Why Stock Markets Crash: Critical Events in Complex Financial Systems

by Didier Sornette

The scientific study of complex systems has transformed a wide range of disciplines in recent years, enabling researchers in both the natural and social sciences to model and predict phenomena as diverse as earthquakes, global warming, demographic patterns, financial crises, and the failure of materials. In this book, Didier Sornette boldly applies his varied experience in these areas to propose a simple, powerful, and general theory of how, why, and when stock markets crash.
Most attempts to explain market failures seek to pinpoint triggering mechanisms that occur hours, days, or weeks before the collapse. Sornette proposes a radically different view: the underlying cause can be sought months and even years before the abrupt, catastrophic event in the build-up of cooperative speculation, which often translates into an accelerating rise of the market price, otherwise known as a "bubble." Anchoring his sophisticated, step-by-step analysis in leading-edge physical and statistical modeling techniques, he unearths remarkable insights and some predictions--among them, that the "end of the growth era" will occur around 2050.
Sornette probes major historical precedents, from the decades-long "tulip mania" in the Netherlands that wilted suddenly in 1637 to the South Sea Bubble that ended with the first huge market crash in England in 1720, to the Great Crash of October 1929 and Black Monday in 1987, to cite just a few. He concludes that most explanations other than cooperative self-organization fail to account for the subtle bubbles by which the markets lay the groundwork for catastrophe.
Any investor or investment professional who seeks a genuine understanding of looming financial disasters should read this book. Physicists, geologists, biologists, economists, and others will welcome "Why Stock Markets Crash" as a highly original "scientific tale," as Sornette aptly puts it, of the exciting and sometimes fearsome--but no longer quite so unfathomable--world of stock markets.

Genres: Finance, Economics, Business, Science, Nonfiction, Mathematics, History

448 pages, Hardcover

First published November 18, 2002

Reviews

[Jacob · Rating 5 out of 5]

It is fair to compare this book to the Black Swan by Nassim Taleb. Here an attempt is made to analyze and quantity instabilities of the Black Swan variety; mostly the [stock] market(s) but the final chapter contains an analysis of civilization itself. The book is VERY rich in concepts and ideas, much more so than most books. It also assumes a lot from the reader. If, for example, Ising model, K-selected, or Polya's Urn doesn't ring a bell for you, this may be too difficult a read.

[Mario Cianfarani · Rating 4 out of 5]

Thoughts on what is a very interesting book on capital markets behavior that is not well understood.

Stock market crashes are caused by the slow build-up of long range correlations leading to a global cooperative behavior of the market and eventually resulting in a collapse in a short, critical time interval
• Crash may be caused by local self-reinforcing imitation between traders. If traders are more likely to imitate other traders which increases up to a certain point – critical point – by placing the same type of order – all buys or all sells – increases the probability of a crash. Frame work for understanding needs to be probabilistic b/c crashes are not certain outcomes
• Bubbles manifest themselves as overall super-exponential power-law acceleration in the price decorated by log-periodic precursors – a concept related to fractals
• Bubbles across epochs share a common underlying background as well as structure
o Rationale: rooted in the fact that humans are endowed with the same basic human emotions across history – fear and greed.
• Large crashes are not the same as a single large decline – they are fundamentally different events. i.e. A 1% decline is not the same event as a 10% decline and cannot be modeled as such
Financial Crashes are Outliers
• In the Bachelier-Samuelson financial world, in which distributions are normally (Gaussian, bell shaped) distributed, all returns are scaled according to a fundamental “ruler” called Standard deviation
• In reality, returns are not Gaussian
• They are not far from exponential law
• Under Gaussian assumptions: Oct 19/1987 -22.6% & Oct 20, 1987 rebound +9.7% should not occur. They are essentially impossible
• Under exponential law: rebound of 9.7% less extraordinary, once every 22,026 days or 88 yrs, -22.6% should occur every 520 million yrs – still qualifies as an outlier
• Can’t look at individual return data and assume that each successive return is uncorrelated. Drawdowns preserve the information in busts of activity that demonstrate local dependence.
• Body and tail of distributions are made up of 2 different populations that have different physics, scaling & properties.
• Large outliers are not scaled up versions of small fluctuations
o Distribution is made up of 2 different populations – the body & tail, which have different physics, scaling & properties
• Drawdown calculation, rather than daily or weekly returns or any other fixed time scale returns, are more adequate time-elastic measures of price moves
o Simply looking at daily returns and their distributions destroys information that the daily returns may correlated at specific times
Positive Feedbacks
• positive feedback asserts that the higher the price or the price return in the recent past, the higher will be the price growth in the future
• There is growing empirical evidence of the existence of herd or “crowd” behavior in speculative markets. It Is Optimal to Imitate When Lacking Information

