Friday, March 11, 2016

Nassim Taleb is the Donald Trump of Probability


Nassim Taleb is officially pissed at me.



In deference to Bartlett versus Ritchie, let's get one thing straight first. Mr Taleb, "straw man" is two words, not one.

But at least there is some honesty here. Taleb's annoyance has everything to do with the 13,000 views this video has received on YouTube, despite the obscurity of the topic addressed and its length (almost ten minutes). I now appear on his shortlist of "bad apples" - those who dare to question his authority on probability.



Flattery goes a long way ... though actually I cannot remember two hundred digits of Pi. If I could, I would definitely list that as my "sole attribute" when responding to Quora questions - even on answers to chess questions which I notice Taleb has been reading alongside my finance/math responses (yeah Taleb, that is just a little creepy). In any case, clearly I must be taken out lest my dangerous thoughts infect the precious minds of Taleb's followers.



Now parenthetically, it is remarkable how Taleb really cannot manage a couple of sentences without being disingenuous. Are we to believe that Taleb's concern is for his co-authors, not himself? Really? Perhaps for Pablo Triana, Taleb's understudy who suggested we need a Council of Nicaea comprised of non-technical people to opine on what was good and bad mathematics. Sure, I can see why he might need protection from any responses to that COLOSSALLY FANTASTICALLY STUPID position.

But the real question is how Taleb plans to "deal with me" for having the temerity to answer a question "outside my field of expertise". A question like "What do quants think of Nassim Taleb?" that - oh wait - that would be entirely within my expertise actually.  And actually, Mr Taleb you can refer to me as Dr Cotton from now on, Mr Taleb, and you can have the honestly to remind your readers, Mr Taleb, which cereal box you got your degree from because no, installing a giant oversize cardboard cutout of yourself outside a conference center doesn't actually make you an expert on anything. While we are cutting through the delusions, Mr Taleb, I'll point out that you find my writing repetitive only because you read it over and over again.

Now curious about this list I found myself on, I got in contact with another person in that same club: Eric Falkenstein who has written some pretty funny articles about Taleb (Scott Locklin deserves a spot, incidentally, for this articulate gem alone). Eric revealed to me that Taleb - ever the cheap thug - had contacted his boss directly in an attempt to intimidate him. Well,  I honestly can't wait to see what happens if Taleb tries that one on me. Not sure how it will turn out exactly, but I reckon there will be tears streaming down my boss' face every time he retells the story over beers.
                                     
                                         Trumpability

And now I am wondering why I never noticed before the uncanny similarity in the bullying tactics of Nassim Taleb and those of Donald Trump. Trump insulted his way to the nomination. Taleb insulted his way to a kind of popular probability presidency. Taleb tosses around autism whereas Trump's game is Islam (I think it is pretty clear which is the most offensive, given that one is a medical condition).

Trump tries to make people feel better by bringing down everyone in the public sphere. Sure, most of them are civic minded people working hard to make the country better and, often, forgoing more lucrative options in the private sector. They deal with incredible frustration on a daily basis and stomach churning compromise as they inch their way to what they earnestly (for the most part) believe in. That doesn't matter to Trump. He wants to make people feel good about the fact that there are simple solutions perceived in an instant by amateurs that have somehow escaped the career professionals for years.

Taleb played exactly the same game. By standing up the straw man of naive frequentism and then knocking it down (in a rather longwinded fashion) he invited the lay audience to make fun of "autistic savants" working in banks in quantitative roles. Having never had anything to do with Fixed Income he offered zero perspective on the actual causes of the crisis and zero actionable intelligence thereafter - unless you think that "models bad" or "debt bad" is advice. Personally, I find it right up there with "China bad" or "Mexico bad". Hey Taleb, should we get rid of "waste and abuse" too?

Then there is the absence of a platform. Trump's tremendous confidence is rooted in the knowledge that his supporters don't give a damn that he has virtually no actual policy. What exists is so counter to basic arithmetic it beggars belief, but that doesn't matter either. Taleb, similarly, offers no actual theory to speak of. His arguments are in no danger of being moulded into actual logic (by himself or others). And worse, he is too lazy or ignorant to point out to readers where the existing literature would step in. On top of that, he actually thinks it is clever to reduce knowledge. Taleb appeared before congress and suggested, entirely seriously, that it would be better to have no research into financial stability because knowledge is a bad thing.

