Because they work so fast, what’s to stop robotic trading agents from over-reacting and getting things horribly wrong?
Trading at the speed of light
 | |
 |
The ten seconds it takes a good financial trader to react to a pricing change on a screen and execute a trade is a lifetime for an algorithmic trading agent—the blindingly fast software routines used by investment banks to execute buy and sell decisions autonomously. Now, having made fortunes for financial institutions, robotic traders are poised to conquer new fields. Expect to see them cropping up in other brain-intensive activities—from marketing to education, from battle-field analysis to writing software.
Read the full article, click here
Listen to the podcast: Working at warp speed
| |
 Approx. 10 mins |
Professor David Cliff, a former director at Deutsche Bank and now head of the Science and Engineering Natural Systems Group at Southampton University, explains why autonomous trading agents have become so successful.
Listen to the podcast, click here
To save the MP3 file, right-click the link and use the "Save Link As" option.
Total Posts: 2 - Pages (1): [1]
Well traditional economics is based on the assumption of rational agents, however the real world seems to prove otherwise. Most agent based paradigms tend to be modelled very closely to the real world, hence intelligence which is a function of rationality is bound within a certain range.
Thus as intelligence is not absolute there is the risk of an agent over-reacting, hence an agent will get things wrong due to the limitations of its model rather than speed of execution. The speed of execution is infact a desirable characteristic which has been proved to beat human traders.
Thus as intelligence is not absolute there is the risk of an agent over-reacting, hence an agent will get things wrong due to the limitations of its model rather than speed of execution. The speed of execution is infact a desirable characteristic which has been proved to beat human traders.
Total Posts: 2 - Pages (1): [1]