[R-SIG-Finance] A simple variant of Luxor strategy with Dukascopy data

[Brian G. Peterson]

On Tue, 2015-09-29 at 09:46 -0700, Tsvetan Stoyanov wrote:
> I am trying to replicate Table 3.1 in Jaekle&Tomasini using quantstrat
> and Dukascopy data.
> While I have similar percentage of winning trades, net profit and
> drawdown are quite different.
> For example, they have P&L of ~$66,000 while i get ~$19,500 for a
> period from 2003-05-04 to 2008-07-06
> which is half a year shorter than their period.
> 
> It should not make such a difference I think, unless they used
> substantially different data. 
> 
> The trade stats quantstrat code are included bellow, where I have
> set .threshold = 0.0001,  
> to reduce slippage and set .txnfees = 0.
> 
> I am new to quantstrat and will appreciate some help in reconciling
> this difference.

The most likely difference is that the Tomasini and Jaekle data appears
to have used instant execution assumptions, which quantstrat does not.

Other people have also failed to get the same numbers as the book with
widely varying data sets, including futures, IB data, and other cash FX
sources.

Regards,

Brian

-- 
Brian G. Peterson
http://braverock.com/brian/
Ph: 773-459-4973
IM: bgpbraverock