[R-SIG-Finance] strategy classes supported by quanstrat

Brian G. Peterson brian at braverock.com
Tue May 19 10:25:36 CEST 2015

On 05/18/2015 08:45 PM, John Williams wrote:
> Hi ...
> Is Quantstrat effective for backtesting stat arb strategies?


You'll need to be a bit more specific about what type of stat arb you're 
talking about if you want more guidance.

"statistical arbitrage" is an awfully large category.

> What are the limitations of backtesting with quantstrat?

Isn't this spelled out pretty clearly in the documentation?

Put another way, you'll get more useful answers if you ask better questions:


Briefly, guessing at what you may be asking...

quantstrat models signal-based strategies in a path-dependent manner.

quantstrat also models orders.

**What are the implications of these two statements?**

If you don't want to do what I've described above, you should probably 
use a different backtesting approach.

quantstrat's path-dependent performance will be a nearly linear function 
of the number of signals (and consequently, the number of orders).

Strategies that derive much of their performance from microstructure 
will not be well-modeled by quantstrat, or any other backtesting 
framework which is available in open source or commercially, because 
these types of microstructure processes are inherently stochastic, and 
require a huge amount of historical data both on the market and on your 
execution history to model.



Brian G. Peterson
Ph: 773-459-4973
IM: bgpbraverock

More information about the R-SIG-Finance mailing list