Think Twice

Think Twice

TT1:  Inside/Outside View

TT2: Biases – Overview

TT3:  Experts and the Crowd

TT4:  Situational Awareness

TT5:  More is Different

TT6:  Evidence of Circumstance

TT7:  Grand Ah Whooms

TT8:  Sorting Luck from  Skill

TT1:  Inside/Outside View

We have a tendency to favor the inside view over the outside view.  Inside view considers a problem by focusing on the specific task and by using information that is close at hand and makes predictions based on that narrow and unique set of inputs.  These inputs may include anecdotal evidence and fallacious perceptions.  These assessments are generally built in the evaluation model and usually paints an optimistic view.

Outside view asks if there are similar situations that can provide a statistical basis for making a decision.  Rather than seeing a problem as unique, it looks to find comparable problems, and if so, what happened.  Outside view is generally considered an unnatural way of thinking as it disregards cherished, or close, information.  Outside view creates a valuable reality for decision makers.

Three reasons why people take the inside view:

1.      Illusion of superiority:  the least capable people tend to have the largest gap between what they think they can do and what they can achieve.  When people do rank themselves poorly, then tend to dismiss those traits they rank as poor.

2.     Optimism:  Most people see their future as brighter than others.

3.     Illusion of control:  People behave as chance events are subject to their control.

People who believe that they have control have the perception that their odds of success are better than they actually are.  People who don’t have a sense of control don’t experience the same bias.

Anecdote:  short and interesting story about a real person or event.  An account considered unreliable.

Positive anecdotes paired with a low statistic tends to be more powerful than a negative anecdote paired with a high percentage of success.  We are susceptible to anecdotes, especially from friends, and sources we consider reliable.  Although the anecdotes have no real merit to the subject at hand,   Anecdotes guide us to the inside view.

Planning fallacy:  People have difficulty estimating jobs and costs.  When they are wrong, they usually underestimate the time and expense.  Only about ¼ of people use base rate data when making projections.

When people are poor about making judgments of themselves, they tend to be pretty good at judging other people.  People tend to think of themselves as different or better than the people around them.

How to invoke the outside rule:

1.      Select a reference class.  Find a group broad enough to statistically encompass like events, but narrow enough to be useful.

2.     Assess the distributions of outcomes.  Take a close look at success and fail rates.  Study the distributions, find the average and the extremes.  The statistical rate of success and failure must remain reasonably stable over time for the reference class to be valid.  Also, be careful if small perturbations can lead to large scale change.  Cause and effect are difficult to determine thus affecting the ability to rely on the reference class.

3.     Make a prediction.  With valid data and an awareness of distributions, one can make a forecast.  Estimate your chance of success or failure.  Realize your probabilities will be somewhat optimistic.

4.     Assess the reliability of your predictions and fine tune.  When cause and effect is clear, you can have more confidence in your prediction.  If predictions are off, then adjustments need to be made.  Best decisions arise form sameness, not uniqueness.

TT2: Biases – Overview

Anchoring biases was reviewed:  Even after explaining the bias, you can still view it in effect.  Psychologists believe it is attributed to anchoring being a predominantly subconscious trait.  Anchoring is symptomatic of insufficient considerations to alternatives.  Failure to entertain options or possibilities can lead to dire consequences.  Including unwarranted confidence in final decisions.

Reasoning:  People reason from a set of premises and only consider compatible possibilities.  As a result, people fail to consider what they believe is false. 

A problems presentation significantly influences our approach, evaluation and choice.  How its described, how the subject feels about it and their individual knowledge all contribute.  Supported by prospect theory.

Mental models are internal representations of an external reality.  They are incomplete and inherintly trade detail for speed.  Mental models replace more cumbersome reasoning processes.  Ill conceived mental models are difficult to alter and consistently provide poor decisions.

Our brains are very good at getting answers quickly and often efficiently.  By getting the right solution expeditiously means homing in on what seems to us to be the most likely outcomes and leaving out a lot of what could be.  These processes have worked well for us in the natural world, but is not as well suited for today’s environment.

