Probabilistic Momentum with Intraday data

March 30, 2014
By

(This article was first published on Systematic Investor » R, and kindly contributed to R-bloggers)

I want to follow up the Intraday data post with testing the Probabilistic Momentum strategy on Intraday data. I will use Intraday data for SPY and GLD from the Bonnot Gang to test the strategy.

##############################################################################
# Load Systematic Investor Toolbox (SIT)
# http://systematicinvestor.wordpress.com/systematic-investor-toolbox/
###############################################################################
setInternet2(TRUE)
con = gzcon(url('http://www.systematicportfolio.com/sit.gz', 'rb'))
    source(con)
close(con)

	#*****************************************************************
	# Load historical data
	#****************************************************************** 
	load.packages('quantmod')	

	# data from http://thebonnotgang.com/tbg/historical-data/
	# please save SPY and GLD 1 min data at the given path
	spath = 'c:/Desktop/'
	data = bt.load.thebonnotgang.data('SPY,GLD', spath)
	
	data1 <- new.env()		
		data1$FI = data$GLD
		data1$EQ = data$SPY
	data = data1
	bt.prep(data, align='keep.all', fill.gaps = T)

	lookback.len = 120
	confidence.level = 60/100
	
	prices = data$prices
		ret = prices / mlag(prices) - 1 
		
	models = list()
	
	#*****************************************************************
	# Simple Momentum
	#****************************************************************** 
	momentum = prices / mlag(prices, lookback.len)
	data$weight[] = NA
		data$weight$EQ[] = momentum$EQ > momentum$FI
		data$weight$FI[] = momentum$EQ <= momentum$FI
	models$Simple  = bt.run.share(data, clean.signal=T) 	

	#*****************************************************************
	# Probabilistic Momentum + Confidence Level
	# http://cssanalytics.wordpress.com/2014/01/28/are-simple-momentum-strategies-too-dumb-introducing-probabilistic-momentum/
	# http://cssanalytics.wordpress.com/2014/02/12/probabilistic-momentum-spreadsheet/
	#****************************************************************** 
	ir = sqrt(lookback.len) * runMean(ret$EQ - ret$FI, lookback.len) / runSD(ret$EQ - ret$FI, lookback.len)
	momentum.p = pt(ir, lookback.len - 1)
		
	data$weight[] = NA
		data$weight$EQ[] = iif(cross.up(momentum.p, confidence.level), 1, iif(cross.dn(momentum.p, (1 - confidence.level)), 0,NA))
		data$weight$FI[] = iif(cross.dn(momentum.p, (1 - confidence.level)), 1, iif(cross.up(momentum.p, confidence.level), 0,NA))
	models$Probabilistic  = bt.run.share(data, clean.signal=T) 	

	data$weight[] = NA
		data$weight$EQ[] = iif(cross.up(momentum.p, confidence.level), 1, iif(cross.up(momentum.p, (1 - confidence.level)), 0,NA))
		data$weight$FI[] = iif(cross.dn(momentum.p, (1 - confidence.level)), 1, iif(cross.up(momentum.p, confidence.level), 0,NA))
	models$Probabilistic.Leverage = bt.run.share(data, clean.signal=T) 	
	
	#*****************************************************************
	# Create Report
	#******************************************************************        
	strategy.performance.snapshoot(models, T)    

plot1

Next, let’s examine the hourly perfromance of the strategy.

	#*****************************************************************
	# Hourly Performance
	#******************************************************************    
	strategy.name = 'Probabilistic.Leverage'
	ret = models[[strategy.name]]$ret	
		ret.number = 100*as.double(ret)
		
	dates = index(ret)
	factor = format(dates, '%H')
    
	layout(1:2)
	par(mar=c(4,4,1,1))
	boxplot(tapply(ret.number, factor, function(x) x),outline=T, main=paste(strategy.name, 'Distribution of Returns'), las=1)
	barplot(tapply(ret.number, factor, function(x) sum(x)), main=paste(strategy.name, 'P&L by Hour'), las=1)

plot2

There are lots of abnormal returns in the 9:30-10:00am box due to big overnight returns. I.e. a return from today’s open to prior’s day close. If we exclude this observation every day, the distribution each hour is more consistent.

   	#*****************************************************************
   	# Hourly Performance: Remove first return of the day (i.e. overnight)
   	#******************************************************************    
   	day.stat = bt.intraday.day(dates)
	ret.number[day.stat$day.start] = 0

   	layout(1:2)
   	par(mar=c(4,4,1,1))
	boxplot(tapply(ret.number, factor, function(x) x),outline=T, main=paste(strategy.name, 'Distribution of Returns'), las=1)
	barplot(tapply(ret.number, factor, function(x) sum(x)), main=paste(strategy.name, 'P&L by Hour'), las=1)

plot3

The strategy performs best in the morning and dwindles down in the afternoon and overnight.

These hourly seasonality plots are just a different way to analyze performance of the strategy based on Intraday data.

To view the complete source code for this example, please have a look at the bt.strategy.intraday.thebonnotgang.test() function in bt.test.r at github.


To leave a comment for the author, please follow the link and comment on his blog: Systematic Investor » R.

R-bloggers.com offers daily e-mail updates about R news and tutorials on topics such as: visualization (ggplot2, Boxplots, maps, animation), programming (RStudio, Sweave, LaTeX, SQL, Eclipse, git, hadoop, Web Scraping) statistics (regression, PCA, time series, trading) and more...



If you got this far, why not subscribe for updates from the site? Choose your flavor: e-mail, twitter, RSS, or facebook...

Comments are closed.