Hi Guys,
During my many postings on both MFO and its FundAlarm predecessor, I have assiduously promoted statistical learning and constant application when making mutual fund investment decisions. Statistical analysis methods add yet another tool to the decision making toolkit. The statistical discipline is one that has largely evaded many US investors, including some professionals. We could all benefit from an improved understanding in this sometimes complex and often ignored subject.
My campaign continues with this posting.
I know you all recall that Mark Twin famously said that “There are lies, damned lies and statistics.” And Gregg Easterbrook added that “Torture numbers, and they'll confess to anything.” But that’s a minority opinion designed to amuse and really should be disregarded by serious marketplace investors. I prefer to recognize Francis Galton’s admonition that “Whenever you can, count”. That simple statement is the basic rule of statistical analysis.
Galton is infamously remembered as the father of Eugenics, but he was also an accomplished amateur statistician who gave us the concept of regression-to-the-mean and documented it with endless studies. One such study included a comparison of the percentage of good looking women in London contrasted to those in Aberdeen, Scotland. Galton concluded that London sported the higher percentage of pretty women. Wow, talk about inviting a biting controversy.
Galton also invented the “Galton box” that demonstrated the normal distribution of random, independent events. Here is a Link to one such Internet demonstration that is coupled to equity market monthly returns. Note that the demonstration is a reasonable approximation of a normal bell-curve distribution except at the extremes. Fat tails exist, and a normal distribution assumption fails to capture that significant deviation. Black Swans do extensive wealth damage.
http://www.ifa.com/probability_machines.aspThis Index Fund Advisor (IFA) website offers many excellent investment videos mostly directed towards the neophyte investor. You might profit from repeated visits.
Enough about Galton. He serves as an intriguing introduction to my main purpose for this submittal. Nate Silver has just published an important statistical survey book that offers both statistical rigor and is comfortably accessible to the general public. The book honestly documents the merits of statistical analyses, and clearly identifies its many shortcomings. I encourage you to consider reading this fine introduction to statistical reality.
The Silver book is titled “The Signal and the Noise”. It was released only a few days ago. It is chock full of stimulating examples that engagingly define the rewards and pitfalls of statistical analysis. I anticipate that your investing skill set and acumen will benefit from being exposed to the book’s diverse areas of statistical research.
Behavioral researcher and “Nudge” coauthor Richard Thaler reminds us of a well-traveled Yogi Berra aphorism that “forecasting is hard, especially about the future”. Nate Silver constantly reminds us that experts are statistically as fallible as the amateur at this merciless task. In the investment discipline, CXO Advisory Group documents the unimpressive predictive record of so called market Gurus. They struggle to score a 50/50 record. Likewise, television pundits are no better than a fair coin toss.
Here is a Link to a Nate Silver book review that summarizes 12 “cool” takeaway lessons from “The Signal and the Noise”. I am part way through the book, and it yields many other insights that are not reported in the excellent referenced review.
http://www.businessinsider.com/the-signal-and-the-noise-nate-silver-2012-9Often, it is a challenge to separate the wheat from the chaff (or equivalently, the men from the boys, the sheep from the goats, the useful from the worthless, and numerous other idioms). Sometimes, it is an impossible assignment. Nate Silver intelligently writes “why so many predictions fail – but some don’t”. I encourage you to explore the “why” factor.
Silver’s book does a remarkable job of explaining and illustrating some of the more subtle aspects of statistical analysis. One such area is Bayesian analysis. The Bayesian method is an experimental analysis in which the predictions are continually revised as new evidence is introduced into the projection to modify existing probabilities. The Silver treatment of this somewhat obscure concept makes it understandable to us amateur statisticians. This section alone makes the book a wise and worthwhile purchase. By the way, the book can be bought for about 16 dollars at discount.
In another section, the author and statistician eats humble pie when he examines out-of-sample baseball data that tests his PECOTA ballplayer evaluation tool. Silver originally gained national recognition with his PECOTA formulation that assesses the future performance likelihood of young minor league players based on historical comparisons. His statistical tool was marginally successful, but was eclipsed by the superior record of professional baseball scouts. Practical experience matters greatly to sharpen forecasting wisdom.
“The Signal and the Noise” book can serve to guide when a reliable forecasting task is doable or not doable. Silver’s final chapter correctly concludes that “What you don’t know can hurt you”. Indeed, in the current chaotic investment environment that observation is particularly pertinent and salient.
As a Post Script permit me to note that Nate Silver is the founder and maintains the New York Times FiveThirtyEight statistical Blog. As you are aware that 538 is the United States Electoral College magic number, the Blog is totally statistical by design and totally political by purpose. It is exhaustively comprehensive and colorfully constructed. It is definitely worth a visit. The political statistical modeling cohort and their proponents will get an acid test of their methodology in about a month. Here is the Link to that superior website:
http://fivethirtyeight.blogs.nytimes.com/Observe that the site includes US Senate seat forecasts as an integrated component. The data graphs are fascinating. Their accuracy will be challenged on November 6.
Best Regards.
Comments