Moving window

Dear,
I ran FragStat based on a "moving window" and specified the following analysis parameters
- Use an 8-cell neighborhood
- Landscape metrics moving window square with a side length of 100 meters
- I calculated PD, LSI, LPI, and SHDI. 
After running, I saw the results in ArcGIS, and their values include low values such as -999 across all metrics; I am confused as to why I have such values. I know that my background is 999 and why it is included in the map output.
Another thing I want is to have my results in a 100*100 fishnet, and what statistical methods (I mean, should I have to use mean, sum, median, standard deviation) do you recommend to me for each landscape metric I am using?
I attached my screenshot of the Fragstats input and the results of each LM

Dear Samuel,

I don't see any attachments to this message. If you attached anything it did not make it to the forum as you can see yourself.

I would like to have a copy of your model to see why you get -999 in the outputs.

Normally that should not be the case (for exhaustive moving window), but I cannot tell you exactly what happened unless I can replicate the issue locally.

Best regards,
Eduard Ene