Can We Stop Overreacting To Computerized Trading?

Such are the crazy times we live in, where a single rumor can cause the entire system to crash, even momentarily.

|
Apr 24 2013, 2:00pm

After a fake tweet caused markets to temporarily erase $136 billion worth of paper value yesterday, critics quickly took the opportunity to blame high frequency trading, whereby pre-programmed algorithms trade at lightning speeds.

A Bloomberg article aggregating the FUD quickly became the site’s most read piece of the day.

“It’s one thing for an illiquid stock to do that but how does a multitrillion-dollar market do that?” wondered Walter Todd, a chief investment officer that manages nearly $1 billion. “That’s very disturbing to me. It’s unnerving.”

“Don’t let computers rule your investments,” adds Barry Schwartz, a fund manager that oversees half a billion dollars worth of investments.

There's truth to their criticism. If a human had seen this questionably worded tweet: "Breaking: Two Explosions in the White House and Barack Obama is injured," he or she may have felt suspicious or at least the need to confirm news of such magnitude. Computers, at least for now, aren’t so discerning, and in many cases are designed to react to the market itself, even if the market is being shaken by fake news. That’s why the Dow Jones Industrial Average quickly plummeted 150 points in mere minutes before immediately bouncing back once everyone realized the AP’s Twitter account had been hacked.

As Motherboard’s Brian Merchant noted, such are the crazy times we live in, where a single rumor can cause the entire system to crash, even momentarily. But others clearly saw it as an opportunity to take potshots at computerized trading.

If we’re being honest, that used to be me. When the terror of high frequency trading first started breaking into mainstream consciousness, I too felt concern, in part because I didn’t fully understand it and in part because shit is scary whenever you concede control, especially to emotionless robots with the power to crash markets in the blink of an eye.

But the more I thought about it, the more I came to terms with Wall Street’s robot overlords. This is the essence of technological progression. It’s always a double edged sword. Even as Facebook connects us in brand new ways, there are some who have never felt more alone. The bountiful energy that nuclear power provides comes with the uncertainty of catastrophic meltdown. In play is the universal balance between risk and reward.

"All news is considered by 'the market' in the context of their credibility, and obviously the AP has as much clout as anybody," a trader told me.

Part of the problem is our obsession with speed, where we have now arrived at a point where trades happen far too fast for any reasonable sort of human intervention. Yet for humans, it’s an age-old attraction in our never-ending need for efficiency. To be certain, when a high speed train traveling at 200 mph crashes, the odds of survival are exponentially lower than if you had taken a steam locomotive moving at snail's pace. But if both options are available for a trip between New York and D.C., most will end up trading the difference in safety for what amounts to a significant difference in time saved.

There will always be those whose fear of flying prevents them from taking the more efficient option, but overall, these are the kinds of risks we’ve grown accustomed to accepting. If such bravery weren’t ingrained in our genetic code, we would have never arrived on the moon and Mars would always be the stuff of science fiction. Instead, everyday, we look to go higher and faster. In the world of finance, things are no different.

If we were to rewind history, back to the days of human specialists on the floor of the New York Stock Exchange, we’d discover that such rumors would have a similar impact, although perhaps not as drastic. What would you do if a credible friend suddenly told you the president had been injured in a terrorist attack? There would be no cell phones or internet to immediately confirm the veracity of the report. Do you sell? Do you wait? Even with humans, such rumors can cause confusion and uncertainty. While the valleys weren’t as deep, they were, by definition, much wider.

Perhaps that is the ultimate testament to the resiliency of the current system, that markets self-corrected in a matter of minutes. The fall may have felt chasmic but it was only a tiny blip in the grand scheme of things. In a way, it all happened just as we would have expected it to.

“The world reacted quickly and efficiently on the news,” a trader, who wished to remain anonymous due to company policy, explained to me. “Then when the news was reported fake the market corrected quickly and efficiently. Things are working like they should.”

And when the world operates at the speed of Twitter, it would be foolish to expect finance not to follow suit, especially when trillions of dollars are at stake.

“The reaction was 'normal' because it is exactly what you would expect to happen if the news were valid,” another anonymous trader told me. “All news/information bites are considered by 'the market' in the context of their credibility and obviously the AP has as much clout as anybody when it comes to reporting news.”

Remember, the root problem here isn’t the system or even the speed of information in the age of the retweet, it’s the AP’s suspect social media security.

“Certainly this episode might scare the actively participating investors/traders into distrusting all news sources until corroborated by a number of others,” trader #2 continued. “However, in an environment where speed is paramount and can make the difference between millions or billions of dollars, nobody wants to sit around and wait until a story reported by an otherwise totally credible news source is corroborated elsewhere.”

And like the tragedy in Boston, yesterday’s events were an outlier. Things went haywire for a bit but over the course of time, the system has worked out just fine. While we should respect the worst case scenario, it’s shouldn’t be the foundation of how we operate.

A few seconds of high-frequency trading, visualized. Via Nanex.

“While this incident may make the market a bit more skittish, I don't think a single incident will change the way the market would react to a credible news source tweeting about an event. If this happens again to the AP, they may lose credibility while other major news sources maintain theirs. If another major news organization's twitter is hacked and used similarly it may lead to a loss of credibility for twitter as a news source. I believe it will take much more than this one event to shake market confidence in twitter accounts of major news organizations as a source of news.”

That isn’t to say safeguards to prevent a complete system crash shouldn’t be put in place. But as much as it may be something to strive for, like many complex problems, there is no absolute solution.

“There are certain "black swan" events and other shocks that you simply can't write a program to respond to correctly,” explains trader #2. “How could you write a program that processed all of the news today and traded accordingly? You simply can't think of every case possible. Since you can't predict every outlier event, it seems like a reasonable idea to add some kind of clause to your algo that is something to the effect of "if the market is down X% in Y minutes, sell Z amount immediately and sound the alert for human intervention."

The potential winners and losers are those doing the trading. Invariably this means that after every such incident the algorithms get smarter.

But even that solution isn’t perfect. “If the news today had in fact been real this clause in an algorithm could have saved somebody heaps of money. In the case of today it may have actually caused significant losses, but this is the inherent risk of having computers trade for you, you are only as good of a trader as the program you have written.”

In the end, perhaps the most important takeaway is that what happened yesterday was economically irrelevant, given the speed by which the system fixed itself. Headlines, like that of Bloomberg, will emphasize how much value was “wiped out.” The key word is “momentarily.” Once the incident was over, there was a marginal real world impact. Regular people won’t feel a thing and their 401Ks are unaffected. “A flash crash and recovery doesn't necessarily guarantee a loss for any participant, but it certainly offers an opportunity for a lot of money to change hands,” said trader #2.

So the potential winners and losers are those doing the trading. What this invariably means is that, after every such incident, the algorithms get smarter.

Driving a McLaren Formula 1 car will always be scarier than flooring your mom’s Toyota Camry. And whenever when you’re operating at that kind of speed, there will be crashes, no matter how technically savvy the driver. But the race goes on.

Then again, when you control the rules of the game, it might make sense to enforce artificial speed or vehicle limits, as many racing organizations have done. For the sake of the humans.

Top image via Flickr/Howistheanswer

@sfnuop

Read more about high-frequency trading:

A Single Fake AP Tweet Can Apparently Crash the US Stock Market

What Computer-Powered High Frequency Trading Looks and Sounds Like

Brief Encounters with a Wall Street Arms Dealer

Stories