Modeling Financial Bubbles and Market Crashes

Models are synthetic sets of rules, pictures, and algorithms providing us with useful
representations of the world of our perceptions and of their patterns
• More correct – bounded rationality. They do not have perfect knowledge. “long list of irrational or anomalous behavior shown by human beings in certain specific systematic ways should not confuse us: the relevant task for understanding stock markets is not so much to focus on these irrationalities but rather to study how they aggregate in the complex, long-lasting, repetitive, and subtle environment of the market”

The Risk-Driven Model
• Models investor behaviours, developed to formalize herd behavior or mutual mimetic contagion in speculative markets
• The emergence of bubbles is explained as a self-organizing process of infection among traders
• Its key assumption is that a crash may be caused by local selfreinforcing imitation between traders. This self-reinforcing imitation process leads to the blossoming of a bubble. If the tendency for traders to “imitate” their “friends” increases up to a certain point called the “critical” point, many traders may place the same order (sell) at the same time, thus causing a crash. The interplay between the progressive strengthening of imitation and the ubiquity of noise requires a stochastic description: a crash is not certain but can be characterized by its hazard rate that is, the probability per unit time that the crash will happen in the next instant provided it has not happened yet.
• Since the crash is not a certain deterministic outcome of the bubble, it remains rational for traders to remain invested provided they are compensated by a higher rate of growth of the bubble for taking the risk of a crash, because there is a finite probability of “landing smoothly,” that is, of attaining the end of the bubble without crash. In this model, the ability to predict the critical date is perfectly consistent with the behavior of the rational agents: they all know this date, the crash may happen anyway, and they are unable to make any abnormal riskadjusted profits by using this information

[Stephan Pire]

I greatly recommend this book. It goes from the history of crashes down to the detail on how to predict market events through physics (fractal)

[Q* := Q - {0} · Rating 2 out of 5]

Few books better illustrate the futility of applying (even interesting) mathematics to the complexity of human behavior than "Why Stock Markets Crash." With the benefit of hindsight, we now know LPPL (the "log-periodic power law") fails to model financial crashes in real-time, but it should have been obvious, even to readers in 2004, that the fine-tuning of four(!) parameters would be unsuccessful in modeling an entire market index comprised of millions of participants with unequal resources, knowledge, belief systems, financial goals, and levels of risk aversion. Sornette makes all the usual disclaimers, but those caveats only weaken the project's conceit; you either have a predictive "law" or a mushy hypothesis that needs constant qualification. Sornette's math isn't the problem. The math is easy; it's accurately predicting the behavior of complex, even chaotic, systems using toy models that's difficult. Just ask the myriad economists who missed the 2008 crash.

Sornette makes sure to include everything that might tangentially flutter into his mind, whether it substantively adds to the discussion or not: catastrophe theory, the propagation of forest fires, Mayan overpopulation, hypergeometric distributions, climate change, carrying capacity and the logistic function, taxes, Bayesian analysis, mineral mining, fractals, the evolution of computing power...the list is nearly transfinite. There's even the famous quote by Wigner, now obligatory in any book availing itself of mathematical machinery, concerning the unreasonable effectiveness of mathematics (199). Yes, the sheer range of material is impressive, but one is left with the impression these excursions exist less as a desire for comprehensiveness or plodding explication than as a signal to the reader that Sornette is a smart guy whose encyclopedic reach should convince you his model is correct. They also make the book much longer than it needed to be.