Key example:  giving real estate agents houses to appraise.  Same house, different initial listings.  Most agents anchored on the listing price, then adjust.  However, agents insisted their appraisals were independent.  The bias is pernicious because we are so unaware of it.

Representativeness heuristic:  Often rush to conclusions biased on representative categories in our mind.

Availability Heuristic:  Judging the probability or an event based on what is readily available in memory.  We tend to heavily weigh a probability if we’ve recently seen it.

We also tend to extrapolate inappropriately from past results.  Though statistically a few items don’t represent a confirmed trend, our brain sure thinks they do.  Tough to overcome.  Pattern recognitions deeply rooted in our minds in conjunction with predictive behaviors.

Models based on the past assume the future will be like the past.  As such, we tend to anticipate without giving suitable considerations to other possibilities.

Cognitive dissonance:  the rigidity that comes with the innate human desire to be internally and externally consistent.  Arises when an individual holds two cognitions (attitudes, beliefs, opinions) that are psychologically inconsistent.  The dissonance causes mental discomfort that our minds seek to reduce.  Usually wants to resolve with the least amount of mental effort.

The primary way we deal with it is to justify our actions.  Author indicates a little self delusion is ok when the stakes are low, but significantly problematic when the states are high.

Confirmation bias:  Seeking outside information to affirm a position or belief we hold.  The bias offers us a couple benefits:

1.      Permits us to stop thinking about it, and gives us a mental break.

2.     Frees us from the consequences of reason and possibly some responsibility.

The first allows us to avoid thinking, the second to avoid acting.

However, people tend to be selective in their exposure and retention.  Reinforcement of ideas tend to create a positive stimulus.  Contrary ideas tend to form negative emotions.  As its easy to filter out contrary ideas we should be aware of the bias and keep problem solving in a broader context.

Some discussion on stress, in the form of a time bias.  Stressful situations that focus people on the short term hamper them from making decisions for the long run.

Incentives:  A factor, financial or otherwise, that encourages a particular decision or action.  Incentives can cause conflicts of interest that compromises a persons ability to consider alternatives.

Subprime Example:  Incentives that led to it.

1.      People with poor credit could buy a house.

2.     Lenders earned fees on the loans they made to people with poor credit.

3.     Banks bought individual mortgages and bundled them to investors, for a fee.

4.     Rating agencies were paid a fee to rate the mortgage backed securities.  Many of which were deemed AAA.

5.     Investors in AAA rated mortgage backed securities earned higher returns than other AAA investments.

The subprime event clearly illustrates that what may be beneficial for parts of the system are not necessarily beneficial for the system as a whole.

Key Example:  Accountants to review accounting procedures at a firm.  Half were told they were hired by the company, half by an outside firm.  Those hired by the firm were 30% more likely to be less critical

Once again, the bias/incentive is so ingrained that people are completely oblivious to it.

How to avoid biases:

1.      Explicitly consider alternatives.

2.     Seek dissent.

3.     Avoid making decisions at emotional extremes.

4.     Keep track of previous decisions.

5.     Understand incentives.

Humans tend to consider too few alternatives.  Most cases present an obvious choice.  However, biases hinder the ability to objectively view reality and the hindermint increases with the complexity of the system and the increase in uncertainty.

TT3:  Experts and the Crowd

It is impossible to find any domain in which humans clearly out perform crude extrapolation algorithms, much less sophisticated statistical ones.

Despite the above statement, we still tend to rely on experts.  Most people have a hard time assimilating broad statistical evidence into the judgment at hand.  Wisdom of crowds generally provide better forecasts than experts.

Diversity prediction Theorem:  collective error = individual error – prediction diversity.

Squares are used so that negative and positive values do not cancel each other out.

Average individual error capotes the accuracy of individual guesses.

Prediction diversity reflects the dispersion of guesses, or how different they are.

Collective error is the difference between the correct answer and the average guess.

You can reduce the collective either by increasing ability or by increasing diversity. Both ability and diversity are important.  Implies healthier understanding of the markets. 

