Rise of the Quants

Why Quants Don’t Know Everything

I’ve been traveling a bit more than normal, which makes for an excellent opportunity to catch up on my reading, including reading a magazine article or two. Recently, I was traveling with one of our Executive Managers, who handed me the latest copy of Wired magazine. Skimming through the magazine I came across what turned out to be an excellent article by Felix Salmon, titled Why Quants Don’t Know Everything.

What’s a Quant?

Quant is “business” slang for a quantitative analyst or a person who specializes in the application of mathematical methods, also called quantitative analysis, such as numerical or quantitative techniques to financial and risk management problems.

Over time quantitative analysis has spilled over into other professions. If you have seen the movie Moneyball, you’ve witnessed a great example of what quantitative analysis can do in building a competitive baseball team.

Today many baseball teams employ a similar method, called Sabermetrics, to determine the statistical contribution of individual team members.

More recently, quantitative analyst and New York Times blogger Nate Silver accurately called the results of the 2012 Presidential election in every single state and the District of Columbia. Silver predicted that President Obama would win 332 electoral votes and Mitt Romney would win 206 electoral votes. ²

At first take it looks like the quants win, they’re usually right – that is, at least at first. As time goes on, the longer a system is operated by quantitative analysis the more and more the system creates incentives for everyone, and this includes employees, customers, vendors and competitors, to distort their behavior. This distorted behavior starts to provide more of what the system was designed to measure or produce, despite “real” value being created or not.

The Four Stages of Influence

Mr. Salmon separates the influence of quants, despite discipline or industry, into the four stages: pre-disruption, disruption, overshoot and synthesis.

Pre-disruption

Stage one is called pre-disruption, and is best visible in hindsight. Ripe for quantitative analysis is wherever there may be huge amounts of data that has never been mined. There is untapped potential and realizations in the data, and what it may tell us.

Disruption

Stage two is called disruption. This is where the algorithms start to take over. For example, the moment you start to receive targeted and personalized offers for credit cards and other financial services, solely based on computations, according to your finances and credit worthiness, instead of a relationship with your bank.

During the 2012 Presidential campaign,  when it came to fundraising emails, President Obama’s campaign threw out gut intuition, in favor of raw data about what worked and what didn’t work. For example, one email, headlined “I will be outspent,” raised $2.6 million on its own, when compared to projections with other emails which would have raises far less.

Overshoot

Stage three is called overshoot, where results in all these systems, whether they be metrics, algorithms or processes, are gamed by humans in a rational but often unpredictable way. Donald Campbell, a sociologist in the 1970s, identified this problem as Campbell’s law, which states, “The more any quantitative social indicator is used for social decision-making the more subject it will be to corruption pressures and the more apt it will be to distort and corrupt the social processes it is intended to monitor.”

Translated to the business world, simply put, once quants move past the disruption stage, they often don’t know when to stop. Now empowered, they tend to create systems that encourage cheating. As a manager, as soon as you pick some metric as a way to measure your people, they start focusing on maximizing the metric at the expense of doing the rest of the job.

Mr. Salmon provides an example found in Harvard sociologist Peter Moskos’ book, Cop in the Hood: My Year Policing Baltimore’s Eastern District, where police officers, driven by a purely statistical regime based on arrests, will often arrest a perpetrator even if there is no chance of conviction. Even in the 2012 Presidential campaign one had a hard time determining President Obama’s stand on issues from his website because it was so optimized for donations.

On the game show Jeopardy, the computer, Watson responds to an answer in the category U.S. Cities with “What is Toronto?”

Synthesis

Organizations need to acknowledge that living by the numbers alone, simply won’t work.

The fourth stage, called synthesis, is when quantitative insights are married with good old-fashioned experience. Even in the post Moneyball era, baseball relies on a combination of statistics and scouting. Mr. Salmon points out that organizations are recognizing the need for synthesis, citing how “the National Weather Service employs meteorologists who understand the dynamics of weather systems, can improve forecasts by as much as 25 percent compared with computers alone.”

Even the most powerful computers in the world, that can best a human one-on-one, can in turn be bested by humans aided by computers.

Data Isn’t Everything

Where big data and human intuition intersect we find good synthesis. Mr. Salmon concludes, “As long as the humans are in control, and understand what it is they’re controlling, we’re fine. It’s when they become slaves to the numbers that trouble breaks out. So let’s celebrate the value of disruption by data – but let’s not forget that data isn’t everything.”

Torture numbers, and they’ll confess to anything.  – Gregg Easterbrook

1. Porter, Martin (1984-05-29). “The PC Goes to Bat”. PC Magazine. p. 209. Retrieved 24 October 2013.
2. Vijayan, Jaikumar. “Presidential Election a Victory for Quants.” Computerworld. Computerworld, n.d. Web. 03 Feb. 2014.