Despite the general disappointment of LPPL as a predictive model, however, "Why Stock Markets Crash" is worth reading because it still offers one of the most substantive and courageous attempts at predicting market crashes. If you like mathematics, finance, and the fascinating history of stock market collapses—and you're willing to subject yourself to Sornette's incessant, self-congratulatory meanderings—you'll enjoy the ride, even if it doesn't take you to your desired destination.

[Andrew Davis · Rating 5 out of 5]

An extremely informative text for those interested in economics and econometrics. The author is a leading authority in the field. He covers a range of approaches in analysing variations of the stock markets. This is not a book to be read and put aside, but rather referred to from time to time. The additional benefit of the book is almost 500 references that extend on the topics discussed in the book. Due to limited size of the book, some of them are critical in understanding the various topics.

The book covers in details the GARCH models, complex fractal dimensions and log-periodicity. It conducts a forensic analysis of major crashes and provides guidelines for predicting any future disturbances.

[Mi Lia · Rating 5 out of 5]

If you're looking for an approachable read to understanding stock market bubbles, written by a professor of geophysics, look no further.

Some basic knowledge of calculus might be needed.

[Jordan Coy · Rating 4 out of 5]

This book examines how stock markets crash and the dynamics involved in creating a crash. In the methods of measuring and predicting risk, this book sides with the “fractal” power law view of statistics (popularized by Nassim Taleb & pioneered by Mandelbrot) vs the standard Gaussian “bell curve” view. There are a lot of great observations about the inner workings of market dynamics and the components involved in crashes. Very technical but still accessible to finance students

Black Friday Oct 19 1987– the prime example of the fractal power law view. Under Gaussian view, the -22% drop and subsequent 8% return is statistically impossible, but it happened. (page 50)

Positive feedback: conditioned on the observation that the market has moved up, it makes it more probable that it will continue to move up, so that a large cumulative move ensues. (81)

Feedbacks work with general equilibrium theory/invisible hand of the market. General Equilibrium theory is the formalization of the idea that “everything in the economy effects everything else.” This fits with Adam Smith’s invisible hand principle. Smith never worked out a proof for this theory. It would not be proven until 1954 with K Arrow & Debreu. (83-84)

Herding: many people take similar action to mimic the behavior of others. Influences investment recommendation, IPO prices behavior, earnings forecasting, etc (94)

Patterns in price trajectory belong to fractal geometry-- fragmented geometric shape that can be subdivided into parts, each of which is a reduced-size copy of the whole (121)

Principles of Financial Modeling--Principles that guide the determination of financial models:
Absence of Arbitrage: The law of one price says two similar assets should be sold at the same price so should two same assets in different markets. Therefore there is no arbitrage opportunity to make a return by buying in an undervalued market and selling in an overvalued market. It is not a mechanism but a principle that emerges from large-scale organization of the market. (136-137)

Rational agents: a significant fraction of traders will act rationally upon the available information to optimize their strategy. “Bounded rationality” (138)

Rational expectation bubbles: Prices may deviate significantly over time intervals from fundamental prices. Rational bubbles however keep an anchor point in fundamental prices. Bubbles must obey rational expectations and no arbitrage opportunity. The market is not far off from being efficient and few arbitrage opportunities exist that beat buy-and-hold investment strategies. (140)

Two models: (149-150)
1. Risk-driven: by imitative behavior, noise traders make markets move more unstable. Crashes may be caused by local self-reinforcing imitation leading to a bubble. Rational traders who stay in the market profit from these exaggerated swings.
2. Price-driven: noise traders drive prices up into a bubble. Rational traders recognize the bubble, identify the risks, and and know the corrections that bring the prices back down to the fundamental value. As a result of rational investors, the price is driving the crash hazard rate and the price is being driven up by herding behavior .

