Posted By: webradio
how good are the parts of Z12? R script and images to check - 08/27/16 16:35
I was interested to see how do algos of Z12 perform per asset and trade direction. (One of the reasons was a honest warning on the manual page that test results are selection biased due to using OptF from the future).
So I first modified Z.ini to disable equity curve trading by setting Phantom to 0 and then Z12.par, where I entered .100 as OptF for all assets and algos.
Then I executed a Test. And then used R to create individual equity curves from testtrades.csv.
The following script produces a bunch of png files, one per asset+algo combination, like this:
An attachemnt contains all of them, in case you don't want to run the script yourself.
My personal take-away so far: do take it seriously about "The equity curve above is on the optimistic side due to selection bias" on the Manual page
So I first modified Z.ini to disable equity curve trading by setting Phantom to 0 and then Z12.par, where I entered .100 as OptF for all assets and algos.
Then I executed a Test. And then used R to create individual equity curves from testtrades.csv.
The following script produces a bunch of png files, one per asset+algo combination, like this:
An attachemnt contains all of them, in case you don't want to run the script yourself.
Code:
trds <- read.csv("C:\\Zorro\\Log\\testtrades.csv", stringsAsFactors=TRUE) trds$Open <- as.POSIXct(trds$Open, tz="GMT") trds$Close <- as.POSIXct(trds$Close, tz="GMT") #install.packages("stringi") #do it once require(stringi) for (i in levels(trds[,"Name"])) { for (j in levels(trds[,"Asset"])) { nice.asset.name <- stri_replace_all(str=j, replacement="-", fixed="/") png.file.name <- paste(sep="",i,"_",nice.asset.name,".png") png(file=png.file.name, width = 600, height = 800, units = "px", pointsize = 16) old.par <- par(mfcol=c(3,1)) k <- "Long" idx <- trds$Name==i & trds$Asset==j & trds$Type==k plot(y=cumsum(trds$Profit[idx]), x=trds$Close[idx], pch=19, main=paste(i, j, k), ylab="Eq", xlab=""); abline(h=0) k <- "Short" idx <- trds$Name==i & trds$Asset==j & trds$Type==k plot(y=cumsum(trds$Profit[idx]), x=trds$Close[idx], pch=19, main=paste(i, j, k), ylab="Eq", xlab=""); abline(h=0) idx <- trds$Name==i & trds$Asset==j plot(y=cumsum(trds$Profit[idx]), x=trds$Close[idx], pch=19, main=paste(i, j ), ylab="Eq", xlab=""); abline(h=0) par(old.par) dev.off() } }
My personal take-away so far: do take it seriously about "The equity curve above is on the optimistic side due to selection bias" on the Manual page
Description: Individual Equity Curves For Z12 Members