# How exactly is time series regression useful to market participant?

Assume we run an OLS regression with dependent variable of 3-month holding period returns of a stock, and independent variable of 10 year treasury yields.

Assume regression coefficient is 2, so for every 1% point rise in 10 year yield, we expect a 2% rise the 3-month return of our stock.

Since the coefficient explains the relationship between past 3-month returns and treasury yields, how exactly does that provide any insight to a market participant who always holds for 3 months and then sells the stock? Or does it not provide any insight?

• In my opinion it does not provide any insight, perhaps I am missing something. Usually we look for predictive regressions: variable(s) known at time t that can predict stock returns between t and t+h. Commented Sep 5, 2021 at 14:31
• I’m not trying to predict anything. I want to explain the relationship between stock returns at each holding period and the monthly 10 year yield. So I want to be able to say “1% rise in monthly treasury yield means a 2% rise in 1-month holding period return;” “1% rise in monthly treasury yield means a 5% rise in 3-month holding period return,” etc. Does that make sense? I want to see the sensitivity of the returns, by holding period, to changes in monthly treasury yield.
– Hans
Commented Sep 5, 2021 at 15:22
• So I’m asking if my regression is a valid way of describing that sensitivity.
– Hans
Commented Sep 5, 2021 at 15:28
• Yes, that regression coefficient is your sensitivity that you are looking for. That coefficient is also called beta. Here is an article that walks through the beta calc and talks about some of its implications. investopedia.com/ask/answers/070615/… Commented Sep 5, 2021 at 17:04
• Thank you! And that beta makes sense if I use a 3-month return with the monthly treasury yield? I suppose one could use anything as independent variable (the interpretation would simply change).
– Hans
Commented Sep 5, 2021 at 19:07