Podcast: Using a Supercomputer to Trade the Market
What happens when supercomputers trade the markets?
Last week, we spoke with Didier Sornette, director of the Financial Crisis Observatory at ETH Zurich, who uses one of Europe's most powerful supercomputers to scan financial assets around the globe for unsustainable price movements—"bubble signals"—which, when combined with fundamental analysis, is used to identify mispricings or "deviations from efficient markets," as Didier explains in the following clip:
Predicting Bubbles in Advance
The Financial Crisis Observatory (FCO) was founded in response to the crisis of 2008 and uses “a scientific platform to test the hypothesis that … excesses in the market have a degree of predictability that can be diagnosed in advance,” Sornette said.
Here's his explanation of “the engine powering the Financial Crisis Observatory":
“We are using a concept that looks at the financial markets as fundamentally unstable, like Hyman Minsky famously said in his financial instability hypothesis...and we are essentially pushing this idea further by developing an operational implementation looking at the market as having trenchant phases—pockets of predictability—where it may accelerate for a while and then change its regime... We have developed a methodology called the log-periodic power law singularity method or formula that we calibrate on different time windows at multiple scales...to indeed identify these times when there is a non-sustainable market regime or deviation from efficient markets...that is the core of what is the engine powering the Financial Crisis Observatory.”
How It Works
There are essentially two dimensions to the supercomputer's analysis, Sornette stated. One dimension is technical, performing a daily scan on 25,000 assets for bubble-like characteristics (faster than exponential price behavior up or down), and the other is fundamental, simultaneously ranking them according to their valuation (high or low).
By combining these two dimensions—a value and a bubble score—assets are categorized into four quadrants that each have their own trading strategy: trend-following buyers, contrarian sellers, trend-following sellers, and contrarian buyers.
Conditions Right Now
This month, we’re seeing an abnormally large number of stocks in quadrant two, Sornette noted. These are stocks with strong positive bubbles, accelerating to the upside, but value score fundamentals are weakening.
Currently, Apple, Amazon, Tesla, and a long list of other familiar names fall into this category, according to their May report.
That being said, Sornette cautions investors against using their screen as an automatic buy or sell list for trading but more as a jumping off point for further research on what you hold or may want to hold.
“Don’t use it as a God-given, error-free method,” Sornette noted. “We are dealing with an extremely complex system...which is the result of many conflicting forces,” including central bank interventions, he said.
Listen to this full interview with Didier Sornette on our website by logging in and clicking here. Become a subscriber and gain full access to our premium weekday interviews with leading guest experts by clicking here.
About FS Staff
FS Staff Archive
|03/23/2018||Secure Methods for Generating Income in Retirement||story|
|03/20/2018||Podcast: How to Prepare for a Bear Market in Bonds||story|
|03/16/2018||Technician Eyes 2019 Stock Market Peak||story|
|03/15/2018||Danielle DiMartino Booth: It Won't Take Much More for the Fed to Break the Markets||story|
|03/13/2018||Projections for Peak Oil Demand Are Overblown, Says Dan Steffens||story|
|03/12/2018||Nicoski on Market Breadth, Foreign Investors Fleeing US Bond Market||story|
|03/09/2018||Prepare for Second Half Slowdown, Mild 2019 Recession, Says ITR's Alan Beaulieu||story|
|03/09/2018||America's Massive Looming Social Security Problem||story|
|03/07/2018||Don’t Count the Dollar Out Just Yet, Says Marc Chandler||story|
|03/06/2018||Gary Dorsch: Exploding Deficits, Rising Rates Put US Market in "Dangerous Territory"||story|