Skip to content

[Rcpp-devel] RInside for hft strategy.

2 messages · Harry G, Dirk Eddelbuettel

Original
Harry G
# 
Thinking of using RInside for predicting a theoretical price for a high
frequency trading strategy. Strategy is written in C++.

Let's assume I want to use the lm() function.
Flow i'm imagining:  get the factors in c++ -> have R pre-process them ->
use the R predict() function.

What is the recommended approach for something like this.  Latencies > 50
micros matter.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.r-forge.r-project.org/pipermail/rcpp-devel/attachments/20160317/150965f8/attachment.html>
2 days later
Dirk Eddelbuettel
#
On 17 March 2016 at 12:41, Harry G wrote:
| Thinking of using RInside for predicting a theoretical price for a high
| frequency trading strategy. Strategy is written in C++.
| 
| Let's assume I want to use the lm() function.
| Flow i'm imagining: ?get the factors in c++ -> have R pre-process them -> use
| the R predict() function.
| 
| What is the recommended approach for something like this.? Latencies > 50
| micros matter.

In that case you probably want to stay away from R in all forms, including RInside.

Most people end up using hybrid schemes -- test and train 'off-line' and use
the models and parameters online.  A decade ago I 'invented' a scheme where
the print method for my model emitted valid C(++) header code I could just
compile in ...

Dirk