[Anders]

Thanks for the answer Walter! but I cant seem to get it working. When I implement your answer, I get the following;

mkt = austrial(:,2);
w = log(1 + mkt/100)
nvals = 2:15;
nidx = 1 : length(nvals);
    numElementsToSum = nvals(nidx);
    mSum = conv(w, ones(numElementsToSum, 1), 'valid');
    movsum = mSum(1:end-1);
    returns = mkt(numElementsToSum+1:end)
    Buy=returns(movsum>0);
    Sell=returns(movsum<0);
    ExRetBuy = mean(Buy);
    ExRetSell = mean(Sell);
    StdBuy = std(Buy);
    StdSell = std(Sell);
    [h, pmean(nidx)] = ttest2(Buy,Sell,0.05,'right','unequal');
    [h, pvar(nidx)] =  vartest2(Buy,Sell,0.05,'left');

plot(nvals, pmean, 'b', nvals, pvar, 'g:')

which gives me 14 values for both pmean and pvar as Im looking for, but they are all the same value. It just repeats the values from the tests from n=2. Is there something I have misunderstood with your answer?