risktoollib / rtl Goto Github PK

It's probably a better idea to make assumptions on the structure of the df (i.e. contracts come in order of expiry) rather than assume that the contract expiry order will be implied in the column names with numbers. I'm going to make that change to the python version.

I don't think the 'stats' output returns the right data.

@risktoollib rgdal was archived two days ago, so please do move to submit an updated version of RTL as soon as possible. See https://github.com/r-spatial/evolution https://r-spatial.github.io/evolution/ for details.

It would make it easier to browse documentation
Locally, you would only run

usethis::use_pkgdown_github_pages()

Thanks

Using your example:

cpairs <- dplyr::tibble(year = c("2014","2019","2020"),
first = c("@HO4H","@HO9H","@ho0h"),
second = c("@HO4J","@HO9J","@HO0J"))
ou <- chart_spreads(cpairs = cpairs, daysFromExpiry = 200, from = "2012-01-01",
conversion = 42,feed = "CME_NymexFutures_EOD",
iuser = username, ipassword = password,
title = "March/April ULSD Nymex Spreads",
yaxis = "$ per bbl",
output = "chart")

the 2019 and 2020 spreads only return 170 and 169 days respectively. 2014 returns 200 days as expected.

https://api.eia.gov/v2/petroleum/pnp/pct/data/?&api_key=xxxxxxx

EIAkey is not defined in mutate command. It should be key which is the argument name

is there a need to drop the first row? I don't think so....

df <- df %>% dplyr::filter(!date %in% sort(unique(df$date))[1])

rolladjust drops first row because x$expiry <- lag(x$expiry) creates a null value at the beginning. you should fill in the null with a zero before running
x <- x %>% dplyr::filter(expiry != 1) %>% dplyr::select(-expiry)

API limit is 5000, which results in needing to run multiple queries and combine them to extract a time series. Action is to build logic to return it as a clean series based on user specific start and end date.

argument EIAkey is not defined

argument name is key, but EIAkey is used to call EIA API leading to an error.

chart_fwd_curves.R
distdescplot.R
garch.R
promptBeta.R
rolladjust.R

res output type fails on check

you might get a better fit to the beta curve if you add a third degree of freedom using a constant bias. So the formula would be beta ~ x0exp(x1prompt)+x2