Machines rule the trading world, and are getting even faster, leaving us wonder if humans still play a necessary role in trading.
On August 1st, 2012 at 9:30 am, in a matter of only 45 minutes, Knight Capital Group lost $440 million due to a computerized trading system malfunction. Five days later, a consortium of four firms rescued Knight and diluted shareholders, with the company’s stock dropping from $10 to $3. A victim of their own creation, the machines rule the trading world, and are getting even faster, leaving us wonder if humans still play a necessary role in trading.
Knight is a trading firm and one of the largest market makers in a number of stocks in the United States. Its high-frequency trading system malfunctioned that morning, or was accidently tested before it was ready[i], as it simultaneously entered buy and sell orders in the same 148 stocks, sending prices fluctuating wildly. According to reports[ii], trading volume on the New York Stock Exchange (NYSE) was six times more than average.
Through the 1970s, the most sophisticated mode of communication for the financial world remained the same for nearly 100 years. The telegraphic ticker system gave prices of stocks and the telephone allowed the communication of orders until then. Now, high speed servers send orders through fiber optic cables with speed as the main objective; one millisecond means the difference between profit or loss.
High frequency trading (HFT) programs make money in multiple ways , usually less than a penny per share at a time (as stock quotes are to the penny but real prices are up to six decimal places). Sometimes the programs simply provide liquidity taking the other sides of trades and getting rebates from the exchanges. In the more dubious ‘flash trades,’ programs ‘flash’ buy or sell orders to pinpoint the limit price orders of other participants and then quickly soak up orders until the limit is filled at the most profitable levels for the HFT program.[iii] The brainpower for the trading comes from the mathematics and physics PhDs applying a variety of theorems from other scientific fields to financial data in a lighting fast manner.
Speed is the number one concern among these program traders. The following charts from Nanex and Zero Hedge provide a great visual as to what it is like trying to compete with these machines. The first image is a massive jump in price for this particular instrument. The entire “spike” took only 275 milliseconds to complete. This image represents what a human sees during this spike.
The next picture shows what the High Frequency Trade algorithms sees during that same 275 milliseconds time frame during the “spike” from the image above.