2 classes of investors: (217)
1. Fundamentalists/value investors
2. Trend followers
Both act in nonlinear ways: power law singularity results from growth rate from trend followers. Oscillations result from nonlinear restoring force exerted on price by value investors, bringing price down to fundamental value (217)

3 basic ingredients in price dynamics: (218)
1. Trend following
2. Reversal to fundamental price
3. Risk adversion

Real option effect: new economy companies can use their already developed networks to grasp new opportunities for an untold future. (271)

Basic ingredients found in all crashes:
fueled by well founded economic fundamentals, investors develop self fulfilling enthusiasm by an imitative process/crowd behavior that leads to unsustainable overvaluations and eventually revert back down to fundamental price based on cash flows (272)
Mechanism that sustains bubbles and makes them crash abruptly: buying stocks on margin. (283)

4/5 Very technical but accessible book on market crashes

[Walter Cavinaw · Rating 4 out of 5]

Because I read this book on and off over the course of a year without taking notes, some of the details are more fuzzy.

Didier Sornette took his work on self-organized critical from earthquakes and materials failure and applied it to the stock market.

The central idea of the book is to examine the crash pattern formed through log periodic oscillations around a power law (LPPL = log periodic power law). Because power laws are commonly observed in fractals he relates the LPPL to fractals and discrete self similarity.

One of the more interesting sections looked at the reasons for why this might occur. The description that most stood out to me was around a hierarchical Ising model that can form the log periodic oscillation. Based on my personal experience I believe an Ising model, or agent-based model of market interactions is more realistic so I really liked this description. Some of his students seem to be examining the market from this light.

He goes on to demonstrate the LPPL model on a few instances through time. His group has even published some of their calls in real time. They have a few different formulations of the model, the differences of which I can't remember.

He goes on to speculate on the possibility that their are log periodic power laws in the structure of society such as technology evolution or demographics.

The idea presented in the book was super important and I have spent a significant amount of time reading through papers on this and similar topics. Chris Kaufmann of Parallax Financial Research has also done incredible research in this area (and first got me interested in this topic). His interviews and writing are sparse but worth checking out if these ideas are interesting.

Some readers might find the idea untenable because what it implies for things like technical analysis. In a way it is maybe a more advanced form of technical analysis which has a scientific basis and is tested.

The writing on the other hand is bland. It is unfortunately obvious that it is written by an academic. In part that's why it took me so long to finish.

[Kyle · Rating 5 out of 5]

This is a unique book, and so I don't know that it will work as well for other people. Sornette does not hesitate to use equations, including writing down differential equations and solutions. He uses them sparingly and puts many word explanations around them, but know that if you plan to read this book Sornette is not going to skip the math. I think he does an excellent job of explaining things without the math, but I always appreciate having the equations available.

For the book itself, Sornette explains the theory of log-periodicity (LPPLS, log periodic power law singularity, nowadays) in the context of the stock market. He starts with explaining the theory of randomness and the basic economics that drive stock markets and then talks about his theory of bubbles forming and how one may be able to predict them.

I like this book a lot because it presents an idea, and could work well as an explanation of how creating a scientific model works. The author also admits the shortcomings in the theory and offers it as one way that can sometimes be useful. The fact that it is now about 20 years old also gives it an interesting flavor, and reading some of the cited works is worth it, as well.

I'm not completely convinced of the log-periodicity claim, but it does seem plausible, and certainly worth exploration (it has, of course, been more explored in the intervening years), and using the theory to predict crashes and bubbles seems like it could give you some interesting insights or at least perspectives. I'd recommend looking for some of the critiques online, though, as well.

