My second-oldest daughter Katie's post on Igon Values is very relevant to our field. Who are the useful idiots in our field and how can their self-serving posings be used for profit? Let us start in the Midwest. Or the islands.

Alston Mabry adds:

I thought the "igon values" thing was a joke, but upon reading Pinker's review of Gladwell, I find it isn't. For one who writes about research, as Gladwell does, to not be able to sit down and work linear algebra problems is fine (me neither), but to not know that there is a word, "eigenvalue", which may arise in conversation with scientists — that's embarrassing.

The issue of "igon values" bears on the more general issue of knowledge production and dissemination. I like watching science shows, but even when I'm watching a high-quality show like NOVA, I'm wondering, "Where are the folks, with the same level of expertise, who think at least some of this is crap? What is their critique?" It often seems that a well-communicated disagreement can help the audience understand more of what's important. And actually, a good example is Pinker's review of Gladwell.

In most or all fields of research, one could assemble a group of experts who have similar training and knowledge, but who disagree on important points in the field, especially at the boundaries of discovery. Then you have popular writers trying to understand and condense some field into book form — but if the experts don't agree, how can the popularizers possibly be "right"?

Stefan Jovanovich writes:

Mr. Mabry is far too kind. Scientists talk their books all the time. A scientific reputation is made by presenting a theory that is striking, original and difficult to evaluate.  That theory becomes the scientist's brand. His/her future in academia is tied to the success of that brand.  Few, if any scientists, are crazy/honest/selfless enough to challenge the truth of their brand.  Hence, Heisenberg's comment that "the progress of science can be measured by professors' funerals."

In science, practitioners would rarely be lying about what they are doing. But in markets, who tells the "truth," unless that truth is consistent with the talker's book? Then there is the sheer volume of new "knowledge" produced on a daily basis. How does one cope?

Bruno Ombreux writes:

H G G JrHere is a suggestion. I believe it is linked to the Igon.

I am halfway through a book that is dealing with all these issues: Scientific Method in Practice by Hugh G. Gauch. It sheds light and fosters reflections on such things as scientific questions and methods, disproving or proving hypotheses…

It is a book about science. Since good trading is a science, this is a book about good trading too. Good trading is necessarily scientific, because good trading requires good predictions. Only science can yield good predictions. If trading is not scientific, it can't be good.

This is also a philosophical book. After a few chapters, I have enough philosophical ammunition to completely destroy the Black Swan school, on epistemological grounds. The Black Swan ideas that we cannot have models that work, that variance is either infinite or undetermined, are just as naive, and far less nuanced, version of David Hume's radical skepticism. In one sentence: we can't know anything. Scientific Method in Practice advises not to waste time arguing with radical skeptics. They are not targeting science, but common sense. Common sense is literally what humans can sense in common. In this case, we all can measure variances. Common sense is a key presupposition of science. Without common sense, there can be no science. Without science, there cannot be any debate between scientists and radical skeptics, since the later are saying in effect that the former don't exist.

Incidentally, the fat tails debate wonderfully illustrates one problem mentioned in the book, that is the underdetermination of theory by data. Observing fat tails, I can find offhand a bunch of explanations:

  1. power law
  2. slowly converging normal
  3. Student
  4. truncated Levy flight
  5. mixture
  6. Markov switching model
  7. Agent-based dynamics

The same evidence produces a handful of theories. We are confronted to the issue of theory choice. In this case, I would start by getting rid of those that don't make predictions. Power laws would be the first casualty.

EigenvaluesNow, on to another book recommendation: a first-year course in linear algebra and as such is related to the original topic. "Igon values for dummies," if you want. And it is free. This is a useful book for traders, because it is impossible to understand any recent article on economics or statistics without at least a passing knowledge of linear algebra.

I don't think mastery of Igon values is required to trade well, but other concepts can be very useful. For instance, the notion of projections, covered in chapter 3.VI, really helped me understand multicollinearity in regressions. Multicollinearity is the rule in financial time series. Often, its presence is not a problem, but you'd better know about it and when it can be a problem.

Combining this book's chapter 3.VI.2 about Gram-Schmidt Orthogonalization, with chapter 3.2.3 of this other free book, one gets a clear understanding of multicollinearity.

Jack Tierney writes:

 Being unfamiliar with eigenvalues (whether spelled correctly or not) led me to follow the threads in Katie's article. Those threads, in turn, led to still others. I finally landed on this.

The author laments the increasing propensity of Rhodes Scholars to go into the world of finance as opposed to some of the nobler scientific fields that once claimed so many of those blessed by old Cecil's beneficence.

"This break in an almost century-old pattern coincided with great increases in occupational earnings differentials, which have continued to grow, seemingly exponentially…the differentials in earnings…were often rationalized by Rhodes scholars as reasonable additional compensation to balance the lower standing of business jobs among their peers. "When differentials could become a hundredfold or far more — and as investment banking and similar firms started to recruit young Rhodes scholars who had degrees in math, physics or even history, English and theology — the yawning prospective wealth chasm understandably became impossible for many to ignore…"

So there we have it. Offer enough money and even the brightest will sell out. Let a dilettante like Gladwell emulate them, though, and the wrath of the informed will be merciless (just follow some of the threads and you'll discover that Kate's handling of Gladwell was relatively humane).

However, numerous responses seemed refer to the incalculable worth of the scientific method and were it adhered to, we would all be much better off and far less likely to be exposed to the ditherings of Gladwell et al.

Back in '93 a remarkable book written by a woman embittered by her brother's courtroom experiences hit the best seller list. It was "Whores of the Court" and detailed the lengths to which those supposedly trained in the scientific method quite easily (and lucratively) sold their conclusions. Each side could present "experts" with similarly impressive credentials; each side had access to the same evidentiary material; yet their conclusions could not
have been more different.

