Skip to content

apply a function to a data.frame column with condition

10 messages · Enrico Schumann, Brian G. Peterson, Arnaud Gaboury

Original
Arnaud Gaboury
# 
I work with Binancer [1] and PMwR [2] packages to write a trading
journal.
Here is a sample of my data frame:
portfolio <- data.frame(
  symbol = c("BTCUSDT", "BTCUSDT", "ETHUBTC", "AAVEBTC", "ENAUSDT"),
  time = c("2024-12-17 10:01:30", "2024-12-18 21:32:53", "2025-01-02
10:34:49", "2025-01-02 11:14:43", "2025-01-02 11:15:22")
)

I want to fetch the 'BTCUSDT' price for a specific time if the symbol
of same row end with these 3 letters : 'BTC'. In my sample, it will be
rows 3 and 4.
Here is how I find BTCUSDT price for date 2025-01-02 10:34:49:

get_btc_price <- function(time) {
  klines <- binance_klines('BTCUSDT', start_time = '2025-01-02
10:34:49', end_time = '2025-01-02 10:34:49')
}

I was thinking of this code to write my new column BTCUSDT_price:

df <- portfolio %>%
  mutate(
    BTCUSDT_price = ifelse(str_detect(symbol, "BTC$"), map_dbl(time,
get_btc_price), NA_real_)
  )}

But the code returns:
Error in `mutate()`:
? In argument: `BTCUSDT_price = ifelse(...)`.
Caused by error in `map_dbl()`:
? In index: 1.
Caused by error in `[.data.table`:
! Item 1 of j is 12 which is outside the column number range [1,ncol=0]

I have tried other workarounds but all give me errors.

Thank you for help


[1]https://cran.r-project.org/web/packages/binancer
[2]https://cran.r-project.org/web/packages/PMwR

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: This is a digitally signed message part
URL: <https://stat.ethz.ch/pipermail/r-sig-finance/attachments/20250305/b2d7a372/attachment.sig>
Enrico Schumann
# 
On Wed, 05 Mar 2025, Arnaud Gaboury writes:
You want to compute a new column for your data.frame as
a function of two existing columns.

In base R, if you don't want to use a loop (and there
would be nothing wrong with a loop), you could use
'mapply'; and then cbind the column to your data.frame:

    new.column <- mapply(
        function(symbol, time) {
            if (grepl("BTC$", symbol)) {
                ## your get_price computation, using
                ## 'symbol' and 'time'
                paste("do something with", symbol, time)
                
            } else
                NA
        },
        portfolio$symbol,
        portfolio$time
    )

    ## new.column
    ##                                         BTCUSDT 
    ##                                              NA 
    ##                                         BTCUSDT 
    ##                                              NA 
    ##                                         ETHUBTC 
    ## "do something with ETHUBTC 2025-01-02 10:34:49" 
    ##                                         AAVEBTC 
    ## "do something with AAVEBTC 2025-01-02 11:14:43" 
    ##                                         ENAUSDT 
    ##                                              NA 
    ##
Arnaud Gaboury
#
On Thu, 2025-03-06 at 07:29 +0100, Enrico Schumann wrote:
Thank you for your answer.
A very warm thanks for your PMwR package. It helps me managing my
portfolio in a professional way. Unfortunately there isn't so much
material to manage, follow, compute ratio etc. 
By any chance, do you have any cool build-in functions to cook monthly
data (PL, valuation etc) as we compute every month always the same data
?
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: This is a digitally signed message part
URL: <https://stat.ethz.ch/pipermail/r-sig-finance/attachments/20250306/b177c458/attachment.sig>
Arnaud Gaboury
#
On Thu, 2025-03-06 at 07:29 +0100, Enrico Schumann wrote:
I am sorry but I can't see where I shall put my df ? And my function to
get BTC price:
get_btc_price <- function(my_time) {
    # my_time_ms <- as.integer(as.numeric(my_time) * 1000)
    klines <- binance_klines("BTCUSDT", interval = '1h', start_time =
as.Date(my_time), end_time = as.Date(my_time) + 1)
    return(as.numeric(klines$close[[18]]))
}

