Finding your own way

A common question you hear from newcomers to sports trading is, "How do you...?" and the common answer is, "I can't tell you." Information is the key to success in sports trading and people will protect their valuable information.

Once valuable information is shared it loses its value. You only have to look at a betting market to see information being shared and its value being divided up amongst the quick thinking.

I have spent the majority of my life learning how betting markets work. Even as a child I was fascinated by sports betting. And yet, I never excelled at school, in fact I was the exact opposite of a good student.

You don't need to have had a good education to be a successful sports bettor but you do need to gain one. The ability to learn and research is important. At school I learned nothing. The classical schooling system never worked for me.

I wasn't a bad pupil. I tried my hardest but I wasn't good at anything because I couldn't concentrate when people spoke to me. However, I am good at reading. Through reading I learned what I wanted to know for myself in the style and pace that suited me. I left school with few qualifications because I read things that 16 year olds don't need to know to pass their O-levels and yet I still ended up at Oxford University many years later.

The greatest invention during my lifetime must be the World Wide Web. Although the web wasn't around until I was 27 years old. A web browser is the best research tool available. We must all learn how to use it effectively. And by effectively I mean finding things out for yourself and not just going to a forum and asking, "How do you...?" because more often than not nobody is going to tell you and you've just wasted valuable time.

If you are weak at mathematics (statistics and probability theory especially) then learn it online. With sports markets ever more similar to financial markets you should use financial trading websites to help you understand betting markets more. Both markets work using the same principles of supply and demand. You must also learn some basic programming techniques to increase your work flow.

Read. Learn. Experiment. And keep on doing it until you succeed.


The term framing is one from psychology and communications theory and leads into other terms such as cognitive bias. This is a useful subject to study for betting and trading as framing was a problem of mine when I first became a trader.

One of my main faults as a trader in oil and index futures was the inability to stop looking at the prices. A sudden reversal in the price would have me making the opposing trade to close out for a loss. Minutes later the price would correct and continue trending as I had expected it to when I put the first trade in.

I have corrected this fault of mine through automated trading. All I have to do is come up with back-tested trading rules and leave my bots to do the work of placing trades. With this hands-off approach there is no chance of my risk aversity from making a mess of things.

In the case of commodities trading I set up rules to trade on moving averages that looked at prices every so often. By permitting myself to look at prices between the time frame of the moving average I was disobeying the trading rule. There is a lot of noise in time series data and allowing that noise to hide a trend will destroy any winning trading rule.

Another use of the term framing is in the framing of a gamble. For example, if I say, "You will lose £50 if you call heads but you might win £100 if you call tails" then that may have the risk averse declining the bet. The thought of losing overrides the positive expectation of the gamble. If we analyse the gamble we see that the expectation is indeed positive.

Expected Value = (pWin * sWin) - (pLose * sLose)

where p is the probability and s is the sum to be won.

In our example above the expected value is

(0.5 * 100) - (0.5*50) = 50 - 25 = £25

In the time frame of 1 toss of the coin you might worry about losing £50 but over the time frame of 100 tosses you might expect to win in the region of £2500. This is an example of cognitive bias caused by irrational thought.

People who manually trade tick data can quite easily get caught by the framing effect. A sudden news item can spook a market one way or another and there will be a major price change before the market settles down. Doing this repeatedly will make a winning trading rule turn into a losing one.

Moving averages or other such sampling algorithms can smooth out the noise and generate manageable frames to trade upon. For those with 3rd party Betfair trading software, slowing the rate of price updates down from say 1 second to 5 seconds (or slower) can make a difference. Now, all that is required is a successful rule to trade with.

Further Reading

Leighton Vaughan Williams - citations

Professor Leighton Vaughan Williams is the UK's answer to Professor William Ziemba. Vaughan Williams too is a prolific writer in the field of sports gambling. He can also be seen on television or heard on the radio when a bite is required on the matter of sports prediction and betting.