Finally, there is the manipulation of the audience. Trump, we must acknowledge, is an intelligent human being. So is Taleb, though I suspect less so. But both know that they have an audience who is uneducated and easily swayed by emotion and posturing. They can play the strong man with little chance of pushback. They are fundamentally patronizing and dishonest. Taleb addresses audiences who haven't picked up a textbook in thirty to fifty years, if ever, and knows he isn't going to be challenged on logical flaws in his arguments or callous disregard for literature. Readers want someone to help them beat up on all the bad people out there like academics who dare to write papers about economics.

We should be deeply suspicious of those who offer simple answers to complex problems. As Bartlett said to Ritchie, what are the next ten words in your answer, Mr Taleb? And the next ten after that?




Monday, January 28, 2013

Nassim Taleb has educated all 10,000 quants in the world

According to John Fawcett of startup Quantopian, there are roughly 10,000 quants worldwide. (See the Forbes article covering his company, which looks a little like QuantDom meets Kaggle.Com).  The good news is that Nassim Taleb has educated almost all of them:


(The precise count was  9,629 last I checked)

Tuesday, November 20, 2012

A Review of The Word Anti-Fragility


Your blogger found himself under a certain amount of peer pressure to write a review of "Anti-Fragility", the new book by Nassim Taleb. On the other hand, I have already commented briefly on that vast subject and of late have found Taleb to be a little boring.


                                        Nassim Taleb on Anti-Fragility

So here is an xtranormal video instead, coving some old ground admittedly, just to poke at a few of his fans as the release is made. I look forward to the predictable reactions of allegedly unpredictable humans.

Sunday, July 29, 2012

An overlooked contributor to longshot bias?

Bet on the 6/4 favorite at a major race meeting and you are probably losing five to ten cents on your dollar. Bet the same dollar on a 150/1 outsider and you’ll be lucky to keep sixty cents. This declining return on investment as a function of the odds is known as the long-shot effect, and due to the vast number of races that have been run it is one of the least debatable phenomena in all of social science.  The longer the odds, the more negative the expected value of your bet.




In financial markets, it has been suggested by Nassim Taleb, things are very different indeed. One might even be given to thinking that thousands of companies have raised huge amounts of money to finance vast numbers of projects and every time there was a philosophically challenged lender on the other end of the trade. Popular finance suggests that your fund manager placed bets on the red hot favorite: the survival of a company, and exposed you to less likely but potentially disastrous outcomes. She sought higher yields but, as with taking the 16/1-on favourites at the trots, this came at great price: the occasional massive setback more than countering all the gains. Moral hazard exacerbates the problem, we are warned, because fund managers share the upside with investors but not the downside. 

Despite the persuasiveness of Mr Taleb there is no compelling reason to think that the polarity of the longshot bias - should it exist in a broad sense at all - is flipped as we move from the racetrack to finance. Those choosing to adopt an allegedly philosophical lens known as Black Swan theory should take a harder look at bond markets, because if that theory is correct corporate bonds buyers should be amongst the worst offenders. The facts that get in the way include a recent paper by Giesecke, Longstaff, Schaefer and Strebulaev surveying 150 years of corporate bond issuance and 150 years of corporate defaults. It may be surprising to Black Swan theory fans that over the long term buy and hold investors were compensated for twice the risk they were taking, not undercompensated.

What makes this even more remarkable is the fact that their data includes not only the Great Depression but a period that was even worse for fixed income investors. In percentage terms, the railroad crisis of 1873-1875 hit the bond markets with the same force the Black Death hit Europe.

This leaves the reverse longshot theory on shaky ground, to put it mildly, and individuals would be crazy to piss away their retirement funds on pseudo-science of this caliber. As I noted in this post, even if there were a widespread reverse longshot bias in finance it would still not justify a bar-bell portfolio.

                                                           A Subtle Bias?

I am prompted to mention the longshot bias just now for a different reason. While reading a seemingly unrelated, purely statistical paper the other day it occurred to me that the longshot bias is more subtle than I (an erstwhile assistant to a professional gambler in Sydney) previously appreciated. The longshot bias has been dissected many times over the years, most recently by Snowberg and Wolfers, but what I find interesting is a mathematical angle that may have gone overlooked (at least in my somewhat cursory reading of the longshot literature). I therefore draw tge readers attention to two facts:

  (a)  [From the economics literature] In the absence of a track take, parimutuel markets effect convex combinations of the subjective probabilities of participants. 

