Expert Intuition & Machine Learning

{ Euclidean Q1 2014 Letter }


In our last letter, we explored human behavioral biases and why investors often (mis)behave in the highly correlated manner required for great investment opportunities to emerge.  That topic is important for Euclidean investors.  We believe that human nature – especially the way it causes people to focus on information other than companies’ operating results – is a big reason why the opportunity exists to outperform via value investment strategies. 

However, there are great investors who seem unmoved by these unproductive biases and exhibit deep expertise, as evidenced by long track records of superior performance.  In this letter, we want to explore these questions:  Where does expertise come from?  And, how does an understanding of expertise relate to our use of machine learning to evaluate investment opportunities?

2014 YTD Performance

Year-to-date through March, Euclidean Fund I, LP has advanced 1.72% net of fees in the context of the S&P 500 delivering a total return of 1.81%.  Since its inception, our partnership has returned +95.3%, comprised almost entirely of long-term capital gains, while the S&P 500’s Total Return has been +64.4%.  These gains translate into annualized compounded net returns of +12.6% and +9.2%, respectively.

Where Does Expertise Come From?

In his book Outliers, Malcolm Gladwell explored the enormous amount of practice required in a domain before expertise is attained.  After looking at individuals who had ascended to the top of their fields – including chess grandmasters, surgeons, and composers – Gladwell concluded that a prerequisite for world-class performance is 10,000+ hours of practice.  Across this amount of dedicated time, one can experience the wide range of situations, possibilities, and outcomes necessary to become really good at something.    

Gladwell’s book popularized concepts about expert intuition that were first introduced years ago by Herbert Simon.  Simon’s research showed that experts, through dedicated effort in a specific field, build an efficient mind that recognizes the nature of new situations and quickly identifies likely solutions.  This stands in contrast to his observations of novices who, with their limited set of experiences and training, tend to waste time considering irrelevant information and weighing bad choices.  Simon’s views are well-summarized by this often-cited quotation:

“The situation has provided a cue; this cue has given the expert access to information stored in memory, and the information provides the answer.  Intuition is nothing more and nothing less than recognition.” [1]

From Simon’s perspective, we find answers through pattern recognition.  An expert is one who has built a deep familiarity with the patterns of a given domain and, thus, has a robust body of work from which to reference when making decisions.

The Miracle on the Hudson – A Miracle or a Great Example of Expertise?

Five years ago, Captain Chesley “Sully” Sullenberger successfully landed his disabled airplane on the Hudson River, saving all 155 people on board.  His heroic landing provides an interesting example to consider when reflecting on the nature of expertise.

At 3:25p.m., on January 15, 2009, US Airways Flight 1549 took off from New York’s LaGuardia airport.  The plane was under the command of Captain Sullenberger, a former U.S. Air Force pilot who had logged almost 20,000 hours in flight over his career.  Two minutes later, the transcript from the flight recorder shows the co-pilot saying, “Birds.” [2] The cockpit of the Airbus 320 turned dark, and the pilots heard loud thuds as the plane struck a flock of Canadian geese.  The bird strike disabled both of the plane’s engines while the aircraft was still below 3,000 feet.  Within the next three minutes, Captain Sullenberger did the following:

  1. He attempted to reignite the engines while simultaneously radioing the control tower to request a return path to LaGuardia airport.
  2. When the engines would not restart, he asked if there was a closer, alternate runway where he could land.  The controllers offered him Teterboro Airport in New Jersey.
  3. He determined that they could not make it to Teterboro.
  4. He stopped communicating with controllers and successfully landed the plane on the surface of the Hudson River, saving all 155 people on board. 

What can this heroic landing teach us about the nature of expertise?  Dual-engine failures, such as what occurred on US Airways Flight 1549, are very rare, occurring perhaps once a decade. Captain Sullenberger had never directly experienced this type of situation, yet he successfully landed the plane.  So, if expertise involves pattern recognition and successfully applying a vast body of experience, was expertise applied in this instance?  Or, was his successful landing truly a miracle?