Some of the following citations can be found on Google Scholar and downloaded, in PDF format, for your research. Papers not available for download will have to be photocopied from the cited publication at a university library. You might also try JSTOR, a repository for many journals, but this will require a subscription fee to perform downloads.

The professor has also written books on sports betting and more scholarly works on sports/finance.

Parke, J., et al. "An Exploratory Investigation into the Attitudes and Behaviours of Internet Casino and Poker Players." Report commissioned by eCOGRA (2007).

Williams, Leighton Vaughan. "Can Bettors Win?." World Economics 2.1 (2001): 31-48.

Smith, Michael A., David Paton, and Leighton Vaughan-Williams. "Costs, biases and betting markets: new evidence." Nottingham Trent University, Nottingham Business School, Economics Division Working Papers 2004/5 (2004).

Paton, David, and Leighton Vaughan Williams. "Monopoly rents and price fixing in betting markets." Review of Industrial Organization 19.3 (2001): 265-278.

Williams, Leighton Vaughan. "Can forecasters forecast successfully? Evidence from UK betting markets." Journal of Forecasting 19.6 (2000): 505-513.

Smith, Michael A., and Leighton Vaughan Williams. "Forecasting horse race outcomes: New evidence on odds bias in UK betting markets." International Journal of Forecasting 26.3 (2010): 543-550.

Paton, David, Leighton Vaughan Williams, and Stuart Fraser. "Regulating insider trading in betting markets." Bulletin of Economic Research 51.3 (1999): 237-241.

Vaughan-Williams, Leighton. "The economics of gambling." (2005).

Williams, Leighton Vaughan, ed. "Prediction markets: Theory and applications." (2011).

Williams, Leighton Vaughan, and David Paton. "Does information efficiency require a perception of information inefficiency?." Applied Economics Letters 4.10 (1997): 615-617.

Smith, Michael A., David Paton, and Leighton Vaughan Williams. "Do bookmakers possess superior skills to bettors in predicting outcomes?." Journal of Economic Behavior & Organization 71.2 (2009): 539-549.

Paton, David, and Leighton Vaughan Williams. "Forecasting outcomes in spread betting markets: can bettors use ‘quarbs’ to beat the book?." Journal of Forecasting 24.2 (2005): 139-154.

Paton, David, and Leighton Vaughan Williams. "Do betting costs explain betting biases?." Applied Economics Letters 5.5 (1998): 333-335.

Williams, Leighton Vaughan, ed. "Information efficiency in financial and betting markets." (2005).

Smith, Michael A., David Paton, and Leighton Vaughan Williams. "Market Efficiency in Person‐to‐Person Betting." Economica 73.292 (2006): 673-689.

Williams, Leighton Vaughan, and David Paton. "Why are some favourite-longshot biases positive and others negative?." Applied Economics 30.11 (1998): 1505-1510.

Williams, Leighton Vaughan, and David Paton. "Why is there a favourite-longshot bias in British racetrack betting markets?." Economic Journal 107.440 (1997): 150-58.

Williams, Leighton Vaughan. "Information efficiency in betting markets: A survey." Bulletin of Economic Research 51.1 (1999): 1-39.

Tennis - citations

Tennis (along with football) is one of the most popular sports for in-play trading. There is a wealth of freely available data for which to build trading models.

In-play trading allows the suitably equiped trader to watch and trade the changing fortunes of tennis players whilst matches are in progress.

These academic papers can be found on Google Scholar, JSTOR and in journals stored in university libraries.

Knottenbelt, William J., Demetris Spanias, and Agnieszka M. Madurska. "A common-opponent stochastic model for predicting the outcome of professional tennis matches." Computers & Mathematics with Applications 64.12 (2012): 3820-3827.

Barnett, T., A. Brown, and S. Clarke. "Developing a Model that Reflects Outcomes of Tennis Matches." Swinburne University (2005).

