Taleb is the man who came up with the Black Swan theory, the subject of his latest book, and no doubt lots of misinterpretation by broadsheet columnists (cf the Tipping Point, long tail, chaos theory, etc) The Black Swan, which he introduces in this book, is a highly unlikely event that cannot be dismissed as too improbable - even if you were to count thousands of white swans and no black ones, you cannot conclude that black swans don't exist. It's tied up with the Popperian precept about the falsifiability of a theory, and also with common notions of probability.
Most of this book is about probability - how important it is, how people, even very smart people in jobs that are all about probability, misjudge it, and about how biases in the perception of evidence lead people to wrong conclusions.
Taleb has been a Wall Street trader, and has worked in derivatives, but now is an academic, and came to fame on the crash of LTCM, when he was able to explain the downfall of the company by his Black Swan theory - the Nobel prizewinning founders of LTCM had dismissed as near impossible the eventuality that actually brought them down, which showed a fundamental misunderstanding of the cost of the risk involved. In these sorts of probabilistic events, says Taleb, you can't afford to ignore the upper and lower ends of the distribution curve, because it's not the frequency of the event that's critical, but the magnitude of it. This means that if the extremely rare event is potentially catastrophic, it must be taken account of.
Taleb quite deliberately puts as few numbers in the book as possible - he didn't want it to be a text book, and wanted just a general discussion of his principles. However, this means much of the book is frustratingly vague, not helped by his unstructured, chatty style. There are a lot of decent points made, helped by his voluminous reading (which he never fails to remind the reader of), but he claimed he didn't want to put a quote in that wasn't on the top of his head, so that it didn't become a dry, library-created tome.
Some of the most interesting points are about the biases that people have, unwittingly, when analysing data, and these can apply across almost all general activities. One of the most common is survivor bias. This is an assumption that those items of data that are visible are the whole sample, and drawing conclusions from that, while ignoring items that have failed from the total sample. So, for example, there are books written on millionaires and successful businessmen that attempt to find common traits, and therefore work out what qualities might be required to achieve. So it turns out that successful entrepreneurs are hard-working and risk-taking. Taleb points out that so are most unsuccessful entrepreneurs (some of them may also be lazy and cautious) These are necessary, but not sufficient qualities, but because the sample only inlcuded the millionaires, there was a survivor bias.
Another example is of quotes from reviews that are on the backs of books, or on theatre billboards. They are intended to convince a potential buyer that the review represents a consensus opinion of critics, whereas of course they have been specially selected precisely because of their favourable attitude.
A further example of survivor bias, and a very literal one, that I have long considered, is of holocaust literature. It's natural, when reading the works of Primo Levi, or similar writers, to be in awe of the fortune that he survived, through all the selections and illnesses and transports and marches. What was the chance of his survival? 100%, as it happens - we are only reading about him because he survived to tell us. If we'd chosen to follow his life back in 1938, the chance of our picking a man who would go to Auschwitz and survive would of course be very very small, and it would be highly unlikely that we'd have any story to tell at all - only the survivors tell their tales.
Most of Taleb's discussion relates to traders - that is his background, and that is what he is trying to explain. Much of it attributes the success of traders to pure luck, although he isn't very convincing as to why. In a book that is about randomness, he doesn't explain why the market is random, he assumes that it is, and draws from that the conclusions that the outcomes we see could be generated by chance. In this he is bolstered by his prejudice against unthinking corporate types, MBAs (although he has one), CEOs, and everyone who spends their time in a suit and not reading books. The book is frequently condescending, and seeks, broadly, to show that the rich successes of Wall Street have no inherent skill.
His thesis is that, within the game as it is, there are bound to be winners. If you set up random games of Russian roulette, there would be survivors at the end, who would be the winners. Because of hindsight bias, the survivors would then attribute their success to some skill or technique. This is what happens on trading floors all the time. People take positions, the market moves in their favour, they win, and confidently assert that they were the best at predicting the market move. They're helped in this by the outrageous rewards given for success, which confirm this belief. Of course, those who take such positions that pay off at the extreme ends are most vulnerable to movements in other directions, and they rarely have protection against those movements - the people who do have lowered their exposure, and so aren't the big winners in the first place.
It's all quite evident, and Taleb uses this overview to deride the short-term winners in the market place, and to gloat at those who get caught out, as most of the big players will eventually because, as he claims, they don't understand the real probabilities in the market, believing as they do in their own predictive skills, and so therefore don't adequately protect themselves.
The book is spoiled by these personal prejudices, and by the presence of Taleb throughout, telling the reader just what his habits are, who his friends are, and why his philosophy is the best. It's all far too smug, and not justified by some of the analysis. But it's written in a handily comprehensible way for laymen, and more particularly journalists (who, of course, he derides) to understand, so he becomes the newest guru.