Gary Klein and the Sources of Power

Gary Klein is a prominent behavioral psychologist who built on Herbert Simon’s ideas by studying how people make decisions under conditions of pressure and uncertainty.  Prior to Klein’s research [3], the dominant academic frameworks for understanding optimal decision-making were comparative decision models.  The assumption was that experts have a mental spreadsheet whereby they score the costs and benefits of a wide variety of options to determine the best course of action. 

Klein observed something different.  He noticed that it was the novices, and not the experts, who made decisions in the comparative and deliberative way suggested by prevailing models.  He observed that novices employ logical, deductive and deliberative processes because they do not have a body of work from which to draw.  Experts, however, do not think about what to do.  They simply seek to understand the situation they are in and then they know exactly what to do.  They do not have to start from scratch because their experience gives them the answer.

Klein shared this example in an interview with Fast Company magazine [4]

"I had a conversation with an instructor pilot that really stuck with me," recalls Klein. "When he first started flying, he was terribly frightened. If he made a mistake, he'd die. He had to follow all of these rules and checklists in order to fly the plane correctly, and it was an extremely nerve-racking time. But at some point in his development, he underwent a profound change. Suddenly, it felt as if he wasn't flying the plane – it felt as if he was flying. He had internalized all of the procedures for flying until the plane had felt as if it was a part of him. He no longer needed any rules." 

This helps make sense of Captain Sullenberger’s reactions during the emergency’s first moments.  When his engines shut down, he did not stop to evaluate the pluses and minuses of various options.  He knew intuitively what to do and, you could say, he had all the right rules and checklists hard-wired in his mind.

When he considered whether to return to LaGuardia or fly to Teterboro, Captain Sullenberger described how he made his decision [5]:

“Did we have enough altitude and speed to make the turn back to the airport and reach it before hitting the ground?  There wasn’t time to do the math, so it’s not as if I was making altitude descent calculations in my head.  But I was judging by what I saw out the window and creating, very quickly, a three-dimensional model of where we were.”

He did not deliberate.  Instead, his direct expertise, from thousands of hours of flight time, allowed him to instantly determine how far he could glide the plane and know that it would not be far enough to reach the nearest airports.  His rapid decision, if delayed or incorrectly made, would have led to a bad outcome.

But what about the actual water landing?  Captain Sullenberger had never landed a plane in the water, and so, unlike with the actions he had taken up to this point, he had no direct experience on which to lean.  Instead, as a flight safety expert who taught classes to other pilots, he had studied how other pilots had successfully navigated challenging situations.  In his autobiography, he tells of how during World War II, Allied airmen had to ditch a great number of planes in the English Channel.  From their experiences, procedures were documented for maximizing one’s chances of having a successful water landing.  He recalls:

“The procedures called for the landing gear to be retracted rather than extended.  It described why an airplane should fly as slowly as possible, and why wing flaps should be down for impact.  It also called for the nose to be up in most cases.  These procedural guidelines remain in use today and were in my head on Flight 1549.”

Once he understood the situation he was in, Captain Sullenberger’s experience told him what to do.  He developed his pattern-recognition ability directly through flight time and also by mentally simulating the experiences of others.  These patterns represented chunks of knowledge that allowed him to successfully navigate a situation that he had never encountered.   

Calling this heroic landing "The Miracle of the Hudson" feels somewhat like an insult to Captain Sullenberger.  There was no miracle here.  The good outcome was, purely and simply, the result of his deep expertise.

Expertise, Machine Learning, and Selecting Long-Term Investments

When you learn math in school, a teacher might use a bunch of problems in class and for homework.  On the test, however, a good teacher will ask you to answer questions that you have never seen before.  Why?  Because the teacher does not care that you can memorize the answers to specific questions, but rather he wants you to comprehend important principles that will become part of your foundation for solving new problems in the future.

