The Black Swan, by Taleb
A Review and Some Notes
The Black Swan by Nassim Nicholas Taleb is a great read. It is a must read for anyone that does forecasting -- whether it be academically, in business, in government, or in any other setting. I also recommend it for any business major or anyone who likes to dabble in the stock market. Some of the ideas in The Black Swan are seemingly obvious, but Taleb explains them in detail and gives the reader a new perspective from which to view the world.
One of the main themes on which Taleb touches in The Black Swan, is "Platonification", or our tendency as humans to simplify. We like to explain history by using general themes when, in fact, history is very complex and cannot be simplified into one theme and a few pages. We not only simplify history, but we also generalize problems and make simplifying forecasts. Our tendency to Platonify also leads us to depend on averages and to believe that the future will be average. We then miss the "Black Swans".
A "Black Swan", according to Taleb, is an extraordinary, unpredictable event. The term comes from the discovery of black swans. Until the 18th century, Europeans had only experienced white swans since all swans in Europe are white. When they ventured to Australia, they were surprised to find black swans. Why would any European have predicted a black swan when all experience pointed to the existence of only white swans? Well, we will experience many Black Swans in our lives. They include stock market crashes, the rise of personal computers, the rise of the internet, World War I, 9/11, etc. All these events were unforeseen.
The interesting thing about Black Swans, as Taleb points out, is that they are the drivers of history and of our lives. History does not putter along at a steady pace, it jumps and leaps between huge events.
But what value are Black Swans to us if they cannot be predicted? This is what makes Taleb's book really great. He demonstrates that we rely on faulty forecasting and that we are blind to the possiblity of the next Black Swan. We like to think that we are good forecasters, but if we fail to forecast the most important events, the Black Swans, then what good is our forecasting? A stock market crash can wipe out fortunes and companies, even companies that have existed for decades. Interestingly, those who miss forecasts always have explanations, such as they didn't have all the info, the event was unforeseeable, or they were still pretty close. Again, what good is forecasting if it's subject to so much error? Taleb also shows that experts aren't really better at forecasting, but since they're "experts", they think they're better and it leads to worse outcomes.
Taleb's commentary on forecasting particularly hit home for me. Right out of college I worked for a company that forecasted consumer packaged goods. One of our forecasts was for a new type of tanning lotion. Our first year forecast was around half of the actual sales (if I remember correctly). This lead to the company having production shortages, which they were not happy about. Our excuse was that the outcome was unforeseeable and extraordinary (a Black Swan). It was reviewed by Oprah, became a trendy product, and consumers loved it more than anything we had forecasted before. So why were we getting payed to forecast? I suppose we were close on average.
I have also worked for a semiconductor company, which is in an industry known for its volatility. We received forecasts from a couple very popular market research companies. One day I decided to see how accurate they were. I found that over a five year period, their forecast for average selling prices (which basically determines revenue and profitability -- ASP's are extremely important to semiconductor companies) was 20% off a year before the forecasted date ON AVERAGE. A 20% error for a company's earnings is huge. Yet there we were, buying and believing these forecasts. Of course, we couldn't do any better and we needed something to use for planning, so I don't think we were wrong in using the numbers. (The market research firms also provided forecasts 4-5 years out. Ha!)
In The Black Swan, Taleb does give some advice to those who must rely on forecasts to some degree. Among other things, they include that variability matters, forecasts degrade as time increases, and forecasted variables are often random. (You can see The Black Swan notes below for more info.)
Among all this, one of the most interesting points that Taleb makes in The Black Swan, is that Modern Portfolio Theory is flawed. Modern Portfolio Theory uses an asset's standard deviation as its measure of risk. The problem is that outliers in stock returns make all the difference. Taleb points out that the crash of 1987 was a 20 standard deviation event. The probability of that is something like 5.51*10^(-89), or 0.0...[about 90 zeros]...0551. So Modern Portfolio Theory did not account for the 1987 crash, or any other crash. Thus all the finance majors and MBA's are taught a system that doesn't even really hold up.
Taleb's system, on the other hand, is to invest about 80-85% in low-yielding T-bills, and then the rest in very high risk-reward assets, such as way-out-of-the-money options. Most years you will lose a few percentage points, but when the Black Swan hits, you will make huge gains that will offset your losses (while everyone else suffers huge losses).
While Taleb's ideas may seem theoretical, consider the following: his hedge fund, Universa, made 65-115% during October 2008. Also, consider the following paragraph that he included in The Black Swan, which was published in 2007:
Globalization creates interlocking fragility, while reducing volatility and giving the appearance of stability. In other words it creates devastating Black Swans. We have never lived before under the threat of a global collapse. Financial Institutions have been merging into a smaller number of very large banks. Almost all banks are interrelated. So the financial ecology is swelling into gigantic, incestuous, bureaucratic banks – when one fails, they all fall. The increased concentration among banks seems to have the effect of making financial crisis less likely, but when they happen they are more global in scale and hit us very hard. We have moved from a diversified ecology of small banks, with varied lending policies, to a more homogeneous framework of firms that all resemble one another. True, we now have fewer failures, but when they occur ....I shiver at the thought.
