All Hail Black Box Trading !

Black Box Trading, Papyrofix
To err is the prerogative of humans. However, computers are not granted this privilege, considered “smart” and “speedy” antidotes to most of our problems. The growing reliance on automation is testament to this fact. We readily put our faith in the hands of artificial intelligence, so much so that we are willing to trust it even with our hard earned money.

Black box trading or algorithmic trading has been finding stronghold in Wall Street in recent times. It is defined as “the process of using computers programmed to follow a defined set of instructions for placing a trade in order to generate profits at a speed and frequency that is impossible for a human trader.” Long gone are the days when profits hinged on the diligence, precision and agility of human decision-making.

Complex mathematical and statistical models, and computer programs are now combining to effect optimal investment strategies in financial markets that are executed at speed and frequencies that far surpass any human trader’s. With a set of predetermined conditions on price, quantity, volume, or other specifications, human intervention is at a minimum. 

Large institutional investors, investment banks, pension funds, mutual funds, and other buy-side institutional traders are typical users of algorithmic trading and automated trading systems (ATS) due to the sizeable volumes of trades they execute on a day to day basis; they divide large trades into several smaller trades to manage market impact and risk. Hedge funds are increasingly eschewing finance grads in favour of hardcore coders as complex algorithms slash trade monitoring time in half. Algo-trading is fast, improves market liquidity, eliminates large transaction costs and is devoid of the influence of human emotion. Since the world of finance ticks on high speed, reliance on such ATS does improve the chances of making profits by exploiting the best possible prices available in the markets. It provides a more systematic approach to active trading.

However, algo trades that are not backed by an investment strategy are bound to crash and burn. Trading strategies that may underline an algorithm can vary from trend following strategies to randomized complex strategies. Ideally, automation is advisable only for a clearly outlined trading strategy that will buy and sell on the basis of certain prerequisites that are preferably backtested. Since these codes are ultimately crafted by humans, they are not free from error. Backtesting of a trading system entails assessing the program on historical market data to ensure that the end results are compliant with the intended outcome of the underlying algorithm. Though backtesting is in no way an assurance of future performance, it is a means to evaluate the theoretical accuracy of the code.

Alternatively, forward testing an algorithm through trade simulation using real-time market data can help check its performance in the current market milieu. Often times, forward testing is an expedient means of identifying possible errors in the code. Apart from flawed codes, algo trading can also be plagued by system failure risks, network connectivity errors, time lags in execution, among other technical glitches. Due to a technical issue in their ATS, Knight Capital Group lost $440 million in a single day in August 2010.

However, the most concerning aspect of these computerized trading systems has often been pointed out to be its opaqueness or “black-boxness” itself. A large amount of data is fed in through these quantitative models and the output is not always intuitively justified. Yet, these methods have dramatically improved market liquidity and modified market microstructures as well. The key issue to keep in mind is, even automation is ultimately a human directive and it comes with its own set of risks and challenges.



For Papyrofix
By: Aashika Suresh