Standard Error of the Forecast Formula

[topher]

Have you memorized this? sf^2= SEE^2[1 + 1/n + (X − Xbar)^2/(n - 1)sx^2]
Do we need to memorize this? I’ve been just using SEE instead of doing all that to get the exact sf.

 

[swaptiongamma]

Somtimes I do that too. Use the SEE instead of sf and the prediction interval is close enough to the answer.
But I have also memorized this formula, just in case when the going gets tough.

 

[eltia]

Yea, just memorize this together with the Adjusted R^2 equation. That’s pretty much the only two tricky equation to remember in Quant.

 

[mwvt9]

There was a really good shortcut for this formula last year. You may want to try to search it.

 

[TheAliMan]

r^2adj = (n-1)/(n-k-1) * (1- (1-r^2))
How did I do?

 

[eltia]

I believe the correct equation for Adjusted R^2 is
R^2_{Adj} = 1 - [(n-k-1)/(n-1)*(1-R^2)]

 

[mp2438]

isn’t it 1 - (1-r^2)*(n-1)/(n-k-1). the first “1” is not included in the parenthesis.
AliMan, in your equation, the last term (1-(1-r^2)) could be expressed as (1 - 1 + r^2) = r^2, which is incorrect.

 

[topher]

mp2438, you’re correct on the adjusted R^2.

 

[TheAliMan]

Thanks guys!

 

[topher]

http://www.analystforum.com/phorums/read.php?12,680993,681138#msg-681138
In reference to what mwvt9 said, which is basically saying use the SEE to calculate the confidence interval, and then look for an answer that is slightly wider than the interval you calculated.