Investors in three of the biggest Dow Jones Industrial Average (INDU) stocks were whipsawed by price swings that repeated every hour yesterday, fueling speculation the moves were a consequence of computerized trading.

Shares of International Business Machines Corp. (IBM), McDonald's Corp. (MCD) and Coca-Cola Co. (KO) swung between successive lows and highs in intervals that began near the top and bottom of each hour, data compiled by Bloomberg show. While only IBM finished more than 1 percent higher, the intraday patterns weren't accompanied by any breaking news in the three companies where $3.42 billion worth of shares changed hands.

Regulators have increased scrutiny of computerized strategies that have risen to prominence in the U.S. after more than a decade of market structure reform. The Securities and Exchange Commission and Commodity Futures Trading Commission blamed a broker's trading algorithm for setting into motion the events that caused the May 2010 market crash that briefly erased $862 billion from U.S. equities in less than 20 minutes.

"Somebody probably has software that's running an algorithm that's either selling in 30-minute intervals or buying," Bruce W. Weber, dean of the Alfred Lerner College of Business and Economics at the University of Delaware, said in a telephone interview. "For the market value of Coke to be going up and down in this way, oscillating every hour, is a pretty disconcerting observation. This is not going to raise investors' confidence in the mechanics of our market."

David Fuller's view Yesterday's trading swings mentioned above stand out clearly on Bloomberg's intra-day transaction (GIP) charts for Coca Cola, IBM and McDonald's. A different, less ultra short term but not all that dissimilar algorithm is probably responsible for the daily price swings that we are currently seeing in precious metal futures. Here are 10-day transaction (GIP) charts for gold and silver.

A casual observer might respond: 'So what? It is just market noise.' Well, yes and no. These little fluctuations for the three Dow components above are unlikely to trouble many people and might actually assist value investors. But what about when the price swings become much larger, as we saw most memorably in May 2010. Worse still, a considerable increase in overall volatility within most actively traded markets has been evident during the last five years.

This has driven many private investors to the sidelines - a development that cannot be good for markets. It has also made legitimate hedging operations by commodity producers and consumers much more difficult in futures markets. There is evidence, cited by financial articles and often discussed by Fullermoney, that some of the high frequency algorithmic trading (HFT) may actually be illegal, involving front running and other dubious practices. It certainly looks predatory, not unlike other practices conducted by large financial institutions, not least investment banks. (For earlier comments on this subject, use the 'Search' facility shown upper-left, fourth item down and search under HFT or high frequency.)

Why do these firms use HFT; why do exchanges allow it, and what might be the longer-term consequences?

Financial institutions, some of which only use HFT, are drawn to algorithmic trading because they hope to profit from it and an ever-increasing use of technology is part of our evolution. Clearly, some firms have made a great deal of money via HFT, at least until some other company comes up with a better algorithm, or market circumstances change.

Management likes HFT because a few automated trading systems devised by a clever programmer cost a lot less than a team of investment managers, not to mention a room full of high-octane traders, who have to be closely supervised if the next potential blow-up is to be identified in time.

Exchanges welcome HFT firms because they pay large fees for the right to locate their supercomputers where the action occurs, in hope of gaining a nanosecond's advantage. Exchanges also like the increased turnover which can account for anywhere from 40 to 80 percent of daily volume.

The longer-term consequences of HFT may be a matter of conjecture, but the extent to which it deters traditional investors and traders should ring alarm bells among regulators. Also, trading algorithms are just another 'black box' system, albeit with quicker reaction times.

If mechanised trading systems were an investment panacea, would a once thriving hedge fund company such as Man Group be struggling today? Trading systems can be devised to do spectacularly well in the last market cycle. However, when the cycle changes, as it inevitably does, the mechanised systems currently lack the AI to adapt sufficiently. Consequently, the programmer is always in a reactive position, unless he can work with a visionary genius (rare as hen's teeth) and that person would probably not be interested in the project.

Successful trading systems invite imitators which soon dilutes their combined performance. A successful system, once recognised will also give rise to other algorithms which anticipate the original signals. People who work with or near a successful trading algorithm will attempt to piggyback or even anticipate the signals, causing further dilution of results.

I do not doubt that smarter HFT algorithms will be developed, incorporating non linier calculations. Eventually, they will have sufficient AI to reprogramme themselves, at which point they will become terminator machines.

Meanwhile, we humble humans should remember buy-low-sell-high, and ride market trends while they remain consistent.