Gone In 22 Seconds: Just How Frequent Is High Frequency Trading?

This is a story of journalists and economists, and the confusion that can ensue when they communicate. 

For me, the story starts with a book, Dark Pools, written by Wall Street Journal Reporter Scott Patterson. The book, published in the summer of 2012, is an account of the rise of high-frequency trading or “HFT.” That is automated trading of securities and derivatives using powerful computers driven by complex algorithms. It is widely accepted that trading activity in most markets is dominated by HFT. Firms that specialize in this practice trade in enormous volumes but start and end the day with few if any holdings. Their purpose is to profit from intra-day market moves, not from fundamental investment. They buy and sell in the blink of an eye, churning the markets constantly.

Mr. Patterson’s book contains the following passage:

At the end of World War II, the average holding period for a stock was four years. By 2000, it was eight months. By 2008, it was two months. And by 2011, it was 22 seconds, at least according to one professor’s estimate. 

At a briefing to an audience of congressional staffers on the proposed tax on financial transactions, I opined that such a tax would deter HFT, an outcome that, in my view, would benefit the stability and reliability of the markets. I mentioned the quoted passage in the course of the discussion. Of all of the things I said about HFT, that quote was the one that the audience seemed to remember best.

Also on the panel was the noted economist Jared Bernstein. Later, he mentioned that there was some controversy over the “22 second” estimate. That set me and some of my colleagues on a journey that is best described as surreal. It was also instructive.

The controversy centers on a blog post on HFT by Al Lewis, a Dow Jones Newswires columnist, that included this statement in reference to the “22 second” estimate:

When I tracked down the economist who supposedly came up with this astonishing number, he told me he was misquoted. "I was taking a nap and woke up and said something—too long ago to remember," he replied in an e-mail exchange.

This was strange on several levels. Foremost was the odd e-mail exchange that was recounted. But also, Lewis works for Dow Jones and Patterson for the Wall Street Journal, its affiliate. The blog was posted in February, half a year prior to the publication of Dark Pools. And, curiously, neither the article nor Patterson’s book provided the name of the economist.

The next step was to call Patterson. He was unaware of the blog post, but he filled me in on the economist. He is Professor Michael Hudson who has been described as “a Wall Street Financial Analyst, Distinguished Research Professor of Economics at the University of Missouri, Kansas City and author of” several books. The “22 second” statement was from an interview dated January 1, 2011 on the Real News Network that is available online. This is what he said: 

Take any stock in the United States. The average time in which you hold a stock is--it's gone up from 20 seconds to 22 seconds in the last year. Most trades are computerized. Most trades are short-term. The average foreign currency investment lasts--it's up now to 30 seconds, up from 28 seconds last month. So we're talking about really short-term. The financial sector is short-term.

I watched the tape. Professor Hudson was sitting in a studio, not waking up from a nap. And he was very precise in his statement. It was certainly not an offhand remark. And most definitely not misquoted by Patterson. I have no explanation for the Lewis comment, but it is obvious that he didn’t Google ”Michael Hudson 22 seconds” as I did.

I decided to get in contact with Hudson and get to the bottom of the story. On a phone call, I asked him whether he stood by the “22 second” statement and he said that he did. He told me that he first heard the number from traders. He said that he went on to check it by examining data from the New York Stock Exchange and the Chicago Mercantile Exchange that confirmed the “22 second” estimate. He qualified this by saying that the data was weighted toward highly active stock, like financials and technology, and that stock that was infrequently traded was less weighted.

We now that Professor Hudson did assert that the average holding time for stock is 22 seconds and that he still asserts it. We also know that his actual confirming research was somewhat narrower than his statement, but that does not change its basic meaning.

What is astounding is the power that a single startling number can have. The “22 second” estimate has been repeated many times and it captures the imagination. It should always be used carefully since Professor Hudson’s work is not published and so cannot be easily verified. But slight inaccuracies do not really matter. When trades are executed and unwound over time periods measured in milliseconds, the average holding periods drop fast. A 22 second estimate is startling, but no one reacts by saying that it is impossible.