Generational Dynamics: Forecasting America's Destiny Generational
 Forecasting America's Destiny ... and the World's


Generational Dynamics Web Log for 6-Jul-2009
The Bubble Algorithm - How computers and herd behavior are inflating the stock market bubble

Web Log - July, 2009

The Bubble Algorithm - How computers and herd behavior are inflating the stock market bubble

But Thursday's jobs report is changing attitudes quickly.

My report last week on "The influence of computerized trading programs" has raised some interesting discussions.

In that article, I discussed the fact that the S&P 500 price/earnings index had remained in the range 18-20 for many years (2004-2008), and I inferred that the only possible explanation was that programmed trading algorithms used by the major financial firms all used similar algorithms, and that a P/E valuation of 18-20 has been a part of those common algorithms.

A couple of people wondered if this claim of similar algorithms was too extravagant. Speaking now as a computer software consultant who's worked for a number of financial firms, there's little doubt. There are only so many trading algorithms, and they're all well known -- in fact, they're taught in colleges. Different firms might vary some of the parameters, and attempt to gain a trading advantage that way. But when you have almost five years of a constant P/E valuation of 18-20, there's really no other credible possibility.

This really isn't surprising. Different companies use similar algorithms for accounting, for architecting buildings, for diagnosing illnesses. Only the top-notch research centers are developing new algorithms for things, but once a new algorithm is developed and proven to be successful, it becomes widely used.

For example, here's a paragraph from a recent Naked Capitalism blog entry on a different subject -- "value at risk" (VaR) models for risk evaluation:

"The other perversity of that approach to VaR is that it encourages herd behaviour in volatile markets, before the banks have even made it to the sidelines. In other words, since all the models in all the banks are essentially the same model of the same data, they all start screaming ‘fire’ at the same time, with predictable consequences at the exits. All this and more is well covered by Triana: particularly the way that a long period of low volatility before 2007 meant that VaR endorsed massive positions in assets that were suddenly big loss makers, when things went sour."

It's this concept of "herd behavior" that I want to focus on. This phrase could be used to describe many of the things that I've been describing on this web site for years. These things include the offering of predatory mortgage loans, the securitization of faulty mortgage loans into "toxic assets" (as we now call them), and the Pollyannaish reporting by CNBC, Bloomberg TV, the Wall Street Journal, and other financial media.

A point being made by the above quoted paragraph is that herd behavior increases systemic risk. Most theories about the market assume that each investor is acting completely independently, making independent decisions, so that one person's bad decision is canceled out by another person's good decision. But if all investors are acting in unison, then a mistake or bad decision by one becomes a mistake by all.

Herd behavior and the Bubble Algorithm

It's pretty clear that the herd behavior that pegged buy/sell decisions to a S&P 500 P/E valuation of 18-20 for almost five years was completely abandoned at the beginning of 2009. Starting in January, the computerized buy/sell algorithms were modified to something else. As I discussed in last week's article, I'm suggesting that these algorithms are producing a new stock market bubble.

I would suggest that we call this the Bubble Algorithm.

A web site reader questioned this concept as follows:

"If there are computer programs all buying millions of shares in microseconds, who is selling all these shares. There has to be a buyer for every seller. The total amount of stock is constant at any moment. If we had insight into both sides of the transaction, we would understand better what is happening here. I thought all transactions had to go through exchanges and specialists."

The behavior we're talking about is ordinary human behavior, encapsulated in computer software. We implement all sorts of human actions in computer software, so implementing the Bubble Algorithm really isn't that remarkable.

Even in an ordinary bubble, like the Tulipomania bubble, there has to be a buyer for every seller. This relates to the "greater fool" concept. You have to be a fool to purchase assets during a bubble, but you can buy the assets expecting to sell them to someone else, a greater fool, at an even higher price.

When the stock market is operating a low volume, then lots of tricks are available to insiders.

Consider the following mind-blowing example:

Let's suppose I have a million shares of stock A, selling at $1.00 per share, and you have a million shares of stock B, also at $1.00 per share. Then we each have $1 million in stock.

