ECS to develop automated agents for the stock market

Professor Nick Jennings, ECS Professor of Computer Science, led a team which won the Trading-Agent Competition (TAC) – CAT Tournament (http://www.marketbasedcontrol.com/blog/index.php?page_id=5) which was held in Toronto in July. The other team members were Dr Perukrishnen Vytelingum, Dr Ioannis Vetsikas and Dr Raj Dash. The team is now developing this automated stock market further.

The aim of the competition was to create software that would control a stock market, populated by automated trading agents, in the most effective way. In particular, this involved adjusting the market parameters (the cost of listings, the amount of information given out to traders and the fees charged) to create a market that is more appealing than the others that are available.

The ECS team, IAMwildCAT beat nine other teams, with a score four times higher than the team with the lowest score. The overall outcome is judged based on how the competing markets increase their market share; increase the amount of profit they make and the transaction success rate over a number of trading days.

‘There are now lots of different stock markets/exchanges around the world, many of which have shifted online,’ said Professor Jennings. ‘All of these exchanges compete with one another for business and the one which works best with its different parameters will get the majority of the business. This was the case for ECS in the CAT competition. We won because we adapted the terms and conditions of our exchange to respond favourably to changing market conditions.’

This is the second computerised agent competition that a team led by Professor Jennings won this year, in July ECS also won the RoboCupRescue World Championships (www.robocuprescue.org) which was held in Atlanta.