Ref: 98/201a 21 December 1998

Southampton scientists use a commodity supercomputer to reveal how Lottery players choose their numbers

Using techniques normally applied to the study of large numbers of atoms, a group of Southampton scientists has conducted an analysis of the numbers chosen by lottery prize winners to assess how players make their selections. While there is no scientific way of predicting the actual winning numbers in advance, their study has shown that by choosing unpopular sets of numbers, players could significantly increase their chances of not having to share the jackpot if they did win.

Dr Simon Cox and Dr Denis Nicole of the Department of Electronics and Computer Science, together with Dr Geoff Daniell of the Department of Physics and Astronomy, combined ideas from statistical physics and the power of a supercomputer to discover patterns in the way we fill in our lottery tickets.

Using a concept called entropy they have shown that players prefer certain numbers; for example, seven, the most popular number, is chosen 25 per cent more often than 46, the least popular.

In order to perform these large computations, the team in the High Performance Computing Centre built a supercomputer from commodity PC components. Eight Compaq/Digital Alpha PCs running Microsoft Windows NT 4.0 Workstation and connected by switched fast Ethernet were used. The Compaq/Digital Alpha processor outperforms the Pentium II, and is the only multi-source, commodity CPU optimised for floating-point calculations each 500MHz processor is rated at one Gigaflop. The current cost of the system is around 1500 pounds per processor, a factor of eight less than the equivalent specialist supercomputer; it is one of the largest computational resources at the University.

Working with Mississippi State University (MSU), the team at Southampton were able to run the first ever FORTRAN scientific applications using the standard supercomputer Message-Passing Interface (MPI) on Alphas running Windows NT. The Digital Visual FORTRAN compiler was used as it provides an excellent integrated development environment, complete with a full set of scientific mathematical libraries and OpenGL graphics support.

Further work is focussing on building bigger and better commodity supercomputers. Windows NT is being used as it is a high-quality, user-friendly environment that provides the same scientific application performance as UNIX-based operating systems. In addition to using the computational cluster for running scientific applications, the team is developing reliable remote access mechanisms, UNIX interoperability and lightweight communications for Windows NT and Windows 2000.


The team would like to thank the Engineering and Physical Sciences Research Council, Microsoft Research, MPI Software Technology and 3Com for their support of this work.

Notes for editors:

The research is published in, S.J.Cox, G.J.Daniell and D.A.Nicole, "Using Maximum Entropy to Double One's Expected Winnings in the UK National Lottery", J.R.Statist.Soc.D 47, 1998, 629-641.

Details of the computational aspects are published in, Cox, S.J, Daniell, G.J. and Nicole, D.A., "Maximum Entropy, Parallel Computation and Lotteries", Proceedings of the 1998 International Conference on Parallel and Distributed Processing Techniques and Applications, Las Vegas, 1998

This work has been featured in the following general press: Sunday Telegraph, Meridian TV, Talk Radio, BBC Radio Leeds, Southern Daily Echo, Wave FM and Ocean FM.

For further information:

Dr Simon Cox, Department of Electronics and Computer Science, University of Southampton (01703) 593116; email

WWW site -

Sarah Watts, Public Affairs, University of Southampton (01703) 593807