Acceleration Bands in TTR

[serkor1]

Hi @joshuaulrich

On the Finance Discord-server you (I believe) wrote that you are open for PRs adding indicators found in Ta-Lib but not available in TTR, and that we should open an issue to discuss the implementation. I would like to take a crack at adding Acceleration Bands.

The formula is simple (link):

Upper Band = SMA * (1 + (High - Low) / (SMA))
Lower Band = SMA * (1 - (High - Low) / (SMA))

I guess its a trivial exercise unless you want a C-implementation. How would you want it done?

Best,

[joshuaulrich]

Hi and thanks for your willingness to contribute!

Yes, this should be a simple addition. I asked you to reach out first to make sure you follow the pattern of the existing TTR functions. The stoch() function has similar inputs that are used in a similar way (i.e. two possible moving averages (1) upper and (2) lower, and running high/low prices).

So the acceleration bands function should look something like:

accelBands <- function(HLC, n, maType, ...) {
  HLC <- try.xts(HLC, error=as.matrix)

  # Calculation if HLC series is given
  if (NCOL(HLC) == 3) {
    high  <- HLC[,1]
    low   <- HLC[,2]
    close <- HLC[,3]
  } else

  # Calculation if price vector is given
  if (NCOL(HLC) == 1) {
    high  <- HLC
    low   <- HLC
    close <- HLC
  } else {
    stop("Price series must be either High-Low-Close, or Close")
  }

  hmax <- runMax(high, n)
  lmin <- runMin( low, n)

  if (missing(maType)) {
    maType <- 'SMA'
  }

  # Case of two different 'maType's for both MAs.
  # e.g. stoch(price, 14, 3, 3,
  #           maType=list(maUp=list(EMA,ratio=1/5), maDown=list(WMA,wts=1:10)) )
  if (is.list(maType)) {
    # see stoch() source
    up <- upBand
    dn <- dnBand
  } else {
    # see stoch() source
    up <- upBand
    dn <- dnBand
  }

  result <- cbind(up, dn)
  colnames(result) <- c("up", "dn")

  reclass(result, HLC)
}

It would also be great if you could provide a unit test with example inputs (ideally using TTR's ttrc data set) to make sure the TTR implementation matches the ta-lib results.

[serkor1]

@joshuaulrich I have a solution in my own fork. But I have small challenge here: How do you export the functions? Regardless of whether I run make or make docs with or without @export it does not get exported, so the examples fails on R CMD check, but unit-tests pass.

[joshuaulrich]

I don't use roxygen2 for anything related to the package namespace. There have been too many breaking changes over the years. You usually only have to fiddle with the namespace file when you first create a function, so I was wasting more time dealing with roxygen2 changes than just doing it myself.

I apologize that you spent time trying to debug that.

[serkor1]

Oooooh! The Namespace—of course its the Namespace! Its been years since Ive opened the Namespace and manually edited it! That brought back some memories.