In a fine example of survivor bias, the quote on the front of the book is from Fortune - 'One of the smartest books of all time.' I think not.
[39]
Most of this book is about probability - how important it is, how people, even very smart people in jobs that are all about probability, misjudge it, and about how biases in the perception of evidence lead people to wrong conclusions.
Taleb has been a Wall Street trader, and has worked in derivatives, but now is an academic, and came to fame on the crash of LTCM, when he was able to explain the downfall of the company by his Black Swan theory - the Nobel prizewinning founders of LTCM had dismissed as near impossible the eventuality that actually brought them down, which showed a fundamental misunderstanding of the cost of the risk involved. In these sorts of probabilistic events, says Taleb, you can't afford to ignore the upper and lower ends of the distribution curve, because it's not the frequency of the event that's critical, but the magnitude of it. This means that if the extremely rare event is potentially catastrophic, it must be taken account of.
Taleb quite deliberately puts as few numbers in the book as possible - he didn't want it to be a text book, and wanted just a general discussion of his principles. However, this means much of the book is frustratingly vague, not helped by his unstructured, chatty style. There are a lot of decent points made, helped by his voluminous reading (which he never fails to remind the reader of), but he claimed he didn't want to put a quote in that wasn't on the top of his head, so that it didn't become a dry, library-created tome.
Some of the most interesting points are about the biases that people have, unwittingly, when analysing data, and these can apply across almost all general activities. One of the most common is survivor bias. This is an assumption that those items of data that are visible are the whole sample, and drawing conclusions from that, while ignoring items that have failed from the total sample. So, for example, there are books written on millionaires and successful businessmen that attempt to find common traits, and therefore work out what qualities might be required to achieve. So it turns out that successful entrepreneurs are hard-working and risk-taking. Taleb points out that so are most unsuccessful entrepreneurs (some of them may also be lazy and cautious) These are necessary, but not sufficient qualities, but because the sample only inlcuded the millionaires, there was a survivor bias.
Another example is of quotes from reviews that are on the backs of books, or on theatre billboards. They are intended to convince a potential buyer that the review represents a consensus opinion of critics, whereas of course they have been specially selected precisely because of their favourable attitude.
A further example of survivor bias, and a very literal one, that I have long considered, is of holocaust literature. It's natural, when reading the works of Primo Levi, or similar writers, to be in awe of the fortune that he survived, through all the selections and illnesses and transports and marches. What was the chance of his survival? 100%, as it happens - we are only reading about him because he survived to tell us. If we'd chosen to follow his life back in 1938, the chance of our picking a man who would go to Auschwitz and survive would of course be very very small, and it would be highly unlikely that we'd have any story to tell at all - only the survivors tell their tales.
Most of Taleb's discussion relates to traders - that is his background, and that is what he is trying to explain. Much of it attributes the success of traders to pure luck, although he isn't very convincing as to why. In a book that is about randomness, he doesn't explain why the market is random, he assumes that it is, and draws from that the conclusions that the outcomes we see could be generated by chance. In this he is bolstered by his prejudice against unthinking corporate types, MBAs (although he has one), CEOs, and everyone who spends their time in a suit and not reading books. The book is frequently condescending, and seeks, broadly, to show that the rich successes of Wall Street have no inherent skill.
His thesis is that, within the game as it is, there are bound to be winners. If you set up random games of Russian roulette, there would be survivors at the end, who would be the winners. Because of hindsight bias, the survivors would then attribute their success to some skill or technique. This is what happens on trading floors all the time. People take positions, the market moves in their favour, they win, and confidently assert that they were the best at predicting the market move. They're helped in this by the outrageous rewards given for success, which confirm this belief. Of course, those who take such positions that pay off at the extreme ends are most vulnerable to movements in other directions, and they rarely have protection against those movements - the people who do have lowered their exposure, and so aren't the big winners in the first place.
It's all quite evident, and Taleb uses this overview to deride the short-term winners in the market place, and to gloat at those who get caught out, as most of the big players will eventually because, as he claims, they don't understand the real probabilities in the market, believing as they do in their own predictive skills, and so therefore don't adequately protect themselves.
The book is spoiled by these personal prejudices, and by the presence of Taleb throughout, telling the reader just what his habits are, who his friends are, and why his philosophy is the best. It's all far too smug, and not justified by some of the analysis. But it's written in a handily comprehensible way for laymen, and more particularly journalists (who, of course, he derides) to understand, so he becomes the newest guru.
In a fine example of survivor bias, the quote on the front of the book is from Fortune - 'One of the smartest books of all time.' I think not.
[39]
No comments:
Post a Comment