934 of 1,012 people found the following review helpful
Chapters 15 - 17 are excellent. The remainder is OK.,
This review is from: The Black Swan: The Impact of the Highly Improbable (Incerto) (Hardcover)
Starting with the good (chapters 15 - 17), within chapter 15 Taleb explains where the Bell curve works and where it does not. The Bell curve captures well variables that don't deviate much from the mean. Otherwise, it does not work. Taleb suggests we often fool ourselves in believing that correlation, regression coefficients, or standard deviation convey much information. This is because those coefficients are unstable (and can flip sign when possible) depending on the time selected. This is because the underlying variables are often not stationary enough for these coefficients to be stable.
Chapter 16 is excellent as an introduction to Mandelbrot's fractal geometry as an alternative to Gaussian based investment theory. He supports well that these mathematical tools do capture randomness (of non-stationary variables) far better than the Normal distribution. However, he admits that Mandelbrotian models are not predictive. When looking at the same data set, he and numerous colleagues each came up with different underlying parameters to build fractal-like models. And a small difference in such parameters makes a huge difference in outcome. That's why you will not hear much of fractal geometry within the quantitative financial community. Nevertheless, this is a fascinating subject that deserves further exploration. For this purpose, I recommend Mandelbrot's The Misbehavior of Markets
Within Chapter 17, Taleb further elaborates on the flaws of the Normal distribution. He underlines that half of the return of the stock market over the past 50 years was associated with just 10 days with the greatest daily change. This is an example where stock returns have outliers of such magnitude that using the Normal distribution is not appropriate. Taleb describes the run-ins he experienced with the living legends of modern finance including Myron Scholes and Robert Merton due to his rejection of the Normal distribution assumption that underlies all their models.
The remainder of the book is not nearly as good. Hundreds of pages can be summed up in just stating that we can't predict rare events. Taleb goes way overboard in attributing everything to luck. He thinks MicroSoft beat out Apple just due to luck. Taleb does not consider that MicroSoft open system allowed it to mushroom while Apple locked itself into a proprietary corner. Also, according to Taleb both the rise and fall of Rome were due entirely to luck. But, Rome was best at developing military strategy and transportation networks. However, it eventually suffered from imperial overstretch. Explanations are not always narrative fallacies as Taleb believes. They often beat out ignorance.
When it comes to advice, Taleb's recommendation is interesting. It consists in an asset allocation of 85% risk free investments (T-bill) and the 15% remainder into buying way out of the money Calls and Puts. By doing so, he positions his portfolio to capture the occasional mispriced Black Swans.
This book is somewhat uneven in quality and is not nearly as good as his first book: Fooled by Randomness Revision (Not Available in US): The Hidden Role of Chance in the Markets and Life that has become a classic on Wall Street. Otherwise, it still is an interesting reading.
If you find the subject of this book intriguing, let me suggest a few other books that are more rewarding. Scott Plous'sThe Psychology of Judgment and Decision Making explores the flaws in human judgments far more thoroughly and clearly than Taleb in `The Black Swan.' Perry Mehrling's Fischer Black and the Revolutionary Idea of Finance is also an excellent book. Ideally, that may be who Taleb would have liked to become. Fischer Black was brilliant and as skeptical as Taleb regarding much of the body of economics and finance. Yet, he left a great legacy of elegant models that people still use extensively including the famous Black-Scholes option model. Yes those models were often based on Taleb's dreaded Normal distribution. But, with minor modifications those models have remained valuable. Another recommendation is William Poundstone's Fortune's Formula: The Untold Story of the Scientific Betting System That Beat the Casinos and Wall Street that describes the career of a bright MIT mathematician, Ed Thorp. The latter showed how to successfully deal with uncertainty in gambling and investing. Even Taleb recognized Thorp's unique expertise within `Black Swan.'
Tracked by 7 customers
Sort: Oldest first | Newest first
Showing 1-10 of 93 posts in this discussion
Initial post: Nov 4, 2007 3:40:27 PM PST
Reader Bokkie says:
"The remainder of the book is disappointing. Hundreds of pages can be summed up in just stating that we can't predict rare events."
