[R-SIG-Finance] quantstrat - macross demo problem

Stephen Choularton stephen at organicfoodmarkets.com.au
Mon Feb 7 09:15:59 CET 2011


I just downloaded what I thought was the latest quatstrat and then the 
xts update as I got this:

Failed with error: ‘package 'xts' 0.7-6.11 was found, but >= is 
required by 'quantstrat'’

but when I ordered my long rules 'enter' then 'exit' the problem of 
running into that crossing zero problem was still there. When I turned 
the order back to 'exit' then 'enter' it worked again.

I downloaded this zip and installed locally on windows:


Specify, build, and back-test quantitative financial trading and 
portfolio strategies

*Download: * 	
	*Package source (.tar.gz) 
<http://r-forge.r-project.org/src/contrib/quantstrat_0.4.0.tar.gz>* | 
*Windows multi-arch binary (.zip) 
| *MacOS X leopard binary (.tgz) 

*Logs: * 	Linux x86_32 	Linux x86_64 	Windows x86_32/x86_64 	MacOS X 
Daily build: 	patched 
| devel 
| devel (N/A)
Daily check: 	offline* 	patched 
| devel 
	offline* 	patched 
| devel (N/A)


To install this package directly within R type: 
*|install.packages("quantstrat", repos="http://R-Forge.R-project.org")|*

Version: *0.4.0* | Last change: *2011-02-04 23:07:03+01* | Rev.: *548*


Have I got the wrong quantstrat?

Stephen Choularton Ph.D., FIoD

On 28/01/2011 6:07 AM, Brian G. Peterson wrote:
> On 01/27/2011 12:53 PM, Stephen Choularton wrote:
>> I note that the list
>> "pre","risk","order","rebalance",
>> "exit","enter","entry","post"
>> Contains types that are not in the notes. Can I find some further info
>> on what they are for?
> Well, with the exception of 'order', they all (currently) call 
> RuleProc by default, as described in the documentation.
> In that regard, they are all 'the same', but will be accessed in the 
> stated order. Since the rule functions themselves are composed of 
> *any* R function, you could do just about anything with a given slot.
> That said, the names should be pretty evident, 'pre' and 'post' for 
> pre and post processing at a given observation (we occasionally use a 
> function in these slots to set or remove hold, for example), 
> 'rebalance' for portfolio rules (you'd need to write custom 
> rebalancing functions, 'enter' and 'entry' are synonyms for each other 
> because I mistyped one as the other a couple times and puzzled over 
> why things weren't working.
> Per the documentation (thanks to the Stephen's report, Aleksandr's 
> sleuthing, and Josh's patch), for each observation, rules will be 
> evaluated by type, in the order they were declared within that type. 
> 'Stacking' of rules is possible, as is having one rule cancel orders 
> placed by another rule before they have the opportunity to be worked 
> by the 'order' slot. The 'order' slot, by default, also per the 
> documentation, will call ruleOrderProc, though this may be changed by 
> the user to call custom order processing.
>> When Joshua says he has 'committed' the changes, does that mean if I
>> download quantstrat the amendments will be incorporated in it?
> R-Forge builds changes from SVN daily, so within 24 hours it should 
> show up in the tarball or binary package.
>> And finally (for the moment ;-) thanks for the way you all have helped
>> me solve my problems.
> Regards,
> - Brian

More information about the R-SIG-Finance mailing list