The Gateway to Algorithmic and Automated Trading

Seismic shocks

Published in Automated Trader Magazine Issue 30 Q3 2013

When the earth moves in a big way, the markets can too. As part of our series on weather and seismic information, Automated Trader talks to a scientist who thinks he may have found a way to gain advance warning of major earthquakes.

While weather forecasting may have made large strides in recent decades, scientists have been unable to make similar progress in figuring out how to anticipate earthquakes. In his book 'The Signal and the Noise', Nate Silver devotes a chapter to the reasons why researchers have been stymied in their efforts to predict seismic events. Unlike atmospheric weather factors, which are observable, seismic activity takes place far below the surface of the earth, where no one can obtain the relevant data to build predictive models.

But what if you could know about a major seismic event before it struck? Or conversely, what if you were trading against others who might have that information? The last major market-moving quake, the Tohoku‐Oki earthquake on Friday, March 11, 2011, occurred just before the close of trade. But Nikkei 225 stock futures traded in Singapore fell about 5% and the impact of the quake rippled across global markets, with the markets continuing to fall the next week.

One scientist, Professor Kosuke Heki of Hokkaido University in Sapporo, believes he may have discovered a method that can predict a major earthquake some 40 minutes in advance. Most people have been interested in this research because of the lives that can be saved, and that was clearly behind the wave of media attention his work generated when it was published in 2011.

But there's also no question such information would be enormously valuable to anyone in the markets. Professor Heki agreed to answer Automated Trader's questions about his research via email.

The Study

Heki's paper appeared in Geophysical Research Letters, an American Geophysical Union publication, in September 2011. That was about half a year after the March 11 Tohoku‐Oki quake, which measured 9.0 on the moment magnitude scale and led to a massive tsunami which caused a nuclear crisis.

Heki discovered that the total electron content (TEC) in the ionosphere spiked higher about 40 minutes before the quake. The ionosphere is part of the upper atmosphere and TEC is monitored by climatologists, physicists and other scientists for a variety of reasons. Monitoring TEC is made possible because of GPS, the same satellite-based system that lets your car give you directions.

If the quake is large enough, it seems to disturb the electron content in the ionosphere to a point where it shows up in measurements well before the quake actually happens.

"The Japanese dense network of Global Positioning System (GPS) detected clear precursory positive anomaly of ionospheric total electron content (TEC) around the focal region," Heki wrote in his paper.

In other words, TEC surged before the Tohoku‐Oki quake and Heki said similar "anomalies" were seen before a 2010 Chile earthquake, and possibly the 2004 Sumatra‐Andaman quake.

The reason

Heki said one paper published last year gave a physical explanation as to how an earthquake could disturb the ionosphere in its "preparation stage". He added: "However, this is just one of the candidates of possible mechanisms. Release of positive electric charges from stressed rock (classical hypothesis by Dr. F. Freund, NASA) and its influence on the ionosphere is another candidate," he said in his email.

The critics

Heki said some scientists have doubted whether there has been a genuine enhancement of TEC, although in most cases, after shown evidence, they came to believe the increase did occur. But critics have still questioned the significance of the phenomenon. "Unfortunately, the 2011 Tohoku-Oki earthquake occurred during a geomagnetic storm (just a coincidence, no causal relationship between the storm and the earthquake)," Heki said in his email. The professor said he is working on a paper rebutting the criticisms and that he is satisfied with the evidence in the new document.

The application

But the big question is, assuming that the findings are correct, what would it take to translate this knowledge into practical applications that could make predictions in real time?

Heki said it would be key to distinguish between TEC enhancements that are of "space weather origin" and enhancements that are pre-seismic. "If the GPS network is large and dense enough, such distinction will be possible by using a sophisticated criterion," he said.

Heki said regions with high seismic risk are already covered by good GPS networks, citing GEONET in Japan, PBO in California and SUGAR in Sumatra. The major obstacle is the establishment of a system that has real-time TEC data monitoring that automatically distinguishes between pre-seismic activity and other types and then hiring experts to monitor it 24 hours a day.

Heki said no additional hardware would be necessary. "The most important factor is that the majority of the researchers believe that the TEC information is useful in earthquake prediction. This may take a few more decades even if I do my best," he said.

Heki said that until someone successfully predicted an earthquake, a government institution may not take up the challenge. From this point of view, a private sector (firm) may play an important role," he said, although he added it might be difficult to let real-time GPS data from a public network flow into a private company. Heki said there had been approaches from a few companies, but they consisted of "simple exchanges of technical information".

One thing Heki noted was that even if real-time earthquake prediction became a reality, it would still only apply to the largest of quakes, those with magnitudes of 8.5 or more.

Still, those are the quakes that tend to move markets.

In the meantime…

While it may be some time before such a system is in place, it is possible already to set up automatic alerts of earthquakes, based on pre-determined parameters such as magnitude and location. The United States Geological Survey (USGS) offers a real-time notification service that is free to anyone.

A USGS official said the service has been running for about six years and it has about 375,000 subscribers. Many of those are seismologists or people who work in emergency services such as firefighters. Subscriptions tend to spike up after significant earthquakes. One thing is clear, however; many financial firms have signed up. Automated Trader asked about a handful of the biggest banks and trading firms in the world and the USGS official said there were several hundred subscribers from those firms receiving earthquake alerts.

Put another way, the next time a big earthquake strikes, bear in mind your trading competition: some in the market may have set up automated systems that can make trades the very moment the ground starts shaking.

The big quakes - where and when

The United States Geological Survey lists 13 earthquakes that had a magnitude of 8.5 or greater since 1990.

Date Region Magnitude
Nov 11, 1922 Chile-Argentine border 8.5
Feb 3, 1923 Kamchatka, Russia 8.5
Nov 4, 1952 Kamchatka, Russia 9.0
March 9, 1957 Andreano Islands, Alaska 8.6
May 22, 1960 Chile 9.5
Oct 13, 1963 Kuril Islands 8.5
March 28, 1964 Prince William Sound, Alaska 9.2
Feb 4, 1965 Rat Islands, Alaska 8.7
Dec 26, 2004 Off west coast of northern Sumatra, Indonesia 9.1
March 28, 2005 Northern Sumatra, Indonesia 8.6
Sept 12, 2007 Southern Sumatra, Indonesia 8.5
March 11, 2011 Near east coast of Honshu, Japan 9.0
April 11, 2012 Off west coast of northern Sumatra, Indonesia 8.6

Many of these quakes have not occurred in market-sensitive regions. However, Japan and California, two areas of enormous wealth, are both highly vulnerable to seismic activity.