I've been testing Noxa AME optimized long only on a few stocks and it seems to work well until the trend changes. I did have noxa optimize AME for Lumber Liquidators LL for me 1 year ago both short and long and the returns have been very good 77% for the year.
I do not know what input Noxa AME you are using, here I use Noxa AME on LL data between 2007 - Dec 2010, I choose 44 as the best parameter within this period.
Afterward I extend the data up today, and still using 44 as the input parameter then you see here the result.
If you are using all data till today and you chose the best one, it doesn't mean it will perform good in the future.
CSSA looks more like a bandpass filter while SSA does not. I am wondering how different SSA and CSSA are. Do you have some comparison between SSA and CSSA with the same parameters? I am also interesting in how ESN acts like a filter since the otherput is so smooth.
SSA is repainting indicator (change the indicator value as new data arrive), while CSSA having trick to avoid repainting. There is almost no difference between them for a static data, but it will different for online data.
Now I have implemented causal SSA, causal HHT, Polynomial filters, band pass filter, enhanced MESA, structrual model, and it turns out that all these prodcue similar results, which is kind of expected. I will try them in the context of a mean reverting strategy. One problem is that they are not that smooth, which might cause lot of false signals.
Ages not to post on this thread but as im subscribed to this I got curious....
The trick in this report is the instrument. Just simple to scan and made Walk Forward test of all universe of NASDAQ instruments and with not many trades for sure you will find one which was working....again random luck converted to
Just wonder what would be average result of CSSA for all NASDAQ instruments ??