This helps makes sense of how Captain Sullenberger successfully landed his plane in the water, even though he had never done so before.  He had mastered the principles of flying and emergency landing, such that he successfully executed his first water landing.  We believe this example provides a good basis for understanding how certain investors may achieve above-average results over long periods of changing market conditions and a diverse set of companies.

We believe great investors become great through experience.  By making successful and poor investments, sitting on boards, and paying attention to investment history, a great investor builds a rich foundation of prior outcomes from which he can make sense of new opportunities and make informed decisions.  The nature of this foundation is a set of investment principles.  These principles form over time and are characterized by what generally distinguishes good investment outcomes from poor outcomes in the investor’s experience. 

One challenge investors face in developing these expert principles, however, is that there are irregular cycles in markets.  What works well in general across time does not generate good outcomes all the time.  Also, luck mixes with skill such that poor decision-making can sometimes result in good results. So, until an investor has participated in hundreds of investments and operated across several market cycles, he may be particularly susceptible to drawing bad conclusions from his own experiences.     

This is why we were interested in using machine learning to look for investment principles that would have done well across a variety of market cycles.  We felt that by looking at thousands of opportunities across time, we could reduce the influence of outlier successes and failures, and aspire to develop the pattern-recognition skills that an expert investor develops over decades of primary experience. 

It is helpful to deconstruct Euclidean’s approach in light of Simon’s quote that we introduced previously.  The quote again is: “The situation has provided a cue; this cue has given the expert access to information stored in memory, and the information provides the answer.  Intuition is nothing more and nothing less than recognition." [6]

Summary

Our questions were:  Where does expertise come from?  And, how does an understanding of expertise relate to our use of machine learning to evaluate investment opportunities?

Expertise seems to be the manifestation of a body of experience against which a current situation can be compared and understood.  In Simon’s words, expertise is pattern-recognition, and it develops over long-periods of practice and experience.  As with Captain Sullenberger’s water landing, expertise can also be built through studying and “mentally simulating” others’ successes and failures.  This closely resembles what Euclidean aspires to accomplish by using machine learning to build a foundation for evaluating individual companies as potential long-term investments. 

The ultimate test of the expertise embedded in our process will be our long-term performance.  We feel it is meaningful that by applying lessons learned from the past, we have done well in the fascinating market environment that emerged since we launched Euclidean in mid-2008.  Most of this period, however, has been characterized by inflating valuation multiples for domestic public companies.  As you might expect, in the past, there have been cycles of P/E expansion and contraction.   Thus, if the patterns of history remain an adequate guide to the future, we should not expect valuation multiples to forever rise or remain at heightened levels.

When pessimism builds to the point that multiples compress, we anticipate that being in the right companies and paying the right prices will take on paramount importance.  We expect that we will eventually find ourselves in this type of long-term period.   When we do, we look forward to demonstrating the expertise embedded in our investment process.

*****

We greatly value the privilege of managing a portion of your hard-earned assets and want you to be an informed Euclidean investor.  We are available to discuss the content shared here, individual positions in our portfolio, or any other questions you might have.  Please call us at any time.  We enjoy hearing from you. 

Best Regards,

John & Mike

[1]  "What is an Explanation of Behavior?", Psychological Science, May 1992.

[2]  "Loss of Thrust in Both Engines After Encountering a Flock of Birds and Subsequent Ditching on the Hudson River. US Airways Flight 1549 Airbus A320-214, N106US. Weehawken, New Jersey, January 15, 2009." Aircraft Accident Report, NTSB/AAR-10/03, PB2010-910403 Adopted May 4, 2010 National Transportation Safety Board, Washington, DC.

[3] Sources of Power, How People Make Decisions, by Gary Klein. The MIT Press. 1999.

[4] “What’s Your Intuition,” by Ed Klein, Fast Company, August 2000.

[5] My Search for What Really Matters, by Captain Chesley “Sully” Sullenberger with Jeffrey Zaslow. Harper-Collins. 2009.

