Skip to content

Simple portfolio management - anything in R?

6 messages · Peter Carl, Mark Knecht, Radek Maciaszek

Original
Mark Knecht
# 
Hi,
   Really unformed question here. Sorry about that but I'm just poking
around looking for answers/ideas. Thanks in advance.

   In the last year I've put some money in some a few managed accounts
for long term investing. Current these accounts hold 70-80 different
stocks & ETFs. While the investments themselves are going OK I'm not
very happy with the information the brokers give me in terms of how
the portfolios are doing against their respective benchmarks. To that
end I'm starting looking around for how I could get a little better
view.

   In the 1980s I suppose I might have been doing a very simple
version of this in Quicken, but that program seems quite disliked by
people who buy it on Amazon. Additionally, I'm Linux-based and find it
distasteful to spend money on M$ Windows apps VM unless they are
really good at what they do.

   I suspect the professional folks use tools that the rest of us
never will. Things from Bloomberg or the like. There's also a lot of
online portfolio trackers. Maybe one of them is actually good but I
haven't found it yet and I'm reticent to put too much private info on
yet another web site.

   This whole area seems like a natural for something done in R. Maybe
keep the trades in an Excel spreadsheet or CSV file and then load them
into R, get data from Yahoo, create graphs, do correlations, etc.

   Does anything like this exist? Or am I wrong headed and there's a
better way to do this?

   I'm not thinking of much more than an account balance, trade info -
date, number of shares, prices, etc., and then aggregate the results
into a few numbers for the portfolio.

   Comments?

Thanks,
Mark
Peter Carl
# 
Consider blotter, which is specifically for calculating P&L from trades
and prices (for generating returns for analysis in PerformanceAnalytics).

install.packages("blotter", repos="http://R-Forge.R-project.org")

pcc
Mark Knecht
#
On Fri, Oct 25, 2013 at 8:48 AM, Peter Carl <peter at braverock.com> wrote:
Hi Peter,
   Yes, I think the solution, if it exists or needs to be written,
probably revolves around blotter and PerformanceAnalytics. However
every time I start looking at those things I glaze over for lack of
documentation. Nothing in there seems oriented toward teaching how to
use the packages so I was hoping maybe someone has already written
some example code of something like this:

1) Have a list of stocks & download EOD data from Yahoo for that list (Easy)
3) Read a list of transactions in any human writable format. I.e. -
I'll get the trades from my broker and enter them in some reasonable
file format.
4) Generate some simple data from those transactions: equity curve for
each stock, for the portfolio

   I suspect that if this much existed that a programmer with my
meager skills might be able to go forward. However I can't find the
starting point in the blotter or PerformanceAnalytics help pages.
Maybe I just haven't found the right page yet though.

Cheers,
Mark
Peter Carl
# 
I recommend a double espresso.  Then:

library(blotter)
?blotter

In the example provided, you can see your starting point:
?getSymbols
?initPortf
?addTxns
?updatePortf
?initAcct
?updateAcct
?updateEndEq

pcc
Mark Knecht
#
On Fri, Oct 25, 2013 at 10:21 AM, Peter Carl <peter at braverock.com> wrote:
Peter,
   Thanks for the pointers. I've made a little bit of headway this
Friday afternoon and likely can make more progress over the weekend.

   One quick question, mostly out of curiosity. How/where are blotter
portfolios stored? The initPortf function seems to create something
lasts beyond my current run of R. If I exit R-Studio, reenter and
rerun my code that created the portfolio in the first place it tells
me:
Error in initPortf(name = "Mark1", symbols = "StockList") :
  Portfolio Mark1 already exists, use updatePortf() or addPortfInstr()
to update it.

I do see it's in the blotter environment:
[1] "portfolio.Mark1"

   I'm on Linux, ran updatedb and attempted to locate both portfolio
and Mark1 but didn't find anything meaningful.

Thanks,
Mark
5 days later