Risk, uncertainty and non-/knowledge in the financial market
Computerized trades from the perspective of sociology of finance
Abstract
This research project will analyze “algorithmic trading” or “high-frequency-trading”, a system-operated or computerized trading practice. Algorithmic trading is defined by a variety of distinctive features, which will serve as focal points for this research project. These novel forms include new forms of risks, uncertainties and non systems-operated knowledge, which are related to the emergence of qualitatively new forms of speed and acceleration of trading.
Paradoxically, this market practice can be described as a stretching of the present through acceleration. The stretching of present is crucial to one important goal of algorithmic trading to cope with uncertainties that emerge in a contingent future by attempting to keep the future as far away as possible. Whereas previous trading practices focused on various ways of forecasting the future – either through quantitative-statistical analysis or classical investment strategies – High Frequency Trading can be defined by its flight from future risks by means of infinitely exhausting the present moment: profits are sought through potentially infinite amounts of transaction within an infinite small amount of time.
However, in certain market constellations the operations of algorithms can result in domino effects that may tip the market out of balance. For some this seems to be even more dangerous than in a situation involving classic (human) actors, since (1.) the algorithmic trade avalanche can take place within fractions of a second and not only over a period of days and weeks; and (2.) the emergent behavior is entirely new and unforeseen, creating crises which we are only with hindsight beginning to understand. Thus, it is not surprising that a disconcerted European Central Bank uneasily describes this type of trading as “disorderly” and sees in it a possible “breaking of trading systems,” of which the so called “Flash Crash” on May 6, 2010 was just a precursor.
One of the main conundrums of this trading practice lies in the discrepancy between its initial aims and its unintended effects. By aiming to resolve problems of insufficient information and incorrect predictions (about future stock market movements) algorithmic trading excludes long term investments and unreliable (human) factors but at the same time it creates entirely novel risks and uncertainties as well as new forms of (human) non-/knowledge.
All the differing explanations of what algorithmic trading is circle around one fact about these strange phenomena. Whatever their cause, human or machinic, they contain in themselves implacable black holes of knowledge. These are cases where non-/knowledge cannot simply be banished by knowing more, or knowing more things faster. Or rather, there is a slippage in the kinds or modes of knowledge they involve. On the one hand, there is the concatenation of an indiscernibility of algorithmic variances and responses to changes in the price/volume of overall trade that generate new forms of non-/knowledge. On the other hand, the apparent success of algorithmic trading practices indicates that the interacting machines assemble, operate and trade information correctly, which means they ‘know’ what they are doing.
I argue that it’s never before witnessed integration of material, human, and ideational features makes algorithmic trading a challenge to prevailing sociological concepts of rationality, risk, and institutionalized trading. A first analysis shows a variety of different paradigms can be distinguished in current research. I will explain key features that make algorithmic trading a distinctive and novel social formation, and show how my proposed multi-paradigmatic approach sheds new light on algorithmic trading.