  (b)  [From the statistical literature] No convex combination of calibrated forecasts is calibrated.

The first observation is quite well known, I believe, because betting proportional to one's best estimate of true probabilities is optimal under surprisingly weak assumptions. And there are other markets, I dare presume, which effect linear pooling of opinions. But the second observation might be less well known. The terminology 'calibrated' must be read as in this paper, by the way, where the linear combination of probabilistic forecasts is contemplated and improvements suggested. Those improvements attempt to overcome the problem of underconfident forecasts: those that assign lower probabilities to odds on favourites and higher (than true) probabilities to longshots. The point: the linear combination of forecasts creates this type of bias even if the individual bettors are themselves subject to no longshot bias at all.

The authors do not contemplate markets where convex combinations of probabilistic forecasts are implicitly combined, but it seems a few simple theorems would follow easily and it might be able to prove underconfidence in a broad number of markets. In particular, it is already apparent from the two obervations listed above that at least one stylized marketplace comprising only individuals whose forecasts are free of statistical bias can nevertheless conspire to create underconfident forecasts in the risk-neutral measure. That is, a longshot bias can exist in the absence of any individual behavioural biases whatsoever.  


 


Sunday, April 1, 2012

Benchmark Solutions: Probably a Black Swan for the Corporate Bond Market

Your blogger was a little taken aback by a wire story that ran last night. I'm not entirely sure the journalist involved took accurate notes and pending further clarifying discussion I am not going to repeat her name here. Still, here is the release for those that might be interested.

While equity markets have long seen technological and quantitative progression, the bond markets have remained essentially the same for one hundred years. That’s why the talk on the street is the new transparency service offered by Benchmark Solutions, one so advanced it threatens to change the very nature of the U.S. corporate bond markets. Benchmark’s product is “light years ahead”, according to one customer, and the firm re-prices every corporate bond every ten seconds. That’s something a bulge bracket firm manages once or twice a day so bond market makers have already incorporated the service into their operations. More and more market participants are looking at the so-called Magenta Line, the time series of Benchmark’s constantly updated estimate of bond and credit default swap prices.

The imminent upheaval has prompted some analysts to ask whether the arrival of the Magenta Line is a Black Swan event. The provisional answer is “very probably, yes”. Black Swan events were characterized by post-probabilist philosopher Nassim Taleb as unanticipated developments that have great impact and are rationalized after the fact. And interestingly, Taleb himself may have failed to predict the surprising, benevolent impact of mathematics on bond markets.

That’s the view of Eric Schmalzbauer whose responsibilities at Benchmark include reference data. Schmalzbauer notes the philosopher’s recent submission to an oversight committee for the newly formed Office of Financial Research, in which Taleb insists that many quantitative methods have worked only on computers but not in the real world. That has been true in the past, the philosopher’s argument runs, and therefore, in the absence of a compelling reason, it should be true in the future as well.

Applied mathematician AJ Lindeman notes, in passing, that computers used by Benchmark are very much in the real world and located somewhere in upstate New York. “Our main cluster is not far from here”, said Dr Lindeman, responding to questions from his weekend retreat, “and the mathematics is sound”. When asked to predict whether Benchmark was a Black Swan he suggested it might fill a void. “Yes, I’d like to think this is a Black Swan example because the Black Swan Theory Wikipedia page currently contains no examples at all. I think there used to be an examples section”, continued Lindeman, “but they only had an Earthquake in Japan. People took out the whole section after realizing that was a little bit silly”.

When pressed for his favorite Black Swan event, colleague Joe Leo was cautious, restricting his attention to those nominated by Taleb himself. “I’m still not sure that Harry Potter or Viagra are good examples, though perhaps Harry Potter taking Viagra would count”. After pausing to think for a few seconds, the developer qualified this remark “… so long as nobody thought of it before it happened”. Leo then swore something under his breath, scratched his chin and walked away muttering.

Tim Grant hints that criticism of financial mathematics may have created a special kind of platonic blind spot for pundits, though it isn’t clear if it is big P Platonic or the small p platonic variety. “There is a lot of silent evidence that mathematics works extremely well, but we don’t see it because is silent”, says the new head of sales. Grant suggests a term of art for inductive errors related to applied mathematical progress, a specific philosophical trap he calls mathematical induction.

