Understanding the Statistical Probability of a Major Earthquake

–The 9th annual Great Shakeout will take place on October 19, 2017.

–The event is a coordinated state wide event to practice earthquake safety and preparedness.

–Every state in the USA has experienced an earthquake of some kind. Alaska and California have had the most.

Although the sophistication of earthquake prediction and the science of seismology has increased exponentially in the past decade or two there is much that is still not known. There is so much that is still left to be discovered. That’s evident when the best earthquake scientists in the world give disclaimers like this (emphasis added):

“(Thomas H.) Jordan, director of the Southern California Earthquake Center, is quick to make clear this is not an earthquake prediction. Predicting exactly when and where a catastrophic earthquake will strike next is impossible, scientists say.

You sure don’t hear scientists toss around the ‘i word’ much in today’s world. Just think about some of the other forms of severe weather that humans have to deal with and compare how they can be predicted. At one point, the best meterologists could do is give me the probability of rain in a region. Later, they came up with more sophisticated forms of radar and other technologies that allowed them to give more precise probabilities. Now I have an app on my phone that gives me extremely accurate up to the minute forecasts of when rain will start and my exact location.

On the other hand, they may have day to day weather forecasting wired but there’s still plenty of imprecision when it comes to severe weather. Hurricane forecasting is only good once the tropical storm forms and scientists can start to track it. The fact that it’s a big, slow moving system makes it fairly easy to keep an eye on but it’s unpredictable movement prevents an accurate projection of when and where it will make landfall. Even when landfall is just hours away there is never a 100% confidence projection of where it will strike and how it will impact surrounding areas. The best that can be done to mitigate damage is for residents of coastal states to be aware of when the risk of a hurricane is highest and to know what they can do to protect themselves.

Tornadoes are even more difficult to predict. Thus, the concept of the ‘Tornado Watch’:

“A watch is issued when conditions are favorable, for example, either for a severe thunderstorm or tornadoes,” AccuWeather.com Senior Meteorologist Dan Kottlowski said. “It doesn’t mean severe weather is imminent.”

The ability to measure things like wind shear and upper atmospheric conditions has improved dramatically but one of the major components of tornado favorability is ‘warm, humid lower atmospheric conditions’. Or as we call it in South Carolina ‘May through September’. Tornado Warnings have improved in their accuracy–at one point, they meant that there had been a ‘visible sighting of a tornado’. That presumably meant if some yahoo in a trailer park thought he saw one and called it in that would necessitate the issuance of a tornado warning. In the pre-digital ‘dark ages’ some communities relied on a loud siren to sound the alarm. Now tornado warnings are issued more frequently based on meteorologists observing storm rotation in highly sensitive radar. All of this technology has allowed meteorologists the ability to provide a few minutes more advance warning when a tornado is imminent which can be a big deal and mean the difference between life and death. Even so, they can do little before one actually develops.


This all underscores just how little is known about earthquakes. For now, the prevailing wisdom is that weather doesn’t play a part and historical data analysis validates this to some degree. Even so, the US Geological Survey offers the disclaimer:

“As far as we know, there is no such thing as “earthquake weather”. Statistically, there is an equal distribution of earthquakes in cold weather, hot weather, rainy weather, etc. If weather does affect earthquake occurrence, we do not yet understand how it works.”

Scientists have learned a lot about earthquakes from studying the data of past events. Technology has produced more sensitive measuring devices which means even more data to work with. For example, it is estimated that there are 500,000 detectable earthquakes worldwide each year. 100,000 of these can be felt without instruments and approximately 100 cause damage. Every state in the US has experienced an earthquake of some sort with the most in Alaska, California and Nevada. Earthquakes are so common that the USGS reminds us that there is a ‘100% chance’ of their occurance on a given day:

There’s a 100 percent chance of an earthquake today. Though millions of persons may never experience an earthquake, they are very common occurrences on this planet. So today — somewhere — an earthquake will occur.

It may be so light that only sensitive instruments will perceive its motion; it may shake houses, rattle windows, and displace small objects; or it may be sufficiently strong to cause property damage, death, and injury. It is estimated that about 700 shocks each year have this capability when centered in a populated area. But fortunately, most of these potentially destructive earthquakes center in unpopulated areas far from civilization.

The article goes on to once again emphasize that “Earthquake prediction may some day become a reality, but only after much more is learned about earthquakes.”


The fact that legit scientists concur that it’s impossible to predict earthquakes with any degree of accuracy that hasn’t stopped people from trying. Google ‘earthquake prediction’ and you’ll find a dozen of websites and even more theories. Some examine legitimate information outside of the scientific mainstream. One site analyzes a variety of components ranging from animal behavior to lunar phases. The emphasis is more on the prediction process which is why they look to ancillary factors that aren’t directly related to the earthquake. Anecdotally, dogs sure seem to sense them coming in some cases minutes before humans so the ‘animal behavior’ isn’t as silly as it sounds. Given the influence of lunar and solar gravity on ocean waves there’s at least some logic to associating them with earthquakes.

Unfortunately, this ‘DIY’ earthquake research quickly goes from plausible to tin foil hat stuff. There are efforts to predict earthquakes with an increasingly bizarre range of factors including cloud formation and movement, sun spots and even astrology. Many reek of a theoretical liability common to sports handicappers–mistaking correlation for causation. Some are just plain nonsense. Still others use algorithmic analysis of data from past and present seismic events to predict future earthquakes. While there may be some value in using this data to ascertain probability (a completely different concept than prediction) there’s a definite vibe of hokum with many ‘forecasters’. This isn’t helped by the number of earthquakes they claim to have ‘predicted’ giving them a winning percentage that would make the sleaziest boiler room sports tout blush.