[6]  "What is an Explanation of Behavior?", Psychological Science, May 1992.


We share these numbers because they are easy-to-communicate measures that show the results of our systematic process for buying shares in historically sound companies when their earnings are on sale. [7] [8]

It is important to note that Euclidean uses similar concepts but different measures to assess individual companies as potential investments.  Our models look at certain metrics over longer periods and seek to understand their volatility and rate of growth.  Our process also makes a series of adjustments to company financial statements that our research has found to more accurately assess results, makes complex trade-offs between measures, and so on.  These numbers should, however, give you a sense of what you own as a Euclidean Investor.  In general, higher numbers for these measures are more attractive.  The key measures are:

  1. Earnings Yield – This measures how inexpensive a company is in relation to its demonstrated ability to generate cash for its owners. A company with twice the earnings yield as another is half as expensive; therefore, all else being equal, we seek companies with very high Earnings Yields.  Earnings Yield reflects a company’s past four-year average earnings before interest and tax, divided by its current enterprise value (enterprise value = market value + debt – cash).
  2. Return on Capital – This measures how well a company has historically generated cash for its owners in relation to how much capital has been invested (equity and long-term debt) in the business.  At its highest level, this measure reflects two important things.  First, it is an indicator of whether a company’s business is efficient at deploying capital in a way that generates additional income for its shareholders.  Second, it indicates whether management has good discipline in deciding what to do with the cash it generates.  For example, all else being equal, companies that overpay for acquisitions, or retain more capital than they can productively deploy, will show lower returns on capital than businesses that do the opposite.  Return on Capital reflects a company’s four-year average earnings before interest and tax, divided by its current equity + long-term debt. 
  3. Equity / Assets – This measures how much of a company’s assets can be claimed by its common shareholders versus being claimed by others.  High numbers here imply that the company owns a large portion of its figurative “house” and, all else being equal, indicates a better readiness to weather tough times.
  4. Revenue Growth Rate – This is the annualized rate a company has grown over the past four years.  

[7] All Euclidean measures are formed by summing the values of Euclidean’s pro-rata share of each portfolio company’s financials.  That is, if Euclidean owns 1% of a company’s shares, it first calculates 1% of that company’s market value, revenue, debt, assets, earnings, and so on.  Then, it sums those numbers with its pro-rata share of all other portfolio companies.  This provides the total revenue, assets, earnings, etc. across the portfolio that are used to calculate the portfolio’s aggregate measures presented here. 

[8] The S&P 500 measures are calculated in a similar way as described above.  The market values, revenue, debt, assets, earnings, etc., for each company in the S&P 500 are added together.  Those aggregate numbers are then used to calculate the metrics above.  For example, the earnings yield of the S&P 500 is calculated as the total average four-year earnings before interest and taxes across all 500 companies divided by those companies’ collective enterprise values (all 500 companies’ market values + cash – debt).


Euclidean’s Largest Holdings as of March 31, 2014

We provide this information because many of you have expressed an interest in talking through individual positions as a means of better understanding how our investment process seeks value.   

In our last letter, we wrote that 23.8% of the Fund’s capital was invested in the for-profit education industry at year-end.  This has since changed to 22.5% as of March 31, 2014.

We are available to discuss these holdings with you at your convenience.  We are happy to explain both why our models have found these companies to be attractive as well as our sense of why the market has been pessimistic about their future prospects. 

Euclidean’s 10 largest positions as of March 31, 2014 (in alphabetical order)

1.     Apollo Education Group - APOL

2.     Avnet - AVT

3.     Big Lots - BIG

4.     Delek US Holdings - DK

5.     DeVry Education Group - DV

6.     Hi-Tech Pharmacal - HITK

7.     Humana - HUM

8.     Preformed Line Products - PLPC

9.     Tyson Foods - TSN

10.  USANA Health Sciences - USNA