A diverse collective always beats the average and frequently beats everyone.

For the crowd to be effective, three things must be in place.  Diversity, aggregation and incentives:

Incentives help reduce individual errors.

Diversity reduces the collective error.

Aggregation assumes everyone’s information is included.

Inappropriately relying on intuition.  Intuition can play into successful decision making.  The trick is to know when it guides you right and when it leads you astray.

Kahneman two systems of decision making.

1.      Experiential system: fast, automatic, effortless, associative, and difficult to control or modify.

2.     Analytical system:  slower, serial, effortful and deliberately controlled.

Experiential system uses perception and intuition to generate impressions of objects or problems, and the individual may not be able to explain them.  An analytical system involved in all judgments where or not the individual is conscious of the decision.

So, intuition is a judgment that reflects an impression.

Experts collect experiences in which analytical systems is aggregated into experiential system.  Effectively, experts internalize the salient features of the system they are dealing with, which frees higher brain functions.

Traits of experts:

1.      Perceive patterns in their expertise.

2.     Solve problems faster than novices do.

3.     Represents problems at a deeper level.

4.     Can solve problems qualitatively.

Intuition works well in stable environments where feedback is clear and cause and effect relationships are linear.

Intuition fails with a changing system, especially if it deals with phase transitions.  Intuition is loosing relevancy in an increasingly complex world.

Note:  True experts become experts by actively using deliberate practices to train their experiential system.  It’s repetitive, has clear feed back and not a whole lot of fun.  Most experts do not come close to meeting these traits.

Crowds nor statistics should be used with blind faith.  In many cases experts use numbers with no predicative value in their evaluations and forecasts.  This effect, the mismatch, extends to other areas where people evaluate and rank people on questions or tests that do not forecast performance.  Example: Interviews for jobs

If breakdowns occur in the collective error, it can multiply and dramatically undercut the wisdom of crowds.  Diversity is usually the main culprit as we are social and imitative.  Information cascades occur when people start following others.  The cascades explain fads, booms and busts.

Diversity can also break down in smaller groups if there is a dominate personality and/or the absence of facts, or if the group tends to think alike.

How can you make the expert squeeze work for you:

1.      Match the problem you face with the most appropriate solution.  What approaches are the best.

2.     Seek diversity:  knowing a little about a lot tends to make you a better predictor.  However, it lacks the depth of those who know a lot about a little.  What separates levels of expertise is not what experts think, but how they think.  Diversity increases predictive ability.

3.     Use technology when possible.  Identify the nature of the problem and determine the best method to solve.  Technology can do better if data is valid and helps eliminates biases.

TT4:  Situational Awareness

Group decisions, even poor ones, influence our individual decisions.  About one third of people will conform to the group answer, even if it is wrong.  The real question is what is going on in the heads of the people who do conform.

1.      Distortion of Judgment:  People conclude their perception is wrong and the groups decision is right.

2.     Distortion of Action:  People suppress their own knowledge to conform with the group.

3.     Distortion of Perception:  Not aware that the majority opinion distorts their estimates.

Studies indicate the groups decision affects the individual’s perception, not necessarily their higher functions of judgment or action.  Seeing becomes what the group wants you to see.  In people who remained independent from the groups decision had higher activity in the amygadala, which is known for preparing for action (fight or flight).  Suggests that standing alone is unpleasant.

Our situation influences our decisions enormously.  Hard to avoid as this is largely unconscious.  Making good decisions in the face of subconscious pressure requires a high degree of knowledge and self awareness.

If someone gives you a cue word, it creates an associative path and it will color your decision.

People around us also influence us.

1.      Asymmetric information.  Information someone knows, but you don’t.

2.     Peer pressure or the desire to be part of a group.  A collection of interdependent individuals.  Conformity does lead to diversity breakdown within a group.

Fundamental attribution error:  the tendancy to explain behavior based on an individual’s disposition versus their situation.  We naturally associate bad behavior with bad character, except with ourselves.  Although we like to believe our choices are independent of your circumstances, the evidence strongly suggests otherwise.

