# TODO:
# - add support for dynamic position sizing
# - add param to trades to specify single position vs. multiple positions
# - add support for trading costs/slipage
# - test with different trading rules (TTR's)
# - add support for adding to positions (pyramiding)
# - add relative returns (R = (P - (P - 1)) / P)?
# - expand stats for analysis of results
# - add support for derivatives
# - check and/or add support for intraday data 
# - create test data sets and tests
 

#library("TTR")
library("quantmod")
#library("PerformanceAnalytics")
#library("blotter")

crosses_over <- function(data, indicator, value) {
	co_r = cbind(data, indicator)
	co_r[lag(co_r[, "..2"], 1) <= value & co_r[, "..2"] > value]
}

crosses_under <- function(data, indicator, value) {
	cu_r = cbind(data, indicator)
	cu_r[lag(cu_r[, "..2"], 1) > value & cu_r[, "..2"] <= value]
}

# when_to_trade:	0 = at the start of the period
#			1 = at the end of the period
trades <- function(data, open_long=NULL, close_long=NULL, open_short=NULL, close_short=NULL, long_amount=1, short_amount=1, when_to_trade=0) {
	data$Position = 0
	data$LongTrades = 0
	data$ShortTrades = 0

	# Handle longs
	if(!is.null(open_long)) {
		ol = open_long
		ol$Trade = long_amount

		clc = close_long
		clc$Trade = -long_amount
	
		cl = clc[paste(index(first(ol)), '::', sep="")]
		longs = rbind(ol, cl)
		longs = longs[ (longs [, "Trade"] + lag(longs [, "Trade"], 1) == 0)]
		longs = rbind(ol[1], longs)

		data[index(longs), "LongTrades"] = longs$Trade

		for(row in 2:nrow(longs)) {
			if(row %% 2 == 0) {
				if(when_to_trade == 1) {
					data[paste(index(data[grep(index(longs[row-1]), index(data))+1]), "::", index(longs[row]), sep=""), "Position"] = long_amount
				} else {
					data[paste(index(longs[row-1]), "::", index(data[grep(index(longs[row]), index(data)) - 1]), sep=""), "Position"] = long_amount
				}
			}
		}
	
		if(nrow(longs) %% 2 == 1) {
			if(when_to_trade == 1) {
				data[paste(index(data[grep(index(longs[nrow(longs)]), index(data))+1]), "::", sep=""), "Position"] = long_amount
			} else {
				data[paste(index(longs[nrow(longs)]), "::", sep=""), "Position"] = long_amount
			}
		}
	}

	# Handle shorts
	if(!is.null(open_short)) {
		os = open_short
		os$Trade = -short_amount

		csc = close_short
		csc$Trade = short_amount
	
		cs = csc[paste(index(first(os)), '::', sep="")]
		shorts = rbind(os, cs)

		shorts = shorts[ (shorts [, "Trade"] + lag(shorts [, "Trade"], 1) == 0)]
		shorts = rbind(os[1], shorts)

		data[index(shorts), "ShortTrades"] = shorts$Trade

		for(row in 2:nrow(shorts)) {
			if(row %% 2 == 0) {
				if(when_to_trade == 1) {
					data[paste(index(data[grep(index(shorts[row-1]), index(data))+1]), "::", index(shorts[row]), sep=""), "Position"] = 
						data[paste(index(data[grep(index(shorts[row-1]), index(data))+1]), "::", index(shorts[row]), sep=""), "Position"] - short_amount
				} else {
					data[paste(index(shorts[row-1]), "::", index(data[grep(index(shorts[row]), index(data))-1]), sep=""), "Position"] = 
						data[paste(index(shorts[row-1]), "::", index(data[grep(index(shorts[row]), index(data))-1]), sep=""), "Position"] - short_amount
				}
			}
		}

		if(nrow(shorts) %% 2 == 1) {
			if(when_to_trade == 1) {
				data[paste(index(data[grep(index(shorts[nrow(shorts)]), index(data))+1]), "::", sep=""), "Position"] =
					data[paste(index(data[grep(index(shorts[nrow(shorts)]), index(data))+1]), "::", sep=""), "Position"] - short_amount
			} else {		
				data[paste(index(shorts[nrow(shorts)]), "::", sep=""), "Position"] =
					data[paste(index(shorts[nrow(shorts)]), "::", sep=""), "Position"] - short_amount
			}
		}
	}

	data
}

abs_returns <- function(data) {
	data$AbsReturn = (data$Price - lag(data$Price, 1)) * data$Position 
	data
}

#compound_returns <- function(data) {
# Do we also need comparable simple/compound and/or log() returns, i.e.:
# data = cbind(data, Return.calculate(data$Price, method="compound"))
#}


wealth <- function(data, start=10000.0) {
	# data$Wealth = lag(data$Wealth, 1) + data$Return 
	data$Wealth = 0

	# TODO: handle NA value in first row more elegant
	data[1, "Wealth"] = start + 0
	prevWealth = matrix(data[1, "Wealth"])

	for(row in 2:nrow(data)) {
		#print(row)
		data[row, "Wealth"] = data[row, "AbsReturn"] + prevWealth
		prevWealth = matrix(data[row, "Wealth"])
	}

	data
}

stats <- function(data) {
	stats=0
	stats$pl = 0

	stats$pl$all = sum(data[(data$Position != 0), "AbsReturn"])
	stats$pl$longs = sum(data[(data$Position > 0), "AbsReturn"])
	stats$pl$shorts = sum(data[(data$Position < 0), "AbsReturn"])

	stats
}