I sell you one share of A for $2.00, and you sell me one share of B for $2.00. Then all of a sudden, we now each of $2 million in stock. Isn't that incredible?

Once you understand that example, then you can see how all kinds of variations can be played. I can buy a large position in one stock, knowing that there are plenty of insiders who will buy some of it at a slightly higher price, thus boosting the value of my entire position, and increasing the size of the stock market bubble. If a lot of people are doing that, then you have the Bubble Algorithm.

It's "herd behavior" that makes a bubble possible. As we quoted from Garet Garrett's 1932 book, "The bubble that broke the world:"

"Mass delusions are not rare. They salt the human story. The hallucinatory types are well known; so also is the sudden variation called mania, generally localized, like the tulip mania in Holland many years ago or the common-stock mania of a recent time in Wall Street. But a delusion affecting the mentality of the entire world at one time was hitherto unknown. All our experience with it is original."

In a "normal" situation, individual investors make individual decisions about individual stocks, evaluating each stock purchase based on fundamentals, an appraisal of the underlying corporation.

Program trading is a negation of the "normal" situation. No fundamentals are used. The trading algorithms detect small changes in the prices of stocks and attempt to take advantage of them by buying or selling within a few milliseconds.

This whole concept is a rejection of anything fundamental, and rejection of all common sense, and is perhaps the best illustration of the mass delusion of investors today. The only difference today is that these delusions are implemented in computer software.

Bubble Algorithm - the crash

There are two parts to a bubble -- growing the bubble, and the crash.

So the Bubble Algorithm has the same two sides. We've discussed the side that grows the bubble. What happens when the crash comes?

Let's return to the example I gave before. When the bubble is growing, you can buy a large position in a stock at, say, $100 per share, and then sell a small portion of that stock at $101 per share. When you do that, the value of your entire holding goes up to $101 per share. This makes you a lot of money, and it also increases the size of the bubble.

But what happens when the market starts to go down? Suppose you have a large position at $101 per share. If the price falls to $100 per share, then you can't just sell of a small portion; you have to sell the entire position. That's the sell side of the Bubble Algorithm, and you can see that the crash is going to be much more rapid than the growth was.

Putting on my hat as a software development consultant, there's little doubt in my mind that something like the Bubble Algorithm has been implemented the major financial institutions today, and that the buy side and sell side of the algorithm have both been implemented.

Switching from the buy side to the sell side, whether by human investors or by a computer running the Bubble Algorithm, is what's commonly known as a "panic."

If the growth of a bubble, whether by humans or computers, is a manifestation of mass delusion, then a panic is a sudden end to the delusion, and a huge dose of reality. The only difference is that the delusion can go on for weeks, months or years, while the panic may require only a few hours or days.

The jobs report changes the mood

We may or may not ever know what triggers a switch from bubble to crash. The cause of the 1929 panic is still debated. All we can say for sure is that it will be some chaotic event (in the sense of Chaos Theory), and that it can't be predicted.

Thursday's jobs report may or may not be such a trigger, but it has certainly changed the mood. It was a huge disappointment to economists and politicians alike.

To see why, take a look at the following graph from the Calculated Risk blog:

Percent Job Losses in Post WW II Recessions <font face=Arial size=-2>(Source: Calculated Risk)</font>
Percent Job Losses in Post WW II Recessions (Source: Calculated Risk)

This graph very well represents the kind of data that economists and politicians look at it. It compares the job losses, month by month, in all the post-WW II recessions. If you look at it long enough to get a feeling for what's going on, you can see that in the previous post-war recessions, job growth began to spike up right about now, if not earlier. Politicians and economists were CERTAIN that the same would happen now. The fact that it hasn't happened is a signal to economists that their assumptions have been disastrously wrong.

From the point of view of Generational Dynamics, what's wrong with that analysis is that this isn't an ordinary post-war recession. We're in the middle of a generational stock market crash, the first since the one that began in 1929, and job losses are following the post-1929 pattern.

