I got some interesting results running the genetic programming routines. I would like to apply the formulas to watched stocks and see if they can generate buy and sell signals for me to try a little testing in the real market. Problem: genetic programming uses functions that aren't available in the context of the watch screen. I most often get an error about daysfromstart or sometimes stockcapital being unknown identifiers when I try to apply the formulas. At this point I have no clue about the logic, if any, in the formulas, so I don't want to try modifying them. I don't know the language, and even if I did, I would get a headache trying to figure them out. Many are long and convoluted, and I failed calculus in college. Or maybe I'm just lazy. Anyway, I want to try this now without taking days or weeks to figure it all out.
Can someone tell me what I'm doing wrong, or provide a workaround way to apply these formulas in real world? Maybe in context of a portfolio these two functions would work? I haven't been able to figure that out either.
I'm very excited about the possibility of an open source black box trading system that I can run on my PC. Seems like voodoo magic. If it worked in real time as it appears to work on historical data, it should theoretically beat the market consistently by a good margin. Almost seems too good to be true, but I am optimistic and want to give it a try.
A general comment on the state of the project and help docs: Something seems missing - the part where you connect the pieces for the end user. The really cool and unique part of MOV, is the genetic and ANN programming. Portfolio tracking and management is available in hundreds of systems. Almost every bank and broker has one online these days. If I were the developer, I would focus on making the unique part fully functional and then add the nice to have features, bells and whistles. IMHO.
Any help would be greatly appreciated!
PS. If I can make it work I will post my results.
I have been doing some more experimenting. I found the program variables and eliminated ones that don't work outside the context of the genetic programming (set to zero). I was able to apply resulting buy rules to a table of historical quotes, and to a watch table. The result is a reasonable appearing number of trues and falses. However, when I try the same thing with any sell rule, I get all trues for every table row. That's obviously not helpful.
Any suggestions? Am I on the right track?
Log in to post a comment.
Sign up for the SourceForge newsletter:
You seem to have CSS turned off.
Please don't fill out this field.