Situational awareness does not manifest itself the same across cultures.  Eastern versus western thought.

1.      Easterners focus on environment, westerners on the individual.

2.     Westerners believe they are more in control.

3.     Easterners are more open to change.  The differences are attributed to two distinct philosophical decisions.

Example:  Shoot the boss due to some action on the bosses part and how the media treats it.  The west focuses on the individual; bad temper or mentally unstable.  East is focused on relationships.  Didn’t get along with coworkers, maybe influenced by a similar act.

It is a mistake to believe our decisions are independent of our experiences (situation).

Subliminal advertising does not work as the link is weak.  Effective priming must be sufficiently strong and the individual must be in an environment that sparks behavior. 

Default setting:  People tend to go with the default cue over opting out.

It is a mistake to perceive that people decide what is best for them independent of how the choice is framed.  Choice architecture:  We can nudge people toward a particular decision based on how the choices are arranged for them.

Structuring choices creates a context for decision making.  Can also be applied to large groups.

It is a mistake to rely on immediate emotional reactions to risk instead of an impartial judgment of possible future outcomes.

Affect:  How positive or negative emotional impression of a stimulus influences decisions.

How we feel about something influences how we decide about it.  If we feel strongly about something, we tend to go with system 1 (fast, automated, reflexive) rather than system 2 (slow, analytical, reflective).  Highly situational and largely unconscious.

Two core principles.

1.      When outcomes of an opportunity are without potent affective meaning, people tend to overweight probabilities.  Example:  system that saves 150 lives versus system that saves 98% of 150 lives.  The 98%, although worse than without it, scored as a better system as the 98% proved to be a potent positive stimuli.

2.     When out comes are vivid, people pay too little attention to probabilities and too much attention to the outcomes (such as gambling).

Probability insensitivity:  paying more attention to the outcomes than to the probability of the outcome.

Final mistake:  Explaining behavior by focusing on people’s disposition, rather than considering the situation.   Restatement of attribution error.  Situation is generally much more powerful than people, especially westerners, acknowledge.  The combination of the group and the setting lay the groundwork for behavior that can dramatically deviate from the norm.

Factors affecting situation:

1.      Situational power is most likely in novel settings, where there are no previous behavioral guidelines.

2.     Rules which can emerge through interaction, can create a means to dominate and suppress others, because people justify their actions as only conforming to their rules.

3.     People that are asked to play a certain role, for a prolonged period of time, may not be able to break their roles later on.

4.     In situations that lead to negative behavior, there is often an enemy.

Power of Inertia:

Inertia:  Resistance to change, also show s the situation shapes real world decisions.

Discussion on ‘we’ve always done it that way’. Can be changed through a fresh look, often by outsiders.

Discussion on how strict regulation can hamper change and sometimes counter common sense.

Coping with situation bias:

1.      Be aware of your situation.  A) Creating a positive environment and focusing on processes, keeping stresses in check, having thoughtful choice, and make sure to diffuse forces that encourage negative behavior.  B) coping with subconscious influences.  Requires awareness of the influences, the motivation to deal with it and the willingness to address possible poor decisions.

2.     Consider the situation first and the individual second.  Evaluate the decisions of others by starting with the situation, and then turning to individuals.

3.     Watch out for the institutional imperative:  imitating what your peers are doing.  Companies tend to want to be in the in group.  Alsok, incentives.  Executives can reap financial benefits from following the crowd.  If this dynamic is in effect, it is difficult not to be drawn to it..

4.     Avoid inertia.  Periodically review your processes.

We tend to:

1.      Think of ourselves as good decision makers.

2.     Think we weigh the facts and consider the alternatives and select the best course of action.

3.     Think we are immune from the influences of others.

4.     Convince ourselves that facts and experiences carry the day.

In reality, decision making is

1.      Inherently a social exercise.

2.     Primes, defaults, affect and the behaviors of those around us influence our decisions, mostly at an subconscious level.

A thoughtful decision maker becomes aware of the biases and influences and finds ways to manage them.

TT5:  More is different.

