Thread: GAMMADIST Function

I would appreciate any help I could get on this. I have a series of returns data (percentages) that I've calculated a mean, standard deviation, skewness, and kurtosis for. I want to chart the returns demonstrating the skewness and kurtosis. In other words, showing that the data is not a normal distribution with a shifted bell curve. I think the only way to do incorporate this is using the GAMMADIST function but I'm not too sure. Does anybody know how I apply the skewness and kurtosis, or if I can for that matter, to the return stream and chart it to show an appropriately shifted bell curve?

Thanks a bunch for any help. Hopefully I was able to lay this out in an understandable way.

Let me slim this down. Is there a way to create a bell chart with skewness and kurtosis in Excel?

Thanks for any help.

Andy,

Thanks for the response. What I actually need is that bell chart to take into account skewness and kurtosis in the display. I know about the SKEW and KURT equations, but the question I have is how to use them when charting or if excel has the capabilities to do this? Are they just explanitory statistics can they be employed some way when creating a bell chart so that the picture displays these characteristics? I think the only way to do this is to just use a histogram but I'm hoping that there's another way.

Instead of the chart being a normal bell curve like this:

http://www.racerocks.com/racerock/ed...ormalcurve.jpg

I want the curve to display skewness and kurtosis like these

http://grants.hhp.coe.uh.edu/doconno...PosNegPict.jpg

http://allpsych.com/researchmethods/images/kurtosis.gif

Thanks for any help.

Shg,

Thanks for the response. I think you're on the right track. After looking into the logdistribution I think I may have hit a snag. In my return stream that I'm trying to plot, I have negative values and you can't take the log/ln of a negative number (I don't think). Any other suggestions?

Thanks a ton.

I've calculated Skewness, Kurtosis, the Mean, and Standard Deviation as of now. I'm happy to calculate others if it's necessary to chart this data appropriately. Any thoughts on which of the PDFs might be best suited for this?

Got your message. Will try to get back tomorrow.

You need to use the CHI^2 statistic to test for normality. I am curious, however, why you want to know this? Can you send me the data along with a purpose statement?

Steve Druley