Fluorite | Level 6

## use model to explain

i am a novice at sas. Examine whether the portfolio alpha is still significant with more factors. The question: Which is the best model to explain these two highest/lowest alpha portfolio returns.  I have the data.  use four models, and which four models? How can i get the output like photo。 thanks all.

4 REPLIES 4
Diamond | Level 26

## Re: use model to explain

I don't understand anything about your screen capture from Excel. Can you please explain what we are seeing so that we can understand it?

Next, what does this have to do with SAS?

--
Paige Miller
Fluorite | Level 6

## Re: use model to explain

The question is which is the best model to explain these two highest/lowest alpha portfolio returns
from 2011 to 2015? (Note: you should judge the model performance by goodness-of-fit
and parameter significance, not by alpha alone. Since some of these models are nested,
try to use F-tests to compare them while you can.).

The question is about examine whether portfolio alpha is still significant with more factors. I have no idea about, someone told me the output is like that. Using four model to deal with it probably.

Diamond | Level 26

## Re: use model to explain

What are the numbers? What is the red cell?

--
Paige Miller
Super User

## Re: use model to explain

@victory wrote:

The question is which is the best model to explain these two highest/lowest alpha portfolio returns
from 2011 to 2015? (Note: you should judge the model performance by goodness-of-fit
and parameter significance, not by alpha alone. Since some of these models are nested,
try to use F-tests to compare them while you can.).

The question is about examine whether portfolio alpha is still significant with more factors. I have no idea about, someone told me the output is like that. Using four model to deal with it probably.

I don't see any "goodness-of-fit" statistics. Depending on the procedure you are running to create the "model" different fit statistics are often available. Where are the suggested F-tests?

I'm not even sure that I understand what your "alpha" may be. In many modeling discussions alpha is a set level of interest for a significance test, such as alpha=0.05. So typically alpha is not dependent on or the result of a model .

Discussion stats
• 4 replies
• 784 views
• 1 like
• 3 in conversation