R/report.R

In palencar/TraderBot: Quantitative analysis bot.

library("data.table")
library("memoise")
library("moments")
source("R/dbInterface.R")

mergeBacktest <- function(path = "result")
{
  files <- list.files(path, pattern = "*.rds")

  oper <- list()
  i <- 1

  for(file in files)
  {
    name   <- paste(path, file, sep = "/")
    operations <- readRDS(name)

    if(is.null(operations) || nrow(operations) == 0)
      next

    if(!"POSIXct" %in% class(operations$open))
    {
      operations$open <- as.POSIXct(operations$open)
      operations$last <- as.POSIXct(operations$last)
    }

    symbol <- unlist(strsplit(file, "[._]"))[1]
    timeframe <- unlist(strsplit(file, "[._]"))[2]
    oper[[i]] <- data.frame(symbol, timeframe, operations)
    i <- i + 1
  }

  dataTable <- rbindlist(oper, fill = TRUE)

  if(nrow(dataTable) == 0)
    return(dataTable)

  dataTable[, days := as.numeric(difftime(last, open, units = "days"))]
  dataTable[, grade := (profit_pp - maxDrawdown)]

  return(dataTable[order(-grade)])
}

showPlot <- function(dataTable, xy, class = "state_timeframe")
{
  df <- data.frame(dataTable)
  if(class == "state_timeframe")
    df$state_timeframe <- as.factor(paste0(df$state, df$timeframe))
  if(class == "none")
    df <- df[xy]
  else
    df <- df[c(xy, class)]
  df <- unique(df[complete.cases(df),])

  if(nrow(df) > 0)
  {
    if(class == "none")
      aesstr <- aes_string(x = xy[1], y = xy[2])
    else
      aesstr <- aes_string(x = xy[1], y = xy[2], color=class)
    ggplot(df, aesstr) + geom_point(alpha=0.05) + geom_smooth() + scale_y_log10()
  }
}

showReport <- function(dataTable)
{
  dff <- NULL
  i <- 1

  symbols <- unique(as.vector(dataTable$name))

  for(symbolName in symbols)
  {
    obj        <- dataTable[which(dataTable$name == symbolName)]
    profit     <- mean(obj$profit_pp)
    minProfit  <- min(obj$profit_pp)
    maxProfit  <- max(obj$profit_pp)
    variance   <- var(obj$profit_pp)
    skewness   <- skewness(obj$profit_pp)
    count      <- length(unique(obj$profit_pp[obj$name == symbolName]))
    df         <- data.frame(symbolName, profit, minProfit, maxProfit, variance, skewness, count)
    dff        <- rbind(dff, df)
  }

  dff <- dff[order(dff$profit),]

  return(dff)
}