On 06/14/2012 08:47 PM, Robert Peirce wrote: > If I use a small amount (3 pairs) of dummy information, the intercept, > slope and R^2 are exactly what I expect them to be. If I use something > on the order of 400 pairs it makes no sense at all. When I use Linest, > the results are reasonable. > > I am evaluating stock trading systems against holding the market. The > data pairs are the monthly percentage change in the model vs. the > market. A regression produces the y-intercept (called alpha), the slope > (called beta) and R^2. > > Generally, if I know the total return for the model and the market the > calculation > > "total model return" = "total market return" * beta + alpha > or > y = mx + b > > is fairly close if I get the results from linest. They are way off from > intercept and slope. > > If the dependent variable is in A and the independent in B and there are > 400 pairs, I can use INTERCEPT|SLOPE(A1:A400; B1:B400). The results > aren't even close and I don't know why. Plugging the average returns > for the market into the equation to calculate the expected return for > the model produces nothing reasonable. > > Any ideas? I am probably missing something really obvious. > > What are your R^2 values? Near 1 indicates a high correlation between x and y.
-- Jay Lozier [email protected] -- For unsubscribe instructions e-mail to: [email protected] Problems? http://www.libreoffice.org/get-help/mailing-lists/how-to-unsubscribe/ Posting guidelines + more: http://wiki.documentfoundation.org/Netiquette List archive: http://listarchives.libreoffice.org/global/users/ All messages sent to this list will be publicly archived and cannot be deleted