Easton, Stephen, and Katherine Uylangco. "Forecasting outcomes in tennis matches using within-match betting markets." International Journal of Forecasting 26.3 (2010): 564-575.

Øvregård, Øyvind Norstein. "Trading" in-play" betting Exchange Markets with Artificial Neural Networks." (2008).

del Corral, Julio, and Juan Prieto-Rodriguez. "Are differences in ranks good predictors for Grand Slam tennis matches?." International Journal of Forecasting 26.3 (2010): 551-563.

The Numbers Game: Why Everything You Think You Know About Football is Wrong

Not being a fan of football (I don't like games that can end in draws, although cricket is worse) it has taken me some time to get round to reading The Numbers Game

However, a fascination with statistics and the inevitability of looking at the mathematics of in-play football trading meant that eventually I had to read the book. Most football fans could do with reading this book too as they could learn a lot from it.

The usual terrace chants of "Sack the manager!", "Get it in the net!" and "We need a new striker!" should be replaced with "We need a sports statistician!", "Keep it out of the net!" and "We need to replace the weakest link!" but probably they don't make for good chants.

Football as any statistician/scientist will tell you is a badly designed experiment [1]. Goals are a rarity compared to other sports and so luck plays its part. This is why football is played over a long season where the luck evens out over the games so that the richest club wins the league (another statistic in the book).

The game is slowly but surely dawdling its way into the 21st century with in-depth match analysis by forward thinking coaches like Roberto Martinez, Brendan Rodgers and their backroom staff of mathematically minded helpers. Not to mention companies such as OptaSports who provide subscribers with a wealth of data with which to analyse games and improve teams. The future is big data.

A football team is only as good as its weakest link rather than its best striker is a conclusion in the book. So when Real Madrid started its Galacticos programme many years ago it was the weaker players in the team that prevented Los Blancos from getting La Decima until recently.

The media and fan's fascination with forwards and talented mid-fielders means that they are the higher paid players who are always in the news (sometimes for football related matters) rather than the equally important defenders and goal keepers keeping the ball out of the net. A good team is balanced to score goals but also to keep a clean sheet too.

This book is filled with interesting statistics about the beautiful game that might make football fans think twice before shooting their mouths off (okay it won't) but we can hope. The book will also provide some interesting research avenues for the sports trader.


[1] Skinner, G.K. and Freeman, G.H. "Are soccer matched badly designed experiments?" Journal of Applied Statistics Vol. 36, No. 10, October 2009, 1087-1095

See also

Amazon -  The Numbers Game: Why Everything You Know About Football is Wrong

Book reviews

Football research citations

All football related articles

Conditional Logsitic Regression for Horse Race Prediction

Some of you may have seen William Benter's video where he describes some of his work in the Hong Kong horse racing betting markets. Briefly he touches upon the statistical methods used.

In this video you get a better idea of a similar method using conditional logistic regression in a lecture given by Dr. Noah Silverman at UCLA.

Comparisons are also made between parallel processing on multi-core CPUs and massively parallel processing with GPUs. Something we will get round to on Betfair Pro Trader.

Experiences of a hedge fund quantitative developer

A very interesting article by Michael Halls-Moore (an ex hedge fund quant developer) who runs the QuantStart website. The site contains a lot of useful articles that are pertinent to sports betting quants and bot developers.

In one such article, entitled My Experiences as a Quantitative Developer in a Hedge Fund, we get to understand the work cycle of a quant developer. First, and foremost, is building time series data and cleansing it, then performing statistical research on the data to find trading signals. Finally, the routing of a trade is also handled by quant developers. With sports betting now a globalised industry there is much to be learned there.

Michael also appears on a video that I posted a few months back entitled What is a Quant Trader? Again, lots of useful insight that would be of interest to people in sports betting. I recommend that you read the QuantStart website.

See also

What is a quant trader?

Mathematics of gambling