Still other earthquake predictions have understandable ulterior motives. During my research I found several stories about the inevitability of catastrophic earthquakes in certain regions. After doing some digging, it turns out that they originated from companies that have a vested economic interest in fomenting as much earthquake fear as possible. If the entire world is in hysteria over earthquakes they’ll need consultants to help them plan for it and ‘earthquake proofing contractors’ to help them prepare. Unfortunately, the mainstream media isn’t very discerning in what they report and consider the level of hysteria to be a ‘feature not a bug’ of this type of spurious information.


The most viable theory of earthquake prediction today is a radical change from the prevailing wisdom of just a few years ago. Seismologists are starting to focus on a potential link between small earthquakes and the likelihood of bigger earthquakes. This is a 180 degree shift from the theory that smaller quakes diminished the chance of a large earthquake due to the ‘release of pressure buildup’. The current paradigm suggests that small to moderate earthquakes could be a precursor to more significant earthquakes in other areas along the fault on which they occur. This would be of particular interest in Southern California–moderate earthquakes often occur in sparsely populated areas of the Central region of the state. Under the new model this suggests a higher probability of a more serious earthquake all along the San Andreas Fault including densely populated Southern California. Using the old model, a moderate quake in Central California would have been interpreted to have minimal significance to SoCal.

This new theory along with some good old fashioned database analysis has allowed scientists to come up with some general theories of probability about a serious earthquake in Southern California along the San Andreas Fault. In an average week with no significant seismic activity, there is a 1 in 10,000 chance of a 7.8 or higher earthquake along the fault. The week after moderate seismic activity along the fault the probability of a 7.8 or higher earthquake along the fault drops precipitously to 1 in 100. Even with this information there’s a lot of ambiguity. What constitutes ‘moderate seismic activity’? Is there any significance to where this moderate seismic activity happens along the fault? What do these numbers mean in context of the state’s population? The San Andreas Fault runs almost the entire length of the state from Humboldt County in the North to the Salton Sea in the South. The significance of even a severe earthquake would vary widely depending on where it occurred along the fault.


Based on this information how closely should the population of California follow this information and how concerned should they be? It’s interesting that Thomas H. Jordan, director of the Southern California Earthquake Center, made the same point that I thought initially when reading this data–even in a week following moderate seismic activity “there’s still a 99-in-100 chance that the large earthquake won’t happen during that particular week.” Then there’s this money shot which is buried deeply in the text: “There are other limitations to the model. It needs to be tested to see how good it is at forecasting, Jordan said.” Translation–‘your mileage may vary’.

The problem is that the general public has a downright pitiful understanding of probability and what it means. Most people don’t worry about heart disease though the probability of dying from this disease throughout your lifetime is 1 in 7. The flu probably has more widespread concern with everyone short of the Starbucks barista trying to give you a flu shot. The chance of dying from the flu is 1 in 70 throughout your lifetime. On the other hand, our country is terrified by terrorism and the odds of dying in a terrorist attack is laughable. People line up to buy Powerball tickets whenever the jackpot goes up despite only a 1 in 292 million chance of winning. With that caveat in mind, here’s some other probabilities to provide a comparison:

Odds of being killed by heart disease: 1 in 7
Odds of being killed by the flu: 1 in 70
Odds of being killed in a car accident: 1 in 111
Odds of being killed in an act of murder by another US citizen: 1 in 249
Odds of being killed by accidental gunshot: 1 in 8,359
Odds of being killed by police: 1 in 8,359
Odds of being killed by a tornado: 1 in 60,000
Odds of being killed by an earthquake: 1 in 110,000
Odds of being killed by an asteroid: 1 in 1.9 million
Odds of being killed by a shark: 1 in 3.7 million
Odds of being killed by a foreign born US immigrant terrorist: 1 in 46.1 million

We can also put it into perspective within the context of Southern California sports:

Odds of the Golden State Warriors winning the 2018 NBA Title: 1 in 1.7
Odds of Kevin Durant winning the NBA MVP Award: 1 in 6
Odds of Steph Curry winning the NBA MVP Award: 1 in 20
Odds of the Anaheim Ducks winning the Stanley Cup: 1 in 30
Odds of the Los Angeles Kings winning the Stanley Cup: 1 in 35
Odds of the San Jose Earthquakes winning the MLS Championship: 1 in 40
Odds of the Los Angeles Rams winning the 2018 Super Bowl: 1 in 60
Odds of the Los Angeles Chargers winning the 2018 Super Bowl: 1 in 100
Odds of the Los Angeles Clippers winning the 2018 NBA Title: 1 in 105
Odds of the Los Angeles Lakers winning the 2018 NBA Title: 1 in 170
Odds of Lonzo Ball winning the NBA MVP Award: 1 in 500

Sports odds calculated from my power ratings and statistical data. Your mileage may vary.

I don’t know about you, but knowing that even the lowest odds of a major earthquake along the San Andreas Fault is roughly as likely as the Los Angeles Clippers winning the NBA title makes me feel better.

About the Author: Jim Murphy

For more than 25 years, Jim Murphy has written extensively on sports betting as well as handicapping theory and practice. Jim Murphy has been quoted in media from the Wall Street Journal to REASON Magazine. Murphy worked as a radio and podcasting host broadcasting to an international audience that depended on his expertise and advice. Murphy is an odds making consultant for sports and 'non-sport novelty bets' focused on the entertainment business, politics, technology, financial markets and more.