josephdunn

# original idea: http://mcarreira.typepad.com/mc_notes/2014/03/a-simple-model-for-the-pl-of-a-market-maker-how-many-losing-days-in-a-year.html
ndays <- 1e5; pl <- unlist(lapply(1:ndays, function(x) { sum((rbinom(7*60*10, 1, 0.525) - 0.5) * 2) })) ; 1 / (length(pl[pl < 0]) / ndays * 252)

# updated code demonstrating win vs. loss asymmetry with M/T. thanks @marascio and @MarcosCarreira for the suggestions.
# wins are 1.00, losses are -1.05 (normalized)
ndays <- 1e5; pl <- unlist(lapply(1:ndays, function(x) { sum(ifelse((rbinom(7*60*10, 1, 0.525) - 0.5) * 2 == -1, -1.05, 1)) })) ; 1 / (length(pl[pl < 0]) / ndays * 252)

# same as above - ifelse() is slow compared to writing the vector correctly the first time, but this is more cryptic.
# wins are 1.00, losses are -1.05 (normalized)
ndays <- 1e5; pl <- unlist(lapply(1:ndays, function(x) { sum(((1-rbinom(7*60*10, 1, 0.525))*-2.05)+1) })) ; 1 / (length(pl[pl < 0]) / ndays * 252)

josephdunn

import numpy as np

ndays = int(1e5)
ntradesday = 7*60*10
pl = np.sum((np.random.binomial(1, 0.525, (ndays, ntradesday))-0.5)*2, 1)
1.0 / (len(pl[pl < 0]) / float(ndays) * 252)

josephdunn

winProb <- 0.525
numBets <- c(1, 50, 1000, 10000, 100000)
runTimes <- 1000

pl <- function(bets)
{
    sum(rbinom(bets, 1, winProb) - 0.5)
}

runs <- do.call(rbind, lapply(1:runTimes, function(x) { sapply(numBets, pl) }))

josephdunn

hello milktrader!

josephdunn

library(quantmod)
library(PerformanceAnalytics)

getSymbols('HEIA.AS', src='yahoo')
Return.cumulative(Return.calculate(get('HEIA.AS')['2010-12-31::']))

josephdunn

#!/usr/bin/ruby

def winloss
  if rand(2) == 1
    return true
  else
    return false
  end
end

josephdunn

#!/usr/bin/ruby

def winloss
  if rand(2) == 1
    return true
  else
    return false
  end
end

josephdunn

require 'rubygems'
require 'eventmachine'
require 'em-http-request'
require 'json'
 
EventMachine.run {
  bid = 0.0
  ask = 0.0
  spread = 0.0

josephdunn

data <- read.table('TrAEX.xls', header=TRUE, sep='\t', skip=5, stringsAsFactors=FALSE)
data.xts <- xts(data$Quote, order.by=as.POSIXlt(data$Date...time, format='%H:%M:%S'))

josephdunn

require(quantstrat)

# clean up
rm(list=ls(all=TRUE, pos=.blotter), pos=.blotter)
rm(list=ls(all=TRUE, pos=.instrument), pos=.instrument)
rm(list=ls(all=TRUE, pos=.strategy), pos=.strategy)

# set initial values
initDate <- '2011-08-05 06:00:00'
initEq   <- 10000