CEO Jim Toffey is inclined to agree. “You look at the bond markets in 1957. Then 1958. 1959. All the way to the present there’s been virtually no change. But extrapolate that forward even a year or two and you might look like a Turkey. Unless you are British. Then you look like Bertrand Russell’s chicken.”

Dr Stan Raatz, a board member for Benchmark Solutions, suggests that inventions are not always widely appreciated before they exist. Raatz, who has contributed to the philosophical foundations of Benchmark, suggests that no one could have predicted the success of the company, much less funded it or worked for it at an early stage. “People have been harsh on financial epistemology”, notes Raatz who is at pains to defend the Black Swan Theory, “and think that when you take the piss out of epistemology you are left with etymology. Well that’s just not literally true.” 

Stafford Lowe, head of human resources management, believes in a well-rounded approach.  “Like my Prime Minister David Cameron, we’ve found applied mathematics meta-advice and ineffable finance to be a sound complement to precise, logical thinking and the terse, tireless grammar of mathematics.” Benchmark employs verbal meta-advisers to route non-linear and linear problems to the correct teams, in keeping with The Black Swan’s thesis about the distinction between Extremistan and Mediocristan. In Extremistan fractals must be used, notes Lowe, but in Mediocristan more traditional, commodity methods are sufficient.

“The cost savings are enormous”, observes CTO Jim Driscoll, who keeps a copy of The Black Swan on his desk at all times. “Work in Extremistan is expensive because even numbers can’t really be employed - they are thinly veiled linear transformations unsuitable for a non-linear world.” He hinted that the company may be offering new, fifteen year apprenticeships in empirical skepticism to meet a growing demand. “Elephants have long memories”, noted Driscoll, somewhat mysteriously.

Whatever the methods employed, the impact of Benchmark Solutions is now apparent, with industry analysts Celent declaring the arrival of a new, pre-trade transparency era for bonds. But is this really a Black Swan? “Time will not tell”, suggests co-founder Peter Cotton, “because the characterization of Black Swan events becomes cloudier over time. The notion is profoundly unscientific.” Cotton added that he thought that might have come out the wrong way and sound pejorative, due to inadequate media training. “All I’m saying is, I’d Iike to focus on the upside and not worry about the probability of the event itself. Apparently you can build an entire theory of decision making around that notion and it involves going to more parties. Definitely count me in.”

Friday, March 9, 2012

Naked Idiocy: A Review of Econned by Yves Smith



Readers may have noticed”, writes Yves Smith, “that I offer no remedies for the economics discipline”. There are a few other things I noticed reading Econned, the new book from the creator of the successful blog Naked Capitalism. But let the record state that I count myself amongst the hundreds of thousands of subscribers to Naked Capitalism, and appreciate Smith’s extraordinary ability to bring an entertaining news summary to my inbox every day.

Indeed Smith is at her strongest accumulating nuggets of juicy information and in Econned, the coverage of Magnetar's structured finance escapades provides a good example. A sufficiently hungry reader should have no problem picking other morsels from the broth of assertions like “[CDS] … have almost no legitimate uses”  , “criticism from within the profession has fallen on deaf ears” and “All the models in the financial economics edifice assume a normal distribution”. That's why I'm sending you a copy of Econned. And I’m going to help you through the first hundred pages just in case you need a nudge.

Let’s build up some momentum by noting that Nouriel Roubini and Lord Robert Skidelsky think highly of this work. Skidelsky, in particular, is one of Keynes more noted biographers and perhaps like Smith, felt the nuances of the master were lost on some of the math jocks who tried to surf his wake. At one point Smith wonders aloud how this came to be, and why “economists rely on interpretation rather than source”, especially when the golden source was Keynes himself.

Yet on the topic of mathematical models it rapidly becomes clear that Smith has relied on interpretation rather than source, where by source I refer to papers and textbooks. As a consequence, she falls into a tedious rendition of an all too familiar tune: Mandelbrot could have saved the world from Gauss. Had Smith relied on the source, rather than the interpretation, we might have been spared from the same old blunders: like conflating Gaussian with least squares or normal cumulants with value at risk. I was hoping for more of a death-by-a-thousand-cuts than being beaten unconscious with a ludicrously blunt instrument. No such luck.