Wow. When I read that in the summer of 2009, I could hardly believe it.
The government-sponsored institution Fannie Mae, when I look at its risks, seems to be sitting on a barrel of dynamite, vulnerable to the slightest hiccup. But not to worry: their large staff of scientists deem these events "unlikely".
The Black Swan did have some drawbacks. Sometimes Taleb came across a bit cocky and almost preachy. He is obviously very educated and very well-read, and he knows it and uses it. Sometimes his philosophical droning can get a bit boring and tedious.
Nevertheless, his points are fantasitic and his themes are enlightening. If you haven't read The Black Swan yet, now is the time.
Notes from The Black Swan
After reading the first 50 pages or so, I decided that I really needed to remember the main points in The Black Swan, so I compiled notes which are listed below. I have included page numbers, but keep in mind that I read a hardcover version (and I'm not sure when it was printed or what edition it was, sorry). These can serve as abreviated "Cliff's Notes" for anyone that doesn't want to read the whole book. I'm sorry if they are unclear, but they are notes. If you've read the book I think they'll be pretty clear.
- A "Black Swan" is an extraordinary, unpredictable event
- The name comes from black swans, which didn't exist in Europe, but when Europeans went to Australia, they realized that black swans did exist and that not all swans are white
- History is opaque (p8)
- We understand less than we realize -- the world is more complicated than we believe
- RETROSPECTIVE DISTORTION: history seems clearer and more organized in books than in empirical reality
- "Platonification" - when people, especially the learned and authoritative, create categories and oversimplify (comes from Plato)
- History doesn't crawl -- it jumps between major, unexpected events (p11)
- Scalable professions allow you to magnify income without much additional input (p27)
- Examples include things like authors, musicians, and artists
- However, this only works if you're lucky to some degree
- Some strike big, but many get nothing
- When Black Swans are possible, averages don't work because one observation can change everything (p34)
- Consider the life of a turkey: everything is fine, the past, present, and future, until on Thanksgiving everything changes abrubtly and without precedent (p41)
- CONFIRMATION BIAS: we focus on preselected segments of the seen and generalize it to the unseen (p50)
- Books that print statistics about rich peoples' lives do not necessarily show how to become rich -- how do we know that all people don't have similar characteristics?
- No evidence of cancer is NOT equal to evidence of no cancer
- Trying to disprove a theory is better than finding additionaly support for it
- eg, Seeing another white swan proves nothing
- We like to use stories to summarize and simplify; stories appeal to humans (p63)
- Narrative Fallacy: information is costly to obtain and store, so we use narrations (p68)
- We search for causes without enough evidence
- eg, On the morning of Saddam Hussein's capture, Bloomberg printed an article stating that treasury prices rose because of his capture; a couple hours later, after treasury prices had fallen, another article was printed by Bloomberg blaming the fall on Hussein's capture (p74)
- Sensational stories have a greater impact and skew reasoning (p76)
- Many jobs have positive, lumpy returns rather than steady returns (p86)
- These types of jobs require people who can deal with long periods of small returns (consider, artists, entrepreneurs, inventors, pharmaceutical researchers, etc.)
- Most people would prefer $100k for 10 years rather than $1M every 10th year (p91)
- Unfortunately, not everyone has that choice
- SILENT EVIDENCE conceals randomness (p102)
- Losers don't make best-sellers, only the winners do
- This makes it appear that winners have certain characteristics, when they may just be lucky
- We try to attribute certain characteristics to millionaires by only studying millionaires, but we fail to look at the "cemetary of" non-millionaires -- maybe they have the same attributes? (p105)
- Beginners luck happens because all gamblers were "lucky" when they started, otherwise they would have lost $ and stopped gambling; they then become only average and remember their beginning luck (p109)
- "Swimmer's body" is not from swimming -- it's genetic and people with those genetics become the good swimmers (p110)
- ANTHROPIC PRINCIPLE/ANTHROPIC BIAS -- Our presence in the sample makes the odds irrelevant (p117)
- The odds of life on Earth are tiny, but yet we are hear to ask the question
- A millionaire will tell us that the odds of his success are tiny, so it must be talent, but you can't look at the odds from the end point or after the fact (p118)
- The risks at a casino are not the risks they can compute -- like someone winning big. Rather, it's stuff like an employee stealing or breaking a law and other risks that are impossible to see (p129)
- People can't predict even if they think they can (p139)
- Researchers ask people to give a 98% conidence interval of random facts; rather than a 2% error, it's usually between 15%-30%
- The error is greater for more unusual items (like Black Swans)
- More information can make predictions worse, even though people get more confident (p143)
- This is particularly true when information comes in small increments -- people just use the new info to support their original hypothesis -- they're not open to new ideas from new info
- Many "experts", like economists, are no better at predicting than regular people (p151)
- However, since they're experts, they think they're more accurate
- When wrong, they generally have three explanations (and so do all of that make incorrect predictions)
- Tell yourself you were playing a different game -- "I didn't have all the information"
- Outlier. There was something unpredictable, a Black Swan; the chances of this occurance were very small
- "I was almost right. Had things gone a bit different, I would have been right."