That this is an "ordinary" recession is part of the mass delusion that politicians and investors have been suffering from.

It's hard to list all the factors that went into this mass delusion, but certainly the inauguration of President Obama was a major one. Throughout the campaign last year, and even after the election, Obama promised that the world would change after January 21. He would be "guided by facts, not ignore them." He would cure global warming, provide universal health care, and restore the stock market bubble.

As I've written many times, he doesn't have a snowflake's chance in hell of succeeding at ANY of these programs. The fact that it was widely believed that he would is part of the mass delusion that's infected Americans since January.

Thus, on Thursday, President Obama said that the jobs report was "sobering news." Returning to campaign mode, he added, "It took years for us to get into this mess, and it’s going to take us more than a few months to turn it around." And on the Sunday morning news talk shows, Vice President Joe Biden said that the Obama administration had "misread the economy" when it made its earlier forecasts after taking office.

Thus, our new tv drama, "The education of Barack Obama," continues apace, with intriguing new episodes almost every day, as the mass delusion continues to dissolve.

From the point of view of Generational Dynamics, the question is whether this is enough to trigger a panic. That's impossible to predict, of course. But if the public mood continues to deteriorate, and my expectation is that it will, then the current stock market bubble should not continue much longer. Whether that means a gradual fall or a full-fledged crash in the near future can't be predicted.

Will money market funds 'break the buck'?

Many money market funds are in increasing danger because the stocks backing the funds are less than the nominal value of the funds themselves. Last year, some market funds were already forced to "break the buck," meaning that a share in the money market fund is worth less than $1.00. The deteriorating financial crisis means that breaking the buck will occur much more frequently, and may trigger a panic on money market funds.

In a posting in the Generational Dynamics forum, Higgenbotham describes what could happen:

"In the event that there is a run on the money market funds, it would work in a similar way to the bank runs in the 1930's, but at the same time it will be different. The reason it will be different is that a money market fund has a value of 1.000 dollars. That 1.000 value is called "the buck". If the value drops below 1.000 it is called "breaking the buck". The Federal Reserve has said that they will guarantee this value. With credit card charge-offs hovering at 10% and increasing, unemployment at nearly 10% (officially and we know this is a lie), auto sales getting smashed, and all these other problems, the true value of the money in these money market funds is dwindling. Now investors are coming to the point where they need to make a decision. They will need to decide whether they want to be in a money market fund that pays, say, 1.5% interest and has a value of, say, 0.950 but is still trading at 1.000 because the Fed has said they will guarantee the 5% loss (these are the "junk dollars" I have been talking about) OR whether they want to be in safe and secure dollars like treasury bills that pay, say, 0.3% interest and where a dollars worth of treasury bills is really worth a dollar."

Higgenbotham's comments serve as a warning to those who still have money in money market funds. The only safe investments today are cash, FDIC insured bank accounts, and short-term Treasury bills.

(Comments: For reader comments, questions and discussion, as well as more frequent updates on this subject, see the Financial Topics thread of the Generational Dynamics forum. Read the entire thread for discussions on how to protect your money.) (6-Jul-2009) Permanent Link
Receive daily World View columns by e-mail
Donate to Generational Dynamics via PayPal

Web Log Pages

Current Web Log

Web Log Summary - 2016
Web Log Summary - 2015
Web Log Summary - 2014
Web Log Summary - 2013
Web Log Summary - 2012
Web Log Summary - 2011
Web Log Summary - 2010
Web Log Summary - 2009
Web Log Summary - 2008
Web Log Summary - 2007
Web Log Summary - 2006
Web Log Summary - 2005
Web Log Summary - 2004

