how to match daily returns ?

Question

jean claude 2017-2-6

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/323652-how-to-match-daily-returns

回答： Rik 2017-2-9

hello everyone, i have the dependent variable with daily returns y (1 column ), in the other hand daily returns for the independant variable x (1 column of daily returns). How to pick daily returns from x corresponding to y ( so must have same date) because daily returns could be missing some days in x and assign nan for the missing days in such way to make linear regression!

2 个评论
显示无隐藏无

John D'Errico 2017-2-6

Huh? Perhaps you could give an example of what you want to do?

jean claude 2017-2-9

编辑：jean claude 2017-2-9

yep John, example y=[01012000 02012000 03012000, 100 200 300]; x=[02012000 05012000 07012000, 3 5 6]; so to regress y on x i would have x=[nan 3 nan] because in 2 january 2000 it matchs, then i can regress y=[100 200 300] on x=[nan 3 nan]. my problem is to construct variable x (i think unique could help)

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Rik 2017-2-9

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/323652-how-to-match-daily-returns#answer_254037

在 MATLAB Online 中打开

So you have two datasets that are sampled over time that you need to resample over a common/matched timeframe? Am I understanding you correctly?

I had something that sounded similar, but my data had different millisecond timestamps. I used interp1 to interpolate to a timestampvector, which would look something like this:

t=unique([y(1,:) x(1,:)]);
y2=interp1(y(1,:),y(2,:),t);
x2=interp1(x(1,:),x(2,:),t);
plot(x2,y2)

And then you can do all the fancy fitting you wish. You need to convert those dates into something that increments normally. I bet 2 second of google will result in a FEX submission that does that for you.