This is an accurate statement of a book which I quite enjoyed if only for it's intellectual wanderings.
Posted on Jan 1, 2008 4:49:45 PM PST
Gene Parmesan says:
If this reviewer actually read the book, he clearly missed the point.
Yes the book is repetitive. Yes it is longer than necessary (the author readily admits that many chapters are added as support for readers who want additional depth). However, some of the reviewer's points rely on the the same logical mistakes about which Taleb is trying to raise red flags.
In reply to an earlier post on Jan 1, 2008 6:38:30 PM PST
nobrainer, I have read and studied this book in detail. I am not aware that I have "missed the point" as you suggest. I think your criticism must single out my referring to the Black-Scholes option model that Taleb suggests suffers from "Mediocristan" by relying on the normal distribution assumption. This model has been enhanced several times to give the underlying distribution fatter tails to catch some of the "Black Swans" Taleb has in mind. I have mentioned that much more concisely in my review.
Besides this one "point" that I have not missed (as explained above) what other specific points have I missed? I'd be glad to learn from you if there is any learning to be had.
Posted on Jan 4, 2008 11:28:57 AM PST
S. ONEIL says:
In reply to an earlier post on Jan 6, 2008 3:55:24 PM PST
Last edited by the author on Jan 7, 2008 10:39:25 AM PST
S. Oneil, given my extensive background in investment theory and quantitative methods I understand the topics covered by this book very well. Whenever I write a review I share of my analysis to support my opinion. I also often recommend several other books that I feel the readers may enjoy that are either similar to the one reviewed or sometimes superior. Fortunately, over 90% of readers found this review helpful. If you had read any of the recommended books, you would quickly grab their relevance. You isolated "Fortune's Formula" as irrelevant. It is not. Gambling it turns out has much to do with statistics and more specifically probabilities. Your egregious comments come across as an ad hominem attack.
Posted on Jan 7, 2008 3:36:13 PM PST
"Taleb goes way overboard in attributing everything to luck. He thinks MicroSoft beat out Apple just due to luck. Taleb does not consider that MicroSoft open system allowed it to mushroom while Apple locked itself into a proprietary corner."
That is probably the first time I have ever heard Microsoft referred to as an "open" system. While I grant that Microsoft may not have surpassed Apple out of luck alone, it is not because they have an open approach to things - in fact the opposite is true - they mastered the ability to lock data formats (instead of hardware) and drive their competitors out of the market - among a good number of other things. Microsoft and Apple weren't even playing on the same field really, and the comparison is not really a good one.
I think that there is some truth to nobrainer's statement... You missed the point slightly (as have quite a few people here on Amazon and Audible). The point is NOT that narratives can not have some truth to them or that ignorance is better that a narrative - which he never said. In fact, Taleb even uses narrative stories in his book to convey his truth/information because that is how we remember information best. The point is, however, that a narrative story is not the WHOLE truth. The information that is missing can cause you to vastly overestimate the accuracy of the knowledge you have as a result of the narrative you believe - especially when you believe it enough to think that you have all of the information and the correct answer as a result.
"But, Rome was best at developing military strategy and transportation networks. However, it eventually suffered from imperial overstretch."
Take for instance the statement you made above. There are very large volumes written on the rise and fall of the Roman empire. The empire itself lasted for over a thousand years. You summed all this information up in two sentences - a number so small I would feel uncomfortable using it as a summary for my short life much less the span and breadth of an entire empire. And while you may be an astute student of roman history for all I know, the narrative you provided here was not the whole truth. If I were to base my knowledge of Roman history solely on the above statement and then asked to extrapolate answers about Roman society based on that knowledge - I would surely make astoundingly bad judgments.
"When it comes to practical advice, Taleb misfires. His single strategy on how to deal with Black Swans is an asset allocation consisting of 85% risk free investments (T-bill) and the remainder into very high risk investments. I tested this allocation using a small cap index for the high risk component vs a more standard 60%/40% allocation between stock and bond indices respectively. The standard 60%/40% allocation performed way better."