Complex Adaptive System:

1.      Consists of a group of heterogenous agents.  (neurons in the brain, bees in a hive).  Heterogeneity means each agent has different and evolving decision rules that both reflect the environment and attempt to anticipate change in it.

2.     The agents interact with each other and their interactions create structure.  Also referred to as emergence.

3.     The system that emerges behaves like a higher level system and has properties and characteristics that are distinct from those of the underlying system themselves.

Even though individuals may be inept, the colony as a whole is smart.  The whole is greater than the sum of the parts.

The behavior of complex aggregates of elementary particles is not to be understood in terms of simple extrapolations of a few particles.  At each new level of complexity, new properties appear.  Don’t study the ant, study the colony.

Humans have a deep desire to understand cause and effect.  In complex adaptive systems, there is no simple method for understanding the whole by studying the parts.  Searching for simple agent level causes of system elevel effects is useless.

When a mind seeking links between cause and effect meets a system that conceal them, accidents will happen.

First mistake:  Inappropriately extrapolating individual behavior to explain collective behavior. 

Example:  earnings capture the headlines as the driving force in determining share price.  However, studies indicate cash flows may be a better driver.  Both address the question from different perspectives.

Earnings is what the media focuses on.  Economists look at how the market is behaving.  One group focuses on components, one is focused on the aggregate.  The opinion of the market is more relevant than listening to a few individuals.  Just because individuals can be consistently wrong, doesn’t mean the whole is wrong.

Market irrationality does not flow from individual irrationality.  Collective behavior matters more.  You must carefully consider the unit of analysis to make a proper decision.

In a complex system that has many interconnected parts, a small tweak can have significant impacts to the rest of the system.  Or unforeseen consequences.

Second mistake:  How addressing one component of the system can have unintended consequences for the whole.  Even those that arise from the best of intentions.  The decision making challenge remains for a couple of reasons:

1.      Our modern world has more interconnected parts.  So we encounter them more often and usually with greater consequences.

2.     We still attempt to resolve problems in complex systems with a naïve understanding of cause and effect.

Third Mistake:  Isolating performance without proper consideration of the individuals surrounding system.  Analyzing results requires separating relative contributions of the individual from the environment.  When we err, we tend to overstate the role of the individual.

All three mistakes have the same root.  A focus on an isolated part of a complex adaptive system without an appreciation of the system dynamics.  In addition the frequency and scope are only accelerating in our world which means you will encounter them more often.

Advice:

1.      Consider the system at the correct level.  ‘More is different.’  Most common mistake is to extrapolate from the individual agents rather than the whole.  To understand the stock market, understand it at the market level.

2.     Watch for tightly coupled systems.  Complex systems that are tightly coupled can have rapid failure.  One process in the system is closely followed by another.  Nuclear power plant example.  Most complex adaptive systems are loosely coupled.  The removal of a few agents ahs little consequences.  Markets tend to be loose, but boom and busts illustrate when the market becomes tight.  Regulation (good or bad) can couple markets tightly.

3.     Use simulations to create virtual worlds.  Feedback in complex systems ids difficult to accurately obtain due to limited information and a lack of cause and effect relationships.

Understanding how well intentioned intelligent people can create an outcome that no one expected and no one wants is a profound life lesson.

Orderly processes in creating human judgments and intuition lead people to wrong decisions when faced with complex and highly interacting systems.

Our innate desire to grasp cause and effect leads us to understand the system at the wrong level, resulting in predictable mistakes.  Complex systems can work well with dumb agents as well as fail spectacularly when well meaning intelligent people attempt to manage the system.

TT6:  Evidence of Circumstance.

People try to cram the lessons or experiences from one situation into a different situation.  This strategy often fails because decisions that work in one context fail miserably in another.  The right answer to most questions is ‘it depends.’

People ground their choices in theory:  A belief that a certain action will lead to a satisfactory outcome.  You can consider a decision as theory: an explanation of cause and effect.

Theory building occurs in three stages:

1.      Observation:  Carefully measuring a phenomenon and documenting the results.  The goal is to set standards that can be reproduced.