- We always underestimate the time it will take to do projects (p156)
- Everything that affects a project's time frame always seems to lengthen it -- never shorten
- Forecasting fallacies (key ideas that people ignore)
- Variability matters -- don't ignore it; policy should not be set on an average (eg, on average our company needs X cash to survive a downturn, so that's what we'll hold in reserve)
- Forecasts degrade as time increases; 5-10 year forecasts are often worthless (this may not matter to a speculator, but it does to a retiree)
- Forecasted variables are often very random
- Tiny differences in forecast variables can lead to huge differences in the outcome; often the parameters and variables themselves are very difficult to predict (p176)
- If one knows the exact state of everything in a system, he can make accurate predictions (provided he can do the virtually impossible calculation) (p186)
- But this only applies to inanimate objects; if free will exists, then humans are unpredictable and knowing the exact state won't necessarily solve everything
- Economists often make simplifying assumptions like rational behavior, but these often don't hold
- Equivalent data can result in very different predictions (p188)
- Surviving until tomorrow can mean (1) that you are immortal or (2) that you're one day closer to death
- Someone being a good friend can mean (1) that they care for your welfare or (2) that they plan on conning you
- BLACK SWAN ASSYMETRY: You can be confident about what is wrong, but not about what you believe is right (p192)
- Since we are human, we must rely on forecasts, predictions, and causes a lot of the time; the key is to avoid relying on forecasts for the big, important things (p203)
- Make a distinction between positive and negative contingencies -- increase exposure to big, positive black swan events, and decrease exposure to negative ones
- eg, Barbell strategy: 80-85% of assets held in t-bills, 15% in high reward securities (options, private equity, venture capital, etc)
- Don't look for the precise and the local -- don't be narrow minded; do not try to predict Black Swans -- it makes you more vulnerable to the ones you miss
- Seize any opportunity, or anything that looks like opportunity; opportunities are rare -- watch out of them and when they come, drop everything to go with them
- Beware of precise plans by governments; governments can't predict, neither can corporations
- Don't waste time trying to fight forecasters, stock analysts, economists, and social scientists
- Randomness often determines the disproportionate winners and disproportionate losers (p216)
- The Matthew effect: a random win early can lead to multiple wins in the future (p217)
- eg, Microsoft, a well-cited academic researcher (once a researcher is cited, others will continue citing), a book getting a first good review
- Things tend to concentrate randomly: size of a city, bacteria population, used vocabulary (p219)
- Luck not only sends winners to the top, it also can displace winners (p221)
- Only 74 o the S&P 500 in 1957 were still there in 1997
- Capitalism gives everyone a chance to be lucky; socialism keeps people from luck
- There is a "long tail" of websites and books with some success -- from the tail will come the new winners (p223)
- "Gaussian bell curve variations face a headwind that makes probabilities drop at a faster and faster rate as you move away from the mean, while 'scalables', or Mandelbrotian variations, do not have such a restriction." (p234)
- The bell curve makes outliers irrelevant, but outliers are often very relevant: market crashes, Bill Gates, etc.
- Example of Mandelbrotian: double the income is half the probability of occuring
- Regression and correlation are misleading -- using subsamples of a sample shows large differences (p239)
- Gaussian (bell curve) is wrong when the data involves magnitudes where a single point can disrupt everything else -- like yearly profit (a company's history of profits can be wiped out in a single year) (p244)
- Fractals are preserved across scales -- they are self similar (eg, no matter how zoomed in or out they appear the same) (p260)
- Example: If you think 96 books sell 250k copies/year, and exponent is around 1.5, then (500k/200k)^(-1.5) = 34 books will sell over 500k copies/year and (1000k/250k)^(-1.5) = 8 books will sell over 1000k books each year (p264)
- The higher the exponent, the less share of the top 1% (p265)
- eg, The 1987 crash was a 20 standard deviation event -- obviously Gaussian doesn't apply (p276)
- Modern Portfolio Theory -- sigma, variance, Sharpe Ratio, R^2, etc. -- is based on Gaussian, which is incorrect as applied to the stock market (p278)
4 May 2010