This is a request for jcl and the Zorro developers.

Would it be possible to provide a tutorial on using the neural() method for accessing external machine learning libraries? Something like the Alice and Bob workshops would be fantastic, but even a more detailed example than the one in the manual, with explanatory comments, would be extremely welcome.

I'm currently spending most of my development time directly in R, and then only testing and trading saved models via direct calls to R using Rset(), Rx() etc. This is clunky and means that I am missing out on some of the greatest benefits of using Zorro, namely rapid strategy prototyping and direct access to robust testing and training routines.

jcl, perhaps this would make a good post for the Financial Hacker blog. I know it would be popular amongst Zorro users, and would likely convert some R users as well!

Failing that, would it be asking too much to simply post some examples from your own research?

You can find in the r.h header file an outcommented framework for the "neural" function. That can be used as a template that should work for most machine learning algorithms.

And yes, I'll definitely post a tutorial. In fact I would have done this already some time ago, but some other tasks intervened.

Brilliant! Thanks directing me to the framework in r.h; I hadn't seen that before.