Trading in stocks and commodities has become a giant business of super computers (and programs) competing with each other. The machines are analyzing and determining what to sell or to buy within split seconds, often in multi-million dollar trades.
Every penny – or fraction of it – counts when the volume of the trade is large enough. It works well, except when the computer program is faulty, as just happened at Knight Capital Group, Inc., a company in the business of executing trades for others.
Computer Trading
It’s not the first time that computerized trading has run amok. Such problems are commonly termed “glitches.” In fact though, they are not “glitches,” meaning “transient faults.” Rather, they are inadequately designed and tested systems. The latest incidence is being blamed on new “updated” software. Well, it was updated for sure, but not for the benefit of the company. It lost a cool half billion dollars in less than hour, simply by buying high and selling low.
The Problem with Computers
The problem with computers is that they are extremely dumb. Computers are fast and accurate, but don’t understand what they are doing. They’ll make you a millionaire or a pauper with equal ease.
It’s been said “to err is human.” Everyone makes a mistake here or there. That is not the same as unleashing a new or revised computer program without considerable effort to ascertain its proper functionality. Even then, careful observation over an extended period is advisable as unforeseen circumstances may not have been considered in its creation.
Testing, Testing, and More Testing
Testing, testing, and more testing must be the order of the day. If all works out as desired, go back and do some more testing. Finally, when you think the product is absolutely perfect, run it on a parallel system, executing mock trades only, and see what happens.
Even if that works out OK, all bets are off. Good luck and good trading!