As to your comment above about asset allocation, again you missed the point. Taleb is not trying to help you build the best performing allocation, but instead to help you avoid being destroyed by the black swan event. Obviously, over the short term the standard allocation may perform better than Talebs suggested allocation. However, you are making the mistake he is warning against in assuming that past performance is an indicator of future success. I am not suggesting a different approach, just that you should be aware what you are doing and that what you are doing does not make Taleb incorrect in his assessment of the best risk vs reward strategy.
Additionally, if you intend to respond to my comment here, you may wish to refrain from mentioning the "90%" of readers who found your review helpful as a point in your favor. This again to me indicates that you missed the point entirely. A 90% confirmation rate (i.e. 1000 days of food to the thanksgiving turkey) does not negate the 10% or even .0001% of people who ate your review for thanksgiving dinner.
In reply to an earlier post on Jan 8, 2008 10:55:55 AM PST
Last edited by the author on Jan 26, 2009 1:02:23 PM PST
Threadkiller, I congratulate you on an excellent rebuttal. Taleb could not have replied better. I agree with a lot of what you say such as narratives are not the whole truth; that we invariably miss things; that the past is not an indication of future performance. That history can't be fully explained in a phrase.
However, Taleb errs on attributing many events to randomness when non-controversial causal factors were at play. You attempt to pick apart my examples. It does not mean they are invalid. At the onset, Microsoft made it very easy for third party software developers while Apple did not. That's what commonly referred to "open" vs proprietary. That the Roman Empire suffered from imperial overstretch is well accepted by historians of all stripes. This statement does not intend to explain the whole Roman civilization. But, it captures a relevant factor related to its demise.
In terms of practical application, I disagree with your statement. Taleb does outline his asset allocation strategy in detail. He expresses his opinion on how he invests the monies of his clients. But, I have revised my review and I don't denigrate his investment strategy anymore. It is somewhat complex and may well work just fine.
As you requested I won't refer to the 90% approval rating of this specific review. We are engaged in a collegiate debate where advancing the most robust argument has nothing to do with popularity. I understand that. However, the mentioned approval rating was a relevant rebuttal to another writer that nearly called me names while not advancing any material argument.
In reply to an earlier post on Jan 8, 2008 2:06:42 PM PST
Last edited by the author on Jan 8, 2008 2:16:17 PM PST
"Threadkiller, I congratulate you on an excellent rebuttal. Taleb could not have replied better." << You are very sneaky you waskly wabbit ... but that is something I can appreciate. Joking aside, I respect your comments, as not many people can take criticism and bounce back with praise.
I am glad that you agree with some of what I say, though I think credit goes to Taleb for making the point so refined for me in the first place.
My intent was not to pick apart your examples specifically, but instead to indicate the problem with summary information. The point is not that your statements were invalid, but limited by their very nature and brevity and as a result, prone to error. More specifically, any attempt to reduce large volumes of information reduces the accuracy of that information. The more you reduce it, the more likely that information is to provide you with a wrong conclusion. This is a well understood concept in systems and information theory. The error is not IN the statements you made, but with their brevity and their assumption of accuracy. This is the nature of human communication in general and is one of the reasons we get ourselves into trouble.
