Seasonal adjustments for weekly series

I've been trying to learn the software from the Fed that does seasonal adjustment for weekly series. The software is not user-friendly, so it's been a difficult learning curve. Tom Evans at BLS is working on a SAS interface that will make the program easier to use, but the interface is still in development. Tom has been helping me learn the program, and he helps me fix my errors when I make a mistake because error messages are nonexistent in the program. I was also having trouble getting the program to run on my series, and I believe it's because my series are short, so the program has trouble estimating holiday effects.

In the short term, I have found that I can get something like looks like a seasonal adjustment from X-12 if I ask for a trend. I can get it almost as smooth or as rough as I would like by changing the trend filter length. With a limit on the Henderson filters of 101, and with a period of 52 (and sometimes 53), the very longest trend doesn't really look like a trend because you can still see the seasonality. But with a very short trend, it is very close to what I get from the Fed's program. So in the short term, I like what I get from X-12, and it's a lot easier to use.