Computers are the new Dumb Money
Courtesy of Joshua M. Brown, The Reformed Broker
You want the box score on this latest weekly battle in the stock market?
No problem: Humans 1, Machines 0
Because if you think it was human beings executing sales of Starbucks (SBUX) down 22% on Monday’s open, you’re dreaming. And if you believe that it was thinking, sentient people blowing out of Vanguard’s Dividend Appreciation ETF (VIG) at a one-day loss of 26% at 9:30 am, you’ve got another thing coming.
By and large, people did the right thing this week. They recognized that JPMorgan and Facebook and Netflix should not have printed at prices down 15 to 20% within the first few minutes of trading and they reacted with buy orders, not sales. They processed the news about the 1200+ individual issue circuit-breakers and they let the system clear itself.
Rational, experienced people understood that an ETF with holdings that were down an average of 5% should not have a share price down 30%.
Conversely, machines can only do what they’ve been programmed to do. There’s no art, there’s no philosophy and there’s no common sense involved. And volatility-shy trading programs have been programmed to de-risk when prices get wild and wooly, period. Their programmers can’t afford to have an algo blow-up so the algos are set up to pull their own plug, regardless of any qualitative assessment during a special situation that is obvious to the rest of the marketplace.
Warren Buffett once explained that “Paradoxically, when ‘dumb’ money acknowledges its limitations, it ceases to be dumb.” Ordinary investors, in the aggregate, have learned their limitations the hard way over the last few decades. This is why 25% of all invested assets are in passive investment vehicles and Vanguard is now the largest fund family on the planet. Retail players gave up on the fever dream of Mad Money long ago; Mom and Pop are now investing in the missionary position from here on out.
Software, on the other hand, has not learned this lesson. The problem with computers is that they can’t be programmed with humility.
The Financial Times notes that computers now represent more than half of each day’s activity in the stock market. Which means they are mostly trading amongst and against each other:
While it may take weeks before regulators understand why the plunges occurred, one reason for the swing is that automated computer programs have changed how markets function…
Orders are being executed at lightning speeds in huge volumes. But there is another, often overlooked implication: these machines are being programmed to link numerous market segments together into trading strategies. So when computer programs cannot buy or sell assets in one segment of the market, they will rush into another, hunting for liquidity.
Since their algorithms are often similar (or created by computer scientists with the same training) this pattern tends to create a “herding” effect. If a circuit breaks in one market segment, it can ripple across the system faster than the human mind can process. This is a world prone to computer stampedes.
With the exception of HFT, which simply needs volatility and volume to thrive, I would say its a safe bet that quantitative strategies largely blew a gasket on Monday. Especially the really risk-averse ones, the ones that are directed to sell first and ask questions later. On Monday, it paid to ask questions first. Unfortunately, that function isn’t in the code.
This winter, I moderated a panel at the Battle of the Quants conference and interviewed four computer scientists on the leading edge of software-based trading. One of them was a particularly smarmy snot-nose from the West Coast who matter-of-factly informed the crowd that he “could write a program that crushes the market in my sleep.” I knew he was full of shit the moment he referred to himself as “an algorithmist.” Whatever dude, hope you’re sleeping well now.
Gerald Loeb, one of the greatest investors in history, said “If there’s anything I detest, it’s a mechanistic formula for anything. People should use their heads and go by logic and reason, not hard and fast rules.” Loeb, who would go on to found EF Hutton and make big money in the market for decades, said that his big takeaway from the Crash of 1929 and its aftermath was that nothing works all the time. A durable quantitative trading system builds this idea in. A flawed one tunes it out, and you get something like Long Term Capital Management or the blow-ups thatoccurred during the taper tantrum in 2013.
I’ll leave you with a quote from Dominique Dassault of the Global Slant blog, someone who’s more knowledgeable about black box trading than I ever will be:
While in Greenwich Ct. one afternoon I will never forget a conversation I had with a leading quantitative portfolio manager. He said to me that despite its obvious attributes “Black Box” trading was very tricky. The algorithms may work for a while [even a very long while] and then, inexplicably, they’ll just completely “BLOW-UP”. To him the most important component to quantitative trading was not the creation of a good model. To him, amazingly, that was a challenge but not especially difficult. The real challenge, for him, was to “sniff out” the degrading model prior to its inevitable “BLOW-UP”. And I quote his humble, resolute observation “because, you know, eventually they ALL blow-up“…as most did in August 2007.
Dassault’s take is that everyone is running a variation of the same strategy, and they’re all managing risk the same way – dumping oceans of stock the moment there’s a problem.
The Paradox of Dumb Money is that when it realizes its limitations, it ceases to be dumb. I can assure you that none of the thousands of black box quant managers have yet reached this realization about themselves.
The new game for traders is not running away from rapacious algos. Instead, it’s going to become about exploiting their failure to reason.
Pictures via GlobalSlant.com.
