What's Wrong With High-Frequency Trading?

To the Woodshed On Christmas Eve, Yale chief investment officer David Swensen smacked high-frequency trading. His New York Times article (co-authored with law professor Jonathan Macey), "One Way to Unrig Stock Trading," accused high-frequency traders, or HFTs, of benefiting from "rigged markets." The language deliberately echoed that of Michael Lewis in his 2014 best-seller Flash Boys.

Some background--

HFTs analyze trade data to anticipate stock-price movements. Using trade orders placed a fraction of a second previously, HFTs buy stocks that their computer algorithms determine will rise in price and sell those that figure to decline. (Naturally, these trades are all triggered automatically.) There are no guarantees that stocks will behave as expected, and the gains for each successful trade tend to be small, but a modestly positive winning percentage plus high volume can generate healthy profits.

The key is to get there first.

In the grand scheme of things, there's not a lot of money to be made from high-frequency trading--an estimated $3 billion annually, on roughly $40 trillion of stock market trading volume, meaning that HFTs' aggregate revenues are something less than 1/10,000th of dollars traded. For perspective, mutual fund company revenues are about $90 billion, making the fund industry far, far larger than HFTs.

Thus, the strategy succeeds only for the relative few who build their platforms to be microscopically faster than others’. Critically, these winning companies supplement their technological advantages--including the extreme step of locating their computer servers near stock exchanges to take advantage of the shorter distance traveled at light speed by electronic connections--by purchasing insider access. That is, they buy multi-million-dollar data feeds from the stock exchanges that give them trade information milliseconds before other parties.

Three Strikes Swensen levied three charges against HFTs, which I will address in order of accuracy.

1) HFTs cheat.

Yes, they do.

In purchasing "privileged access to market data," to use Swensen's words, HFTs receive favorable treatment from stock exchanges. Without that information, they would not be able to front-run other investors' trades and their profits would disappear. Some HFTs also pay to locate their computer servers inside stock exchanges to minimize data-transmission times.

Wrote Swensen, “These co-located computers detect orders to buy and sell on one exchange and then rapidly send cancelations and orders to other venues where their servers are also co-located. Does this sound like a fair system?”

Hmmm. I don't think that particular example serves him well. If the computers were located across the street from the various exchanges, that wouldn't much change things--the HFTs could still use information gathered from one exchange to send orders to other exchanges. And what's wrong with that? Buying in one place and selling in another because of small price differences in the two marketplaces would seem to be a good thing. Arbitrage improves efficiency.

However, I wholeheartedly agree with Swensen that the stock exchanges should not play favorites. Everybody who trades on a stock exchange is that exchange’s customer. To give the biggest customers a special, faster feed for a special, larger price so that they can profit from the trades of smaller customers is ... unseemly. At best.

As did Lewis before him, Swensen supports a new exchange, IEX, that refuses to sell custom feeds to favored customers. In Swensen’s words, “IEX’s plan is to forgo the high profits earned by the major exchanges from selling speed advantages on the theory that they can make money more ethically by attracting long-term investors.” Yes, that’s a bit high-handed; one can almost see IEX sitting astride that white horse. But I can’t argue with the sentiment.

2) HFTs harm liquidity.

Maybe they do, maybe they don’t.

Wrote Swensen, “Market depth, critically important to investors who trade large blocks of securities, also suffers in the world of high-frequency traders. Startling evidence for the lack of robustness in today’s market comes from a 2013 Securities and Exchange Commission report that found order cancelation rates as high as 95% to 97%, a result of high-frequency traders’ playing their cat and mouse game. Market depth is an illusion that fades in the face of real buying and selling.”

Obviously, I am in no position to dispute what the manager of a $25 billion pension fund says about market liquidity. (If I put an immediate sell on my entire portfolio, no HFT algorithm would even notice those transactions, aside from my stake in relatively little-traded Morningstar MORN stock.) But I can point out that reasonable and informed people have disagreed.

Vanguard CEO Bill McNabb, for example, said last year that HFTs have improved market liquidity and benefited Vanguard shareholders. Before that, in 2011, in the first independent academic study of HFTs, which was published in The Journal of Finance, three professors found that "for large stocks in particular" high-frequency trading "narrows spreads, reduces adverse selection, and reduces trade-related price discovery." In short, high-frequency trading "improves liquidity."

Swensen does not have a history of being wrong. Nor, however, does Vanguard or The Journal of Finance. One side will eventually be shown to have been in error. At this time, however, we do not know which side that is.

3) HFTs hurt the small investor.

Not in any meaningful way.

Swensen played this card briefly, writing "Individual investors, trading through brokers like Charles Schwab, E-Trade and TD Ameritrade, suffer first as the brokers profit by hundreds of millions of dollars from selling their retail orders to high-frequency traders and again as those traders take advantage of the orders they bought." Lewis played those cards much more forcefully, again and again.

It's a bogus contention.

For one, the amounts are small indeed.

The Journal of Finance

article estimated that HFTs made about $7 for each $100,000 of stock trades. That figure has since declined, to possibly as little as one-twentieth of a cent per share. Thus, a retail investor with a $500,000 stock portfolio and an annual turnover rate of 40% would have indirectly paid about $14 to HFTs a few years back and a lower amount today. To once again place this into perspective by citing mutual fund expenses, the annual dollar fees paid to Vanguard on a $500,000 investment in its low-cost

Plus, to the extent that HFTs have helped to reduce bid-ask spreads on stocks--an argument that even Swensen did not directly deny--that change is to the benefit of all investors, the lowly retail buyer included. It's quite possible, if not directly provable, that HFTs have lowered the cost of trading for everyday investors. If that is predatory behavior, then by all means, prey away.

Swensen quickly backed away from populism to stress the more accurate point--the fight over HFTs is not about Wall Street versus Main Street. It is instead an insider battle between two factions of institutions, each with their own set of interests. (Lewis, of course, did not similarly back away, ringing the populist bell throughout his book.) In that quarrel, I side with Swensen; justice is indeed his partner. However, I'm not sure that the unjust's current victory meaningfully harms either the overall market structure or my investment returns.

John Rekenthaler has been researching the fund industry since 1988. He is now a columnist for Morningstar.com and a member of Morningstar's investment research department. John is quick to point out that while Morningstar typically agrees with the views of the Rekenthaler Report, his views are his own.