2.     Classification:  Simplify and organize the world into categories to clarify differences among phenomena.  Based predominately on attributes.

3.     Definition:  Describing the relationship between the categories and the outcome.

Theories improve when researchers test predictions against real world data.  They also tend to improve as researchers start classifying using circumstances, not just attributes in the definition state.  Theories advance beyond simple correlations and sharpens to define causes.

First mistake:  Embracing a strategy without fully understanding the conditions under which it succeeds or fails.

Second mistake:  Failure to think properly about competitive circumstances.

Third mistake:  Failure to distinguish between correclation and causality.  Arises when researchers observe a correlation between two variables and assumes one caused the other.

Three conditions to have causality between two variables.

1.      X must occur before y.

2.     Functional relationship between X and Y.  Cause and effect take on at least two or more values (does person smoke, does persona have lung cancer).

3.     For X to cause Y, there cannot be a factor Z that causes on X or Y.

Fourth mistake: inflexibility in the face of evidence that change is necessary.

Advice:

1.      Ask if the theory behind your decision making accounts for circumstances.  People tend to take decisions from successful experiences in the past into new situations.  Not good decision making if followed blindley, or if components of the new situation are different from the old.

2.     Watch for the correlation trap.  We innately look for cause and effect in complex systems.  We are not beyond making up or perceiving things wrongly to make cause and effect work.  Runs the risk of observing an event, due to chance, and attributing it to a correlation.

3.     Balance simple rules with changing conditions.

4.     Remember there is no ‘best’ practice.

TT7:  Grand Ah Whooms.

When positive feedback takes over.  Feedback can be positive or negative and many healthy systems have both.  Positive feedback promotes change, negative feedback is resistant and provides stabilization.   Too much of either can leave a system out of balance.

Postive feedback reinforces an initial change in the same direction.

Phase transitions:  When small incremental changes cause large scale effects.  They occur in many complex systems where collective behavior emerges from the interaction of its constituent parts.  These systems have critical point or thresholds when phase transitions occur.

Critical points are very important for proper counterfactual thinking.  As in, ‘what might have been’.  For every phase transition you saw, how many close calls were there.  Large scale outcomes are the result of internal workings, not external shock.  Referred to as invisible invulnerability.

Distributions in many systems don’t stray too much from their averages.  However, distributions are heavily skewed for some complex systems.  The idea of an average holds little meaning.

Black swan:  Extreme outcomes in distributions.  An outlier event that has a consequential impact and that humans try to explain after the fact.  People understand black swan events, but not what propagates them.

Positive feedback leads to outcomes that are outliers.  And critical points help explain our perpetual surprise at black swan events as we have a hard time understanding small perturbations that lead to such large impacts.

What are behind critical points:  Answer can be found in the wisdom of crowds.  Crowds ten to make accurate predictions when diversity, aggregation and incentives are in place.

Diversity:  People having different ideas.

Aggregation:  Can bring groups information together.

Incentive:  Rewards for being right.

Diversity is the most likely to fail when humans are involved.  It is important to note that crowds don’t go smart to dumb gradually.  As diversity lessens, there is no impact.  Bat at some point, further reduction of diversity will hit a critical point and cause quantitative change within a system.

During a run up to a crash, population diversity falls.  Agents begin using similar trading styles and the common good is reinforced.  Population becomes brittle and a reduction in demand can have a huge impact on the market.

First Mistake:  Induction:  How you should logically go from specific observations to general conclusions.  Popper argued that seeing lots of white swans doesn’t prove all swans are white.  But seeing one black swan does prove that not all swans are white.  Popper’s point is that we’re better off at falsification than verification.  However, we’re not naturally inclined to falsify something.

When we think about something a certain way, it’s hard for us to think of it in a different way.  Strong tendency to stick to established perspective and slow to consider alternatives.

Good outcomes provide us with confirming evidence that our strategy is good.  The illusion emphasizes the overconfidence bias and we are surprised when it fails.

