Altair RISE

A program to recognize and reward our most engaged community members

Nominate Yourself Now!

Is there anybody out there?

Using RapidMiner for/in conjunction with:

financial modeling
systematic trading
backtesting
time series of asset prices, etc.
anything similar

It would be great to build a (sub)community and learn from one another/avoid reinventing the wheel, help each other out, encourage one another!

@Thomas_Ott, @sgenzer, @tftemme, @IngoRM

Find more posts tagged with

AI Studio

Finance

Accepted answers

hughesfleming68

Hi Noel, you can do it with Rapidminer. I use it exclusively for time series forecasting. You will need another platform for your back testing and some way to manage data exchange. Python and R are helpful if you need to do something that can't be done in Rapidminer natively. This subject comes up occasionally but here isn't that much content. This particular area of machine learning tends to be rather secretive outside of papers published by academics.

regards,

Alex

sgenzer

hi @Noel TBH Tom Ott is the only one I know who really dug deep into financial use cases AND was willing to share publicly. In full disclosure, Tom now works for H2O so I'm sure he's spending more time with that product than RapidMiner.

[N.B. that's not a slam on Tom or H2O at all - we're still good friends and RapidMiner actually uses many H2O libraries in its operators under the hood. It's all good.]

Scott

All comments

sgenzer

Hello @Noel well you certainly know how to tag all sorts of people on this community. It's fortunate as there are no categories or sub-categories in this community; the entire structure is based on tags. Have you explored the Finance tag of the community? This seems like a good place to start. Just click.

Image: https://us.v-cdn.net/6030995/uploads/editor/ke/d9rx3uqowoar.png

You can also search for a tag at any time:

Image: https://us.v-cdn.net/6030995/uploads/editor/3k/82z1zuavs6i3.png

Scott

hughesfleming68

I think Rapidminer is good for this but it is not a back testing tool and it would be challenging to use it on real time data . Building machine learning models on financial time series is very difficult. My guess is that quants who have been able to do it won't be willing to share their ideas on the internet very willingly. Nevertheless, the book "Advances in Financial Machine Learning" by Marcos Lopez de Prado has a lot of useful information on this subject and lots of good advice. You would have to be up to speed in Python to understand the code but with some tinkering, you can get the code to work in Rapidminer by using the Python extension.

Noel

hughesfleming68 and @sgenzer for the responses!

@hughesfleming68- I take your point, but I'm not after anyone's special sauce. RM seems like a great rapid prototyping and investigatory tool and I've learned a ton using it. But, have others used it with time series data, to posit trading strategies, backtest them, etc.? If not, should I instead be focusing on Python techniques if I'm interested in this domain and avoid going down this rabbit hole?

Also, there are nonproprietary things, such as what other software/tech folks have integrated with RM, the basic design of systems they've built, and nuts and bolts RM process-issues people have encountered that could be shared... @Thomas_Ott 's youtube vids are the only "sharing" I've seen.

Any additional thoughts would be greatly appreciated.

hughesfleming68

sgenzer