Tell me more ×
Programmers Stack Exchange is a question and answer site for professional programmers interested in conceptual questions about software development. It's 100% free, no registration required.
up vote 4 down vote favorite

I want to start messing around with Genetic Algorithm's in my spare time. I've been thinking of some ways to incorporate them either into personal projects or work projects.

I work in financial software - so a lot of risk analysis and things like that using derivative markets. Not sure how GA's would fit into that.

Was curious to see if anyone had any cool ideas about how I could use GA's to cure some boredom and do some really cool stuff!

share|improve this question

2 Answers

up vote 0 down vote accepted

There tons of things. Any problem that asks "What is the best..." or "What is the most..." are excellent candidates for genetic algorithms. Ask yourself, "what problems do I face that involve optimizing a quantity?". Now I don't claim that GAs will be the best techniques, but they should be considered amongst a pool of other potential metaheuristics.

One of my projects was to use a GA to fit a function to an arbitrary set of data points, minimizing the least square residual.

You could also use a GA to design a physical object. Some of examples that I've seen are cars and clocks.

share|improve this answer
Do you think I could use a GA to try and predict some facet of the stock market? Just for fun, not guaranteeing it would do anything profoundly useful, haha. – slandau Aug 9 '11 at 18:31
I was actually thinking about this earlier this year. And if you do a quick Google search on it, there have been a few papers written on it (although most not from very credible sources). You should be careful of data dredging though: en.wikipedia.org/wiki/Data_dredging . It does sound like a very interesting idea though. – tskuzzy Aug 9 '11 at 18:35
I'll give this a read. – slandau Aug 9 '11 at 18:43
up vote 1 down vote

You could use them to predict the stock market as long as you have some kind of reasonable scoring function you can make out of it (which is pretty easy for the stock market). It would help hugely if it's something you can get reams of historical data for.

The problem is that even if they work somewhat, it's very hard to figure out what they are doing and nearly impossible to figure out why, especially since recording the evolutionary history takes too much disk space. And it's kind of unlikely they will do well in unusual market conditions - when you really need them to do well.

share|improve this answer
Good point. Thanks – slandau Aug 9 '11 at 18:43

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.