1

I'm trying to find a website which gives me R Squared value of a stock.

So far looked at at least 10 websites. I can find every analytics but R-Square!

I looked at Yahoo Finance, Reuters, fools.com ,...

1
  • You're talking about the R-Square of the stock price correct? Jan 23, 2018 at 17:57

1 Answer 1

3

R-Squared is a measure of a particular kind of performance of a regression analysis. It is highly dependent on the time period and data frequency of the regression in question, to say nothing of the variables used.

Therefore each stock has an R squared:

  • Of a particular model (such as the CAPM)
  • At a particular data frequency (such as monthly)
  • Over a particular length of data (10 years)
  • Chosen over a particular time (the last 10 years)

R-squared changes drastically and predictably based on how you choose these, unlike some other statistics. For example, the CAPM beta measured at the monthly frequency may be quite similar to the beta measured at the weekly or daily frequency (at least, it should be in the same ballpark theoretically), but the R squared values for each regression will be very different, always. The same stock will have a higher R squared value when using lower frequency and shorter time period data.

There is no standard time period or frequency used to measure R-squared, so it's not a statistic that is reported on websites. R-squared is normally only used when you are comparing two stocks side by side so you can be sure that you are using the same data frequency and period for both. For this reason R-squared is not, and should not be, a frequently reported statistic on websites like yahoo finance.

R-squared is also not usually a great measure of something an investor would care about, unlike volatility and beta, but I suppose that could be a matter of opinion.

======== Edit: Some Bonus Info ==========

It looks like a few words about what R-squared is useful for in a finance context are in order.

R squared in a CAPM regression tells us what percentage of the variation in our stock returns can be explained by variation in the market. We can think of a CAPM regression a decomposing the variation in our returns into the component explained by the market (market risk) and the component specific to the firm and not related to the market (idiosyncratic risk). R squared gives us an idea of what fraction of the total risk of the stock is market risk.

This means a high R-squared could be interpreted as the stock has relatively little firm-specific risk when compared with its total risk. We must be very clear that R-squared is a fraction, so it doesn't tell you which has more risk, only which has more of its risk in firm-specific events versus marketwide events. High R-squared means Beta is more important in explaining returns after the fact, not that it is more reliable. Of course, if the market and your stock are completely unrelated, then R-squared will be low, which is where the idea of reliability of beta comes from, but it's not the best measure of this idea.

Please note that the above interpretation is only reasonable when comparing two regressions over the same period and using the same frequency of data.

How do we know if beta is reliable? Beta will have a low standard error and small confidence interval if it is. Remember, a beta of 0 could be highly reliable...meaning we are quite confident the stock is not related to the market. R squared conflates these two concepts and throws in sensitivity to data length to boot.

4
  • Exactly what I was going to say, R-Square is dependent on so many different variables so it's impossible to get a standardized answer. The easiest things is to do the work yourself so you can compare your results. Jan 23, 2018 at 17:59
  • I have heard beta is only reliable when R-Square is high, That is why I'm looking for it. Every website I have seen so far provides Beta,
    – S Nash
    Jan 23, 2018 at 18:02
  • It seems to me that the R-squared could (and should) be given for any beta. i.e. there's an r-squared value for any given beta regression calculation. If you give a beta, you should give the corresponding r-squared. The question is why aren't websites doing this. If someone is telling me a result they got from a regression, the first thing I'm going to want to know is the r-squared for that regression.
    – JimmyJames
    Jan 23, 2018 at 18:27
  • R squared is a very poor measure of reliability because of the reasons I mentioned above. If you do a regression and then throw away half the data and repeat the regression, R squared will most likely tell you the regression with less data is better. If you want to know the reliability of beta, you can look at measures designed for this purpose, such as the standard errors. Folks who tell you high R-squared means beta is reliable have simplified the reality to such a degree that it is no longer truth. There is a nougat of truth there, but overall the statement is false.
    – farnsy
    Jan 23, 2018 at 19:14

Not the answer you're looking for? Browse other questions tagged .