1
$\begingroup$

I have two stock prices with a strong trend which are non-stationary.

I want to calculate the correlation between one stock say $x$ and another stock say $y$, but lagged 7 days back. Why should I have to decompose it? Can I just use the $x$ stock as an indicator of where the price of $y$ stock will be in 7 days?

$\endgroup$
6
  • $\begingroup$ I don't see why not. Just run a regression where stock X is the independent variable and stock Y (lagged 7 days) is the predictor. $\endgroup$ Oct 4, 2017 at 14:43
  • 1
    $\begingroup$ If both series are non-stationary your results are prone to be spurious, they're actually gonna reflect the relation between the deterministic/stochastic trend than the actual the data. It's real common sense in literature to use return instead of spot price. $\endgroup$ Oct 4, 2017 at 14:51
  • $\begingroup$ @lucasfariaslf, they're actually gonna reflect the relation between the deterministic/stochastic trend than the actual the data? I would disagree. The actual data is dominated by these trends, that is, the data basically is the trends themselves. So the real problem here is that we are not interested in the actual data but rather its increments (returns). $\endgroup$ Oct 5, 2017 at 5:22
  • 1
    $\begingroup$ But if I want to buy if they correlated and lagged let's say one week. If I observe the price to drop for stock x then I should short stock y lets say on day 6. Why Do I need returns for that? Since I don't use any beta or diversification why do I need to know returns? $\endgroup$
    – J.Ze
    Oct 5, 2017 at 10:27
  • $\begingroup$ @RichardHardy agreed, shouldn't have said data, did so considering that whichever existing non-stationary information doesn't help us understand the relationship between the stocks. $\endgroup$ Oct 5, 2017 at 12:30

0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.