[R-SIG-Finance] quantstrat luxor.4, timespan optimization

[Joshua Ulrich]

On Fri, Sep 5, 2014 at 1:21 PM, amarjit chandhial
<a.chandhial at btinternet.com> wrote:
>
>
> I am trying to get "luxor.4, luxor timespan paramset optimization" running.
>
> In order to to do this:
>
> (a) I run luxor.1, including timespan=.timespan in the 4 rules.
>
> luxor.1 uses an include file which has
>
>  .timespans <- c('T06:00/T10:00', 'T07:00/T11:00', 'T08:00/T12:00',
> 'T09:00/T13:00', 'T10:00/T14:00', 'T11:00/T15:00', 'T12:00/T16:00')
>
> &
>
> I have  .timespan = 'T06:00/T10:00'
>
> I then save strategy.
>
>
>
> (b) I run luxor.2 paramsets and save strategy.
>
>
>
> (c) I run luxor.4, the optimization, whereby I get the following
> results$tradeStats for the various  .timespans:
>
<snip>
>
>
>
> Well, no, as you can see I get *the same* results for the various times in
> .timespans.
>
> These are the same tradeStats results for the vanilla luxor.1 strategy
> having put .timespan = 'T06:00/T10:00' in the 4 rules,
>
> So luxor.4 is trying to do an optimization but doesn't get past the
> .timespan value, to do the .timespans. That is so because if I follow the
> procedure for a different .timespan including a .timepsan=NULL, the
> script[s] will do the 'optimization' for that .timespan only, giving the
> same tradeStats repeated and not do for the remaining times.
>
>
>
>
> Advice/solution appreciated.
>
I spent some time this morning looking at this specific example and I
don't think the luxor.4 timespan paramset ever worked (at least I
can't see how it ever worked).  Unfortunately, I also do not see a
quick/easy way to make it work.

>
>
>
> As said what I am, and probably lots of others reading this are after is:
>
> A workflow giving
> (1) simple vanilla stratgey (an MA or oscillator)
> (2) In-sample optimization of parameter(s) of simple strategy, Out-of-sample
> run
> (3) periodic optimization & walk-forward procedure
>  with the right results.
>
>
>
>
> Amarjit
>

Best,
--
Joshua Ulrich  |  about.me/joshuaulrich
FOSS Trading  |  www.fosstrading.com