Web Log - December, 2016
Web Log - November, 2016
Web Log - October, 2016
Web Log - September, 2016
Web Log - August, 2016
Web Log - July, 2016
Web Log - June, 2016
Web Log - May, 2016
Web Log - April, 2016
Web Log - March, 2016
Web Log - February, 2016
Web Log - January, 2016
Web Log - December, 2015
Web Log - November, 2015
Web Log - October, 2015
Web Log - September, 2015
Web Log - August, 2015
Web Log - July, 2015
Web Log - June, 2015
Web Log - May, 2015
Web Log - April, 2015
Web Log - March, 2015
Web Log - February, 2015
Web Log - January, 2015
Web Log - December, 2014
Web Log - November, 2014
Web Log - October, 2014
Web Log - September, 2014
Web Log - August, 2014
Web Log - July, 2014
Web Log - June, 2014
Web Log - May, 2014
Web Log - April, 2014
Web Log - March, 2014
Web Log - February, 2014
Web Log - January, 2014
Web Log - December, 2013
Web Log - November, 2013
Web Log - October, 2013
Web Log - September, 2013
Web Log - August, 2013
Web Log - July, 2013
Web Log - June, 2013
Web Log - May, 2013
Web Log - April, 2013
Web Log - March, 2013
Web Log - February, 2013
Web Log - January, 2013
Web Log - December, 2012
Web Log - November, 2012
Web Log - October, 2012
Web Log - September, 2012
Web Log - August, 2012
Web Log - July, 2012
Web Log - June, 2012
Web Log - May, 2012
Web Log - April, 2012
Web Log - March, 2012
Web Log - February, 2012
Web Log - January, 2012
Web Log - December, 2011
Web Log - November, 2011
Web Log - October, 2011
Web Log - September, 2011
Web Log - August, 2011
Web Log - July, 2011
Web Log - June, 2011
Web Log - May, 2011
Web Log - April, 2011
Web Log - March, 2011
Web Log - February, 2011
Web Log - January, 2011
Web Log - December, 2010
Web Log - November, 2010
Web Log - October, 2010
Web Log - September, 2010
Web Log - August, 2010
Web Log - July, 2010
Web Log - June, 2010
Web Log - May, 2010
Web Log - April, 2010
Web Log - March, 2010
Web Log - February, 2010
Web Log - January, 2010
Web Log - December, 2009
Web Log - November, 2009
Web Log - October, 2009
Web Log - September, 2009
Web Log - August, 2009
Web Log - July, 2009
Web Log - June, 2009
Web Log - May, 2009
Web Log - April, 2009
Web Log - March, 2009
Web Log - February, 2009
Web Log - January, 2009
Web Log - December, 2008
Web Log - November, 2008
Web Log - October, 2008
Web Log - September, 2008
Web Log - August, 2008
Web Log - July, 2008
Web Log - June, 2008
Web Log - May, 2008
Web Log - April, 2008
Web Log - March, 2008
Web Log - February, 2008
Web Log - January, 2008
Web Log - December, 2007
Web Log - November, 2007
Web Log - October, 2007
Web Log - September, 2007
Web Log - August, 2007
Web Log - July, 2007
Web Log - June, 2007
Web Log - May, 2007
Web Log - April, 2007
Web Log - March, 2007
Web Log - February, 2007
Web Log - January, 2007
Web Log - December, 2006
Web Log - November, 2006
Web Log - October, 2006
Web Log - September, 2006
Web Log - August, 2006
Web Log - July, 2006
Web Log - June, 2006
Web Log - May, 2006
Web Log - April, 2006
Web Log - March, 2006
Web Log - February, 2006
Web Log - January, 2006
Web Log - December, 2005
Web Log - November, 2005
Web Log - October, 2005
Web Log - September, 2005
Web Log - August, 2005
Web Log - July, 2005
Web Log - June, 2005
Web Log - May, 2005
Web Log - April, 2005
Web Log - March, 2005
Web Log - February, 2005
Web Log - January, 2005
Web Log - December, 2004
Web Log - November, 2004
Web Log - October, 2004
Web Log - September, 2004
Web Log - August, 2004
Web Log - July, 2004
Web Log - June, 2004

Copyright © 2002-2016 by John J. Xenakis.