Yet another “flash crash,” but why?

Although not on the same scale as the $800 billion May 2010 “flash crash,” markets were again spooked this week when some $150 billion in market value disappeared in an instant only to reappear five minutes later.

The apparent explanation is that hackers associated with the pro-Assad regime “Syrian Electronic Army” hacked into the Associated Press’ Twitter account. They then sent out what looked to followers of the AP to be a legitimate tweet stating that there had been a terrorist attack on the White House and that President Obama himself was injured.

A flurry of sell signals hit the markets and with buy orders disappearing there was a free fall in prices. Only when some traders expressed doubts and others made calls to friends in D.C. to try to confirm the tweet did buy orders reappear. The market quickly recovered.

Although many were quick to blame Twitter for being too easy to hack, it also seems likely that the dominance of algorithmic “high frequency” traders is at least partly responsible. Twitter is now fed into the computer programs such traders use to make trading decisions. If the proprietary algorithm receiving the AP tweet was set up to sell in reaction to a serious terrorist attack that could have started the cascade of sell orders. Other “algos” might have been set up to sell upon news that there was a sudden increase in sell orders and the crash would be a logical result.

Only the intervention of sceptical humans led to discovery of the actual situation and put an end to the crash.

Our research suggests that a deeper problem is now at work in the markets. The SEC put in place a set of rules known as Reg NMS (for “national market system”) in 2007 that created a favorable environment for HFTs and for new trading platforms such as “electronic communications networks” (ECN’s).  As a result the trading floor of the New York Stock Exchange, with its longstanding use of human “specialists” to provide liquidity, has lost a significant amount of daily trading volume to these ECN’s where trading between computers dominates for most of the trading day. We find that bid-ask spreads, a proxy for risk, for shares listed on the NYSE increase significantly when those shares are traded on alternative trading platforms.

The result, we believe, is a more fragile and volatile market. While the new trading platforms may offer lower costs to traders and greater speed of execution, these come at the cost of less efficient markets that are subject to runaway events like this week’s flash crash.

Unlike the old NYSE’s specialists, the HFTs who trade on ECN’s are not obligated to provide support to a crashing market. Today many view the old NYSE as an archaic club that took advantage of investors. In fact, specialist firms often put their own capital at risk trying to meet their obligation to provide liquidity. Many such firms went out of business trying to stem the tide of the Dow’s “Black Monday” in 1987.

HFT’s are unwilling to provide this stabilizing role in the markets. Today’s investors live in a world that resembles that circus ride that spins around and around as the the floor drops out from underneath the riders.

The NYSE is set to be sold to new owners who promise to attempt reforms that strike a new balance more favorable to human traders. They should pay close attention to this week’s events but more may be required. A starting point may be to reconsider Reg. NMS implemented by the SEC in 2007.