Date/Time
Friday
27 Oct 2017
4:00 pm - 5:30 pm

Location
470 Stephens Hall

Event Type
Colloquium

Donald MacKenzie
Professor of Sociology, University of Edinburgh

A major challenge in social-science research on algorithmic economic life is how to think about the interaction among trading algorithms. This talk will discuss this issue, drawing upon extensive interviews with practitioners of ultrafast, automated ‘high-frequency trading’ or HFT. It will explore the central divide within HFT. Some algorithms, and indeed some HFT firms, specialize in “making liquidity”: in adding orders to the electronic order books of exchanges and other trading venues. Other algorithms and firms specialize in “taking liquidity”: in executing against existing orders. The divide is simultaneously material (the two sorts of firms have somewhat different technological priorities), legitimatory (some practitioners of “making” view “taking” firms as morally dubious), and cognitive (different “styles” of mathematical modelling are involved).

After sketching these aspects of the divide, the talk will identify two crucial forms of the interaction between “making” and “taking” algorithms. The first arises when market data clearly signal a coming price rise or fall. “Making” algorithms race to cancel their orders; “taking” algorithms race to execute against these orders before they are cancelled; the result is that some algorithms now have less than a millionth of a second in which to respond to at least the simpler forms of signal. The second arises when sophisticated algorithms, increasingly based upon machine learning, seek to predict the behavior of less sophisticated algorithms – an activity made more feasible by the reduction of “jitter” (unpredictable fluctuations in the speed of processing) in exchanges’ technological systems.