I am not the lover of history you appear to be based on your comments and what Amazon has to say about your book purchases and recommendations. However, I know for a fact that there a quite a few historians that sight the decline of morality as one of the reasons for Rome's fall. Your statement only captured one relevant factor you believe to be true. You missed this one and probably thousands more like it - in addition to ones no one will ever know about because they are lost to time. You also filter out a whole bunch of information based on what you perceive to be the truth, and it would be wise to recognize that your filter may be clogged or broken - again not because you are wrong - but because that is how information works. You should always be extremely suspicious of any summary data and any opinions you hold as a result of that summarized data. This is the entire point of "The Black Swan"
You would be absolutely correct in sighting Microsoft's willingness to work with third party developers as the reason for their success - if it were true. In fact, the contrary is actually the case. Remember the lawsuit against Microsoft around 1999? That was all about how Microsoft doesn't play well with others. Remember a program called Netscape, Word Perfect, or how about Lotus 1-2-3? Microsoft didn't necessarily drive them out of business with a superior product (Internet Explorer was and is arguably one of the worst web browsers available) but instead used market leverage and the ability to distribute the software with their operating system as the strategy for "beating" the competition - not by "working" with these third parties. Microsoft uses a zero sum game, you against me, winner take all mentality in everything they do. But again, even my longer summary of what I believe to be the case is not the whole or accurate truth. There were countless (literally) factors that contributed to Microsoft's success. Some we know about, some only Bill knows about, some no one will ever know. Some had a large impact; some a small impact. However, if you don't know ALL of the variables in a system, how can you reasonably point to one variable and say "See that's the major cause". In doing so, you might be missing something larger you were never even aware of and certainly you are likely missing some relevant data. For instance...
I am going to digress a little further from this conversation and attempt to provide you with some information based on my experience and training in the field of software development. I am not sure where you got your information about "open" vs. proprietary software - so please bare with me. For more specific information see what the free software foundation has to say on the matter here: http://www.fsf.org/licensing/essays/free-
"In terms of practical application, I disagree with your statement. Taleb does outline his asset allocation strategy in detail. ....."
First, I did not in any way state that Taleb does not outline his strategy in detail... so I hope that was just some signal noise. Second, instead of trying to point out the error in your approach, let me try a different tactic...
You sighted Taleb's excellent credentials in the end of your review. You also sighted his excellent work when directed at himself. Allow me to suggest that you do the same in the following manner: If you can respect Talebs intellect in some reasonable fashion, is it not possible that he would have also run the "standard" analysis you propose? Is it not possible he sees something different than you do and that difference should give you pause to consider that maybe you are missing something?
As to the specifics of your comment, I am not suggesting that you "ignore" past performance - that would indeed be foolish. Looking back, I should have said "past performance is not a guarantee of future success". I do not play in the investment market so I can not say one way or the other whether or not your assessment of Talebs strategy is correct. My point though, is that you seemed to miss his point and mine concerning any system of information - performance as indicated by past and future models ignores events that the model can not address. Another good book that talks a little about this in terms of human understanding is "Stumbling on Happiness". The author points out that when people talk about tomorrow their prediction is almost always based relative to their current experience. If a loved one has died recently, they can not see happiness in the future. If things are going well, and you ask them about their plans for the future they often say things that indicate they believe there will not be a hurricane, flood, financial depression, etc. that devastates their lives. We as humans and the models we generate are poor predictors of future success - because NO model we generate can encompass events we don't understand or know about and ignores ones we choose not to pay attention to - all of which might be MORE relevant than the things our models DO pay attention to.
In short, your model does not disprove his statement as it can not account for things unknown. On the other hand, things unknown "aka. The Black Swan" can and do invalidate models based on past experience.
Posted on Jan 8, 2008 2:28:44 PM PST
As an afterthought to my last comment:
Neil Gaiman has an interesting parable about this in his book "American Gods." I will not repeat the entire parable here, but the point is this. A model is only as accurate as the information it contains. The more accurate and complete the information, the more accurate the model. The most accurate model, therefore, is the real thing.
As humans are not capable of containing the "real thing" within our minds, we always refer to things as simplified models. As a result, we should always be distrustful and as conservative as possible with these models as they are likely flawed. Pointing to something and saying "I understand the cause" is a fraud to paraphrase Taleb.
In reply to an earlier post on Jan 8, 2008 9:04:49 PM PST
Last edited by the author on Feb 15, 2008 9:42:18 AM PST
Threadkiller, we agree much more than you think. I especially like your concluding paragraph. But, a model needs not perfection to be helpful. A model that is correct 60% of the time adds much information vs a random coin flip. Taleb goes way too far on the model-nihilism bent. And, does not offer any viable constructive alternative. As mentioned in my review, his one alternative is even more flawed than the "Mediocristan" models he criticizes so much.