Contents of AB-FT afl02 zip

8-9-2006

Welcome and hope you find these files of use to you

This zip files includes the following:


AMI NDXrsil.afl                3/3/05   Ken Close-f
ound the attached afl code on my hard drive. Read down the code and you will find a section that does the same thing as SignalPairTrade in Trade.


Trade-Singalpairtrade.afl  3/3/05 Ken Close-
Another SignalPairTrade piece of code from Bruce Robinson---I neglected to mention the previous was also from Bruce


RUTTRft3.afl                      3/15/05    Tom Ttaso-
I'm attaching an afl I just ran. this is not my code. I found something posted and just used it, with a minor change in the ROC calculations.
Anyway, I just ran it using the following:
1. apply to: current symbol which is dfscx
2. range: 10/1/99 to 3/11/2005
3. 0 day delays on both buy/sell
4. no stops used, nor anything else changed.

I ended up w/ an empty result.

I wonder if the reason is that I was trying to optimize for Mdd, and the best value for Mdd would be 0 (i.e. no trading)?

I think you should be able to use the stoploss on the actual fund you are trading. For example,
1. trade rspfx (or whatever) using ruttrft.
2. save the output fnu, XXX.fnu
3. then use the code above, w/ XXX as the fund.

So you're basically applying a trailing stoploss to your equity curve, although you can also certainly apply it to the fund/index you're using for the timing signal.


PSO Optimized.afl   3/15/05    Fred Tonnetti-Take note of the default values in this version (after optimization) ... because there is nothing in the fitness statement to keep the number of trades down the result of course is that the number of trades are quite high ...
 


RUTTRft3a.afl   & 2005-03-16b.xls   3/16/05   Tom Ttaso-I'm making progress in understanding my issue. Couple of things.

1. I was going batty trying to figure out why certain things were not repeatable. I found something interesting. In the AA window, I would click on "pick" (to choose ruttrft3.afl, for instance) followed by "edit" to change the Fitness (between CAR and CAR/MDD). Then, when I started to run PSO, the PSOptimizer (v0.9.9.5) would list the *last* Fitness variable I had used. Not the once I just changed. I would then solve this problem by "picking" the file twice. After the second time, the PSOptimizer would have the proper Fitness variable. Have you ever noticed this? Am I doing something wrong?

2. after I figured this out, Paying attention to that, all of the runs I did using car/mdd gave a result, and all the runs w/ mdd gave me no result (using your equity afl, just a grey line that looks like the BH curve of the fund you were trying to time).

I'm attaching the afl I used as well as an excel file w/ the different runs I tried. I noticed something funny on the 9th iteration, and saw it again on the 11th. that's when I realized the issue in #1 above.

from these results, I'm leaning towards the possibility that trying to minimize mdd gives you the result of "leave your cash under a mattress and you'll have zero drawdown".

 

 

Download AB-FT afl02 zip