Overall, I'm a fan of this type of explain-your-model books. Even if the model turns out not to be as accurate for the phenomena involved, it explains how you one can think through a problem, come up with some sort of solution, and ways to test it. Obviously, this process can vary a lot depending on circumstances, but seeing how someone does it always fascinates me.

[Richard Crowder · Rating 5 out of 5]

Do not buy this as an e-book; a couple of Princeton e-books on mathematical subjects that I've bought had bad misprints in the formulas. For this book, I read the paperback 2017 edition with a new preface by the author.

Stock-market crashes generally take everyone by surprise--they feel like bolts from the blue. They're usually not. Sornette shows how the interplay of greed, fear, and imitation among investors and traders creates an accelerating rhythm of sudden rises alternating with increasingly brief pauses. This "mathematical signature" can begin months or years in advance, but its predictive value rises in the last year before the death of the bubble (which may be relatively calm, but usually is followed by a crash).

Sornette presents the results of several predictions made using this technique. While his track record is not perfect, it is strongly better than what could be expected from chance. Although the math is advanced, the discussion and the graphs make the argument clear to the lay reader.

What about the everyday investors who don’t have access to Sornette’s computational skills? The lesson is straightforward: as markets rise, and especially as they rise sharply, so does the danger of a crash. As they watch a sharp rise, investors should reduce their equity positions to capture gains made so far and limit the danger to their portfolios.

But let's assume that you're not in the stock market and don't plan to be. The last chapter broadens the discussion to consider a wide range of problems confronting the world in the period from the year of publication (2002) to the potential "end of the growth era" around 2050. Many of the trends described have only become more pressing since 2002. This book is both important and fascinating--not just for investors but also for citizens of an uncertain world.

[Lukasz]

recommendation: [The best books on Physics and Financial Markets | A Five Books interview](https://fivebooks.com/best-books/jame...)

recommendation: Chapter 7 of [The Physics of Wall Street: A Brief History of Predicting the Unpredictable by James Owen Weatherall | Goodreads](https://www.goodreads.com/book/show/1...)

[Yates Buckley · Rating 3 out of 5]

An important technical text that evalutes the dynamics of market crashes, through simulation and mathematical modeling.

Thr text is also confusignly written, disorienting in structure, so that by the end you are not sure what to take away to your everyday life.

[Yannick Daelemans · Rating 4 out of 5]

Good book, about a good idea.
The idea that bubbles are statistical anomalies, and how to recognize them, was very good given. However, the book is a bit too wide, going over too many topics sometimes.

[Ethan Drower · Rating 2 out of 5]

I'd skip this one.

[Eli]

Al

[Fabian · Rating 5 out of 5]

Instead of anecdotes and the usual „irrational behavior“ argument, proper data on market crisis. Best book on the topic I have read.

[Iver Gjørvad · Rating 3 out of 5]

Kule konsepter, for teknisk

[Stevenglinert · Rating 2 out of 5]

I read most of this while shit was running on my computer in the GSB library. I'm sure the MBAs love it. So, like Lacan always uses math terminology in this like weird bullshit way and this was similar.

This felt like a guy trying to rewrite Kindleberger with more complex math and a cool cover. Wow such fractal. But he's a geophysicist and like, his attempts at economics and social science came off as forced. Just read Kindleberger and Shiller's Irrational Exuberance and like don't bother.

[Andre Luis · Rating 2 out of 5]

Não pude ler o livro todo. Muito matemático e demonstra muita confiança em equações baseadas em dados históricos.
Apresenta algumas argumentações interessantes sobre como modelar o mercado de capitais e sobre "rational expectations". Válido para lembrar sobre bolhas e seus efeitos.

[Chris · Rating 5 out of 5]

http://seekingalpha.com/article/11887...

[Chang Lan · Rating 5 out of 5]

Ising + Hierarchical Structure = Power Law + Log Periodic

[Matt DosSantos DiSorbo · Rating 3 out of 5]

Diligent technical work, although loses sufficient standing with the 'magic formula' to predict markets at its close.