It might be legitimately argued that psychiatrists/psychologists aren't scientists in the pure sense of the word. Currently, however, we have scientists whose credentials most definitely measure up. Yet on issues ranging from the efficacy of ethanol to global warming to the amount of oil left within the earth's crust, their conclusions couldn't be more disparate. To put it bluntly, our scientists' opinions are for sale and this is occurring as government policy is
more and more determined by their conclusions.

Whose opinions are the most sought after and well rewarded (at least through speaking engagements, articles in the mainstream journals, and in research grants)? Generally, those whose views are the most dire or the least apocalyptic. This, in itself, is a sad development. But increasingly scientists whose expertise lay elsewhere are chiming in on one side or the other. As a result we are faced with promotions that announce that "X Number of PhDs Support Global Warming Theory", or "Y Number of PhDs Claim Peak Oil is a Sham."

I am increasingly exposed to individuals who claim (and firmly believe) that their opinion is as good as anyone else's, that it's unnecessary to study both sides of an issue, that it is quite OK to shout down a speaker whose views diverge from yours, and that it's quite alright to do whatever it takes to get whatever it is one wants.

In such a world, is Gladwell to be condemned or lauded? Are the newly minted Rhodes Scholars so misguided in pursuing wealth? Are scientists who missed the gravy train to be faulted for making a last mad dash for the gold ring on the caboose? Was Linus Pauling correct in observing that peers are nothing more than people who pee together?

Alston Mabry adds:

This post reminded me of the book "Psychology of Intelligence Analysis" which contains these guidelines:

"Start out by making certain you are asking–or being asked–the right questions."

"Relying only on information that is automatically delivered to you will probably not solve all your analytical problems."

"Do not be misled by the fact that so much evidence supports your preconceived idea of which is the most likely hypothesis. That same evidence may be consistent with several different hypotheses."

"Proceed by trying to reject hypotheses rather than confirm them. The most likely hypothesis is usually the one with the least evidence against it, not the one with the most evidence for it."

Chris Tucker replies:

Wow.  Some great reading in that book.  Thanks, Al. This is from Chap. 2 "Why Can't We See What Is There To Be Seen?":

People tend to think of perception as a passive process. We see, hear, smell, taste or feel stimuli that impinge upon our senses. We think that if we are at all objective, we record what is actually there. Yet perception is demonstrably an active rather than a passive process; it constructs rather than records "reality." Perception implies understanding as well as awareness. It is a process of inference in which people construct their own version of reality on the basis of information provided through the five senses.

This is so important in my line of air traffic control.  I am constantly telling trainees that listening is not something that happens to them, it is something one must actively engage in.  Upon hearing a pilot read back a clearance, whether it be an altitude, heading, speed or route, one must pay close attention to what is said and to check it against what is expected to be heard.  It is common for trainees to simply assume that they heard the correct readback and disconcerting to them when it is pointed out that this was not the case.  We spend a great deal of time teaching them how to listen attentively.

Another facet that I have mentioned before is teaching them to get the data from the scope — to look at groundspeeds and recognize overtakes, to look at altitudes and calculate rates of climb or descent, to look at aircraft types and make hypotheses about expected performance, to look at routes and destinations and see who has to get below whom, and to create plans based on all of these.  And then to observe and check hypotheses, again and again to make sure that what one expected to happen is really happening.  And if not, how to take steps to create the reality one intends.

The key to improvement in these areas is a combination of repeated exposure and active thinking about the available data. Exposure alone can make some tasks become automatic, but active thinking and attentiveness can accelerate learning and skill acquisition.

Phil McDonell comments:

Gladwell self styles as a translator from the arcane indecipherable world of science to the everyday world of business and laymen. A good translator must understand the vocabulary of the original source language and must have a command of the vocabulary of the target language. However a command of the two relevant vocabularies is not sufficient. If it were computers would be the best translators.

What Gladwell lacks is semantic comprehension. It is often not sufficient to merely translate the words without a deeper understanding of the content. His Igon Value mistake is a glaring example.

Clearly his substitution of Igon Value for eigenvalue comes from only hearing the word as opposed to actually reading it in a book. Perhaps someone explained it in a phone or lunch conversation and Gladwell seized on it as an interesting buzz word.

Eigenvalues are actually a very beautiful construct in linear algebra. A simple intuitive way to look at them is the amount by which a quantity is stretched in a certain dimension. Suppose a stock or mutual fund has a beta of 2 and an alpha of zero. The equation is:

stock return = 2 * market return + zero

The eigenvalue for the above system is simply 2 because it stretches the market return by a factor of two.

The idea generalizes to 2 or more dimensions. Each dimension of a linear system has its own eigenvalue. If you have ever looked at yourself in a fun house mirror then you can understand this idea. The mirror that makes you look tall and thin has a stretching eigenvalue in the vertical direction and a shrinking eigenvalue (<1) in the horizontal direction.

Like the fun house mirror a matrix can be thought of as a transformation or mapping of one image to another. If one takes the eigenvalues of a matrix and multiplies them together the product acts much like a volume just as length times width of a rectangle gives the area. In Linear Algebra this volume is called the determinant. If any of the eigenvalues is zero then one of the dimensions has collapsed. It also means the determinant will be zero, the system of equations cannot be solved and any regression will be meaningless.

I have never seen any financial model take into account a determinant. Yet there seems to be a grudging acceptance of the idea that when a financial panic hits all the correlations approach 1 as people seek liquidity by whatever means necessary. Rather than simply look at risk from a simple beta model or VAR approach perhaps the proper way to model disaster is the determinant where all the risk are multiplied together.

Dr. McDonnell is the author of Optimal Portfolio Modeling, Wiley, 2008





Speak your mind


Resources & Links