Thank you for your help
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: This is a digitally signed message part
URL: <https://stat.ethz.ch/pipermail/r-sig-finance/attachments/20250306/f65e0a36/attachment.sig>
Enrico Schumann
# 
On Thu, 06 Mar 2025, Arnaud Gaboury writes:
Is the ticker you want to fetch data for always the
same?  If yes, then something like this could be
used. [I don't know/use binance, so with a 'dummy' of
your function.]

    get_btc_price <- function(my_time) {
        paste(42, my_time)
    }
    
    new.column <- mapply(
        function(symbol, mytime) {
            if (grepl("BTC$", symbol)) {
                get_btc_price(mytime)
            } else
                NA
        },
        portfolio$symbol,
        portfolio$time
    )

    ##                 BTCUSDT                  BTCUSDT 
    ##                      NA                       NA 
    ##                 ETHUBTC                  AAVEBTC 
    ##"42 2025-01-02 10:34:49" "42 2025-01-02 11:14:43" 
    ##                 ENAUSDT 
    ##                      NA
Enrico Schumann
# 
On Thu, 06 Mar 2025, Arnaud Gaboury writes:
Thank you.

I am afraid there are no built-in, ready-to-use reports in
PMwR.  That was a deliberate choice: PMwR should only
provide simple tools for computing P/L, returns, or for
backtesting, say. For analyzing the results of such
computations, you have all of R for analysis and summarizing
:-)

For my reporting needs, I typically prepare templates in
plain text, HTML or LaTeX and then fill in stats such as
returns etc. See e.g.
https://cran.r-project.org/web/packages/PMwR/vignettes/FinTeX.pdf

For equity series, i.e. total value of a portfolio, using
NAVseries and its summary method might be a start, see e.g.
https://enricoschumann.net/R/packages/PMwR/manual/PMwR.html#NAVseries-summaries
There are handful of functions for analyzing such series,
e. g. `drawdowns` or `streaks`.  But as I said, most
analysis I do with basic R functions.
Arnaud Gaboury
#
On Fri, 2025-03-07 at 08:55 +0100, Enrico Schumann wrote:
Yes, in this cas I only fecth BTCUSD price, thus my get_btc_price().
For generic, I created get_price().

Your function gives me exactly what I wanted.

Question: why do you paste 42 here ?
    get_btc_price <- function(my_time) {
        paste(42, my_time)
    }
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: This is a digitally signed message part
URL: <https://stat.ethz.ch/pipermail/r-sig-finance/attachments/20250307/ff8354f9/attachment.sig>
Brian G. Peterson
# 
Arnaud,

You may want to look at 'PerformanceAnalytics' for a large number of
performance and risk functions for financial time series, including a
large number of standard tables and charts.

I can also recommend Bernhard Pfaff's excellent book 'Financial Risk
Modeling and Portfolio Optimization with R, 2nd ed.'

Regards,

Brian
Enrico Schumann
# 
On Fri, 07 Mar 2025, Arnaud Gaboury writes:

[...]
Oh, just some random computation with 'my_time' :-)
Arnaud Gaboury
#
On Fri, 2025-03-07 at 07:40 -0600, Brian G. Peterson wrote:
Yes, I know a little bit the 'PerformanceAnalytics' package. I will
certainly have a close look and do different work with it , but first I
wanted to test Enrico package as it is simple and very straightforward.
My portfolio is very simple, thus its management too.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://stat.ethz.ch/pipermail/r-sig-finance/attachments/20250307/165970f1/attachment.html>

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: This is a digitally signed message part
URL: <https://stat.ethz.ch/pipermail/r-sig-finance/attachments/20250307/165970f1/attachment.sig>