Still, having developed a thick skin for this sort of thing your blogger was coasting along and wanting rather badly to believe that this salvo from outside the profession, as Smith refers to her own work, might be a notch above what we’ve seen. If you’d like to take the same upbeat approach please don’t let me stop you. Later in the book Smith even appears to catch a whiff. She notes that “it is all too easy for the top level of the firm to become hostage to the needs and demands of profit centers”, which is undoubtedly true. Sadly though the loop never closes as least as far as mathematics pushing is concerned. Who were these math producing profit centers, I wonder, and could they spell Normal Copula?

        Dumb ...

Thus it becomes a little difficult to take Smith seriously on journalistic grounds. But that is merely the beginning. When she wanders into allegedly logical arguments things get very scary indeed:
Note that there is already some circularity in the[Markowitz’] logic. To create the prized “efficient portfolio” you need to know the future!
Um, no Smith, you don't. The point is very much that you don't know the future. If you knew the future you'd pick one stock, obviously. Now surely that would have occurred to the author of this fine document? Oh wait...
But the reason for diversifying and creating a portfolio in the first place was to diversify risk because the investor does not know what will happen and thus is not willing to put all his money in one instrument. But if you did know future prices, you would simply buy the security that would go up the most over your chosen time horizon.
Amazing! I'm constantly stunned by people who attempt reductio ad absurdum arguments without noticing the meta-reductio ad absurdum: if you argument is so totally facile and derives a contradiction in something that has been carefully considered by others for a long time, then perhaps you didn't understand the premise. 

We can certainly agree with Smith on one thing. This level of logic places the author "outside the profession" which is her desired posture and not likely to be disputed any time soon. (Btw I read this passage several times just to makes sure I wasn’t suffering from lack of oxygen on my Jet Blue flight and had misinterpreted Smith.  I hadn't. Out of mercy the copy of the book I'm sending to you is missing this section. I ripped it from the spine out of mercy).

By positing arguments this stupid the author saves herself from my charge ironically: that her theoretical arguments rely on third hand accounts originating from people who never worked anywhere near a source (those interpreters include semi-famous semi-quants, though these authorities never worked in the relevant asset class, never mind the business units at the center of the mess).

So no, Smith is not entirely guilty of complete reliance on the distant speculations of others. That’s because nobody else firing salvos at models, critiquing financial economics, or purporting to say anything remotely intelligible about anything remotely related to probability could possibly have come up with anything as lame as Smith's arguments which are, surely, her own.

    ... and dumber

Here's what I mean:
But if prediction markets really were all that good at seeing the future, the horse with the best odds would always win the race.
Oh my goodness! In general it's not so easy for devoutly non-mathematical people to stay on that high wire when writing about probability. But even accounting for that, this fall is truly spectacular. Is the author suggesting that an even money favorite should win all the time rather than fifty percent of the time, roughly? Is she suggesting that if that doesn't happen then we can pooh pooh prediction markets?

I can't even delve into the multiple levels of confusion about the basics of probability and markets that must have been swimming in the author's mind in order to pen something like that. Is the bar for prediction markets to be raised so high that if they don't always predict the result we should ignore them and turn to tea leaves instead? But then wouldn't the favorite's odds always be 1000:1 on? Has Smith contemplated the possibility that there might be genuine chance involved and that prediction markets could do a bang up job and still leave the favourite at 6:1?

Addendum: Smith will be happy with the result of the 2016 Kentucky Derby: 



The market picket the first four horses in order. Happy now Smith? If the market had no ability to aggregate information at all this would happen once every 116,280 Kentucky Derby's. 

Unfortunately for you your copy of the book still has page 87 in it, where Smith's excruciatingly idiotic statement can be found. I’ve laminated the section and added some extra stitches so you can’t rip it out either. Econned will open to that page if dropped accidentally and there, on almost the first line, you will have no choice but to read this over and over again:
But if prediction markets really were all that good at seeing the future, the horse with the best odds would always win the race.
See? Just as painful the second time. And even more embarrassingly, the author, completely oblivious, tries to play this hand like a nut flush:
How often does that happen?
because hey, there is nothing quite like a rhetorical question with an answer too obvious to state. Maybe the favorite wins about as often as the market says it will Smith! Did you contemplate that possibility. Oh God have mercy. I don’t know about you, dear reader, but the only thing that saves me from utter despair when reading this sort of thing is the image of Smith and her apparent hero Taleb sitting down to compare notes on the longshot bias at the racetrack. Let’s hope they don’t rely too heavily on interpretation rather than source.