[R-SIG-Finance] tradeStats - Avg.Daily.PL/Med.Daily.PL question

[Mark Knecht]

Hi Brian,

On Thu, Sep 25, 2014 at 12:16 PM, Brian G. Peterson <brian at braverock.com> wrote:
> Mark,
>
> tradeStats does exactly what it says it is doing.  It gives you the number
> mean or median P&L for days on which you have trading P&L from
> (Net.Txn.Realized.PL).  So, on days in which you realize a gain or loss of
> any sort, (position gets closer to zero), they will be counted for the mean
> daily number in tradeStats.
>

OK, your description helps. I guess there's a follow-on question as to when
Avg.Daily.PL & Avg.Trade.PL are ever different (as I've not run across that)
but don't respond on that alone. I'll just do some work and see if I can get
there.

> dailyStats uses days with non=-zero P&L, which isn't quite right, but should
> be much closer to what you're looking for. (it requires a change in price or
> position, so that the mark will differ).  The potential error here is if you
> have days of marked P&L that have no change, e.g. days on which you are
> flat.  I'm not sure it makes sense to count these days most of the time
> anyway, but it is a difference worth describing.
>

OK. I tried running dailyStats a day or two ago and got nowhere with it but
I'll see if I can't figure that out. A quick search shows that Ilya
uses it in his
IKTrading and DSTrading files so that should be enough to get me going.

Thanks for the quick response.

Cheers,
Mark



> Regards,
>
> Brian
>
>
> On 09/25/2014 12:47 PM, Mark Knecht wrote:
>>
>> Hi,
>>     The help file for tradeStats asks for feedback so here I am.
>> Reproducible code is the macd.Rebalancing.R demo file with the symbol
>> changed to "SPY" and these lines added at the end:
>>
>> oB = getOrders(portfolio.st, "SPY", status="closed")
>> tS = t(tradeStats(portfolio.st))
>> pTS = perTradeStats(portfolio.st, "SPY")
>>
>>     The tradeStats function works correctly (as best I can tell) for
>> most values but I'm thinking either I don't understand the
>> Avg.Daily.PL value or there's a bug. Every time I run the function
>> Avg.Daily.PL & Med.Daily.PL is always equal to Avg.Trade.PL &
>> Med.Trade.PL respectively. (NOTE: I'm using daily & weekly data only.)
>>
>>     From the help file I'm not exactly sure what this is intended to mean:
>>
>> Avg.Daily.PL - mean daily realized P&L on days there were
>> transactions, see dailyStats for all days
>>
>>     Is this supposed to be the total trading profit divided by the
>> number of days in trades, or the total profit divided by the number of
>> days from the first entry to the last exit, or something else?
>>
>>     The code from quantstrat-0.8.2 that's involved seems to be this:
>>
>> PL.ne0 <- txn$Net.Txn.Realized.PL[txn$Net.Txn.Realized.PL != 0]
>>
>> DailyPL <- apply.daily(PL.ne0, sum)
>> AvgDailyPL <- mean(DailyPL)
>> MedDailyPL <- median(DailyPL)
>>
>> and
>>
>> AvgTradePL <- mean(PL.ne0)
>> MedTradePL <- median(PL.ne0
>>
>>     If the intention is to say something about how much money the
>> strategy makes "per day" then I don't see anything that tallies the
>> number of days in the simulation run. I got a simplified version of
>> that using:
>>
>> pTS = perTradeStats(portfolio.st, "SPY")
>> last(pTS$End)-first(pTS$Start)
>> Time difference of 2729 days
>>
>> but I'm sure there are easier/better ways.
>>
>>     Anyway, is it working correctly, or if not has it been fixed in a
>> more recent version of quantstrat?
>>
>> Thanks,
>> Mark
>>
>> _______________________________________________
>> R-SIG-Finance at r-project.org mailing list
>> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
>> -- Subscriber-posting only. If you want to post, subscribe first.
>> -- Also note that this is not the r-help list where general R questions
>> should go.
>>
>
> --
> Brian G. Peterson
> http://braverock.com/brian/
> Ph: 773-459-4973
> IM: bgpbraverock
>
> _______________________________________________
> R-SIG-Finance at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only. If you want to post, subscribe first.
> -- Also note that this is not the r-help list where general R questions
> should go.