Some stock market experts think the days of human traders are numbered, but an analysis of a recent trade in shares of Comtech Telecommunications Corp., Melville, N.Y., shows that might not be so.
The average daily volume for Comtech is about 460,000 shares, but on Aug. 16, a block of 100,000 shares crossed. Immediately after that trade, the stock dropped to $32 from $32.80 as a series of 100- and 200-share blocks traded.
Thorsten Schmidt, first vice president of the algorithmic unit at brokerage firm Instinet Group Inc., New York, said his firm crossed the 100,000-share block between two institutional clients he declined to name.
"As soon as the trade was printed and showed up in the market, there was a flurry of subsequent trades, all of which took place at very small sizes," Mr. Schmidt explained. "They were coming out at a very high frequency and were driving down the stock, which was trending down very, very quickly."
Because of the size of the trades and the speed of them, he and his colleagues deduced that they were likely coming from so-called volume participation algorithms, computer programs that trade a stock based on volume in the market.
"A human trader would probably not, in most circumstances, have been as aggressive" selling the stock after a big block crossed, Mr. Schmidt said, adding, "that's not to say all traders are great and all algorithms are bad."
He pointed out many algorithms, including those offered by Instinet, have "safety measures" that exclude blocks from the volume participation methodology.
Tom Whelan, first vice president of portfolio trading and algorithmic trading at Instinet, said it was difficult to quantify exactly how many times situations similar to what happened in Comtech occur. Still, "It's out there and we've seen it. This is really just one small risk component in one algorithm, so I would have to believe there are holes out there" in other trading algorithms.