Second mistake:  Reductive bias.  The tendency to oversimplify complex systems.  People tend to think of systems as simple and linear.  Even though systems tend to be complex and non-linear.

Mistake:  Belief in prediction.  Ours is the only world we know.  Generally there is no way to test the inevitability of the outcomes we see.

Social influence plays a huge part in success or failure.  There is no guarantee of quality and commercial success.  Social influences tends to exacerbate product successes and failures. 

Inequality of outcomes is substantially greater in the social worlds than in the independent worlds.  Luck of the draw is defined early on.  Our world represents one of many possible worlds, and small changes in initial conditions lead to a big difference in outcomes.  Social influences can be the engine for positive (change oriented) feedback.

Advice:

1.      Study the distribution of outcomes for the system you are dealing with.  If evaluations include extremes, then black swans become grey.  The key is to prepare for extremes.  We’re not necessarily scorched by black swans, but scorched by not preparing for gray swans.

2.     Look for phase transitions moments.  Reduction in diversity increases chances of  a system failure.  Also referred to as coordinated behavior.

3.     Beware of forecasters.  Accuracy of forecasts is dismal in systems with phase transitions.  Even by the experts.

4.     Mitigate the downside, capture the upside.  Betting too much in a system with extreme outcomes lead to ruin.  Extremes occur on the positive and negative sides.  Consequences are more important than probabilities.

We tend to over simplify complex systems and thus become mistake prone.  If you see such a system, slow your thinking down.  When navigating through potentially many black swans, the key is to live to see another day.

TT8:  Sorting Luck from Skill

We have difficulty sorting luck from skill.  The result is that we often make predictable and natural mistakes.

Reversion to the mean suggests things become more average over time, while a stable distribution implies things don’t change.  Understanding that change and stability go together  is the key to understanding reversion to the mean.

Any system that requires skill and luck will revert to the mean over time.  Kahneman suggests the following equations:

Success  = Some talent + luck

Extreme Success = Some talent + a lot of luck

When you ignore reversion to the mean, you make the following mistakes;

1.      Your thinking that you are special

2.     Misinterpretation of what the data says.  What looks like cause and effect is really a reversion to the mean.

3.     Making specific inferences from general observations.

Reversion to the main also applies to economic indicators.

We tend to observe financially successful companies, attach attributes (leadership, visionary strategy, financial controls) to that success, and recommend that others embrace those traits to achieve their own success.

In other words we identify traits to success without understanding the role of luck.  When things start to revert back, we become critical of them.

The effect is emphasized by the media.  The presses tendency to focus on extreme performance is so predictable that it has become a reliable counter indicator.

Example:  Forbes, Business Week and fortune feature companies in bullish articles and bearish articles.  Those in the bullish category had for the previous two years generated 40+% returns, the opposite to companies featured in bearish articles.  In the next two years, bearish companies outperformed bullish companies three to one.

Advice:

1.      Evaluate the mix of skill and luck in the system you are analyzing.  To evaluate, ask yourself if you can loose on purpose.  If you can’t then its mostly based on luck.  If you can than skill is involved.  When we win we tend to attribute it to skill and when we lose, to bad luck.

2.     Carefully consider the sample size.  We tend to extrapolate unfounded conclusions from small sample sizes.  The more luck contributes to outcomes, the larger the sample size you will need to distinguish luck from skill.  Streaks ae an indicator of skill.  Favorable impressions lead to future or more interactions.  Although the next experience will revert back to the mean, more information will be gathered by the recipient.  Unfavorable impressions lead to people not interacting or returning.  This ensures additional information will not be gathered.

3.     Watch for change within the system or of the system.

4.     Watch for the halo effect. 

An appreciation of the relative contributions of skill and luck will allow you to think clearly about reversion to the mean.  When outcomes are good, prepare for the not so good and vice versa.

End Notes:

The value of this information pertains to higher risk decisions.  Decisions with low impact normally are straight forward and easy to determine.  We do it numerous times every day.

You should:

1.      Prepare:  learn about potential mistakes and

2.     Recognize:  identify them in context and

3.     Apply: sharpen your ultimate decisions.