Financial Implications of “The Black Swan” by Nassim Nicholas Taleb

The Black Swan by Nassim Taleb is one of the best, and one of the most interesting, books I’ve read in awhile. I’ve heard people reference Taleb’s work a few times before, mainly in financial settings, but I never quite understood why, which is what inspired me to read this. In general, Taleb argues that throughout human history, nearly all majorly significant events were considered highly improbable before they occured (think: Fall of Roman Empire, rise of Nazi Germany, Bubonic Plague pandemic, sinking of the Titanic). Essentially, the events which have the greatest impact on us lie totally outside of our field of prediction, no matter how much past data, observations, and intuitions we use in making these predictions. These momentous, unpredictable events are called Black Swans, known professionally as fat tails, because of they are perceived as rare (Taleb actually argues that these events are much more common than we’d like to think).
Taleb helps us visualize the phenomenon of Black Swans through two fictional worlds: Mediocristan and Extremistan. Mediocristan is a province where “particular events don’t contribute much individually — only collectively” (Taleb 32), and things are distributed rather evenly. Extremistan, on the other hand, is a world of extremes, where “inequalities are such that one single observation can disproportionately impact the aggregate, or the total” (33). Taleb claims that we live in Extremistan, as much as we’d like to believe it to be Mediocristan, since single events can have disproportionately large impacts on our societies. Evidence of us living Exremistan is seen through other metrics as well, most notably through wealth divides (top 1% vs 99%).
Image result for population distribution us
This theory of Black Swans being more significant than regular accumulations of events holds true in all fields in Extremistan, especially (and this is most relevant for us) in finance. One fact Taleb mentions really dumbfounded me, which is that: “In the last fifty years, the ten most extreme days in the financial markets represent half the returns” (275). 10 days in fifty years. The alternative, which is not addressed directly by Taleb, is likely also true: In the last fifty years, the ten most extremely negative days in the financial markets represent half the losses. This is so incredible, and at first this made me wonder why people (like me) try so hard to predict what will happen to a stock’s price day-to-day, when they should really be trying to predict the next Black Swan. I then realized that if a Black Swan can be predicted, it is no longer a Black Swan, and the opportunities for profit are no longer as large since more people are expecting it to happen. Also, once a Black Swan is predicted, a new Black Swan emerges outside of this prediction which becomes the next true Black Swan.
This is also a very sobering thought, considering the work I’m currently doing in predictive trading algorithms. I’m sure Taleb would laugh at the idea of me using historical financial data to predict what will happen to a stock tomorrow, next week, or next month. In his book, Taleb talks a lot about the misleading nature of data, using the example of “1001 days of history”, noting “You observe a hypothetical variable for one thousand days. You subsequently derive solely from this past data a few conclusions concerning the properties of the pattern with projections for the thousand, even five thousand, days. On the thousand and first day — boom! A big change takes place that is completely unprepared for by the past.” I enjoy picturing this concept like this: I know I have been alive every day for the past 40 years, so in conclusion, I can use this data to determine I will be alive tomorrow, next month, and the next 40 years after that! The irony here is that the one event which lies outside of my predictions would be the most significant event in my life. Yet, I am not deterred. I do think that patterns exist in market behavior, and because these patterns repeat cyclically, they can be used to make predictions. The mere existence of Black Swans doesn’t negate the existence of patterns, it’s just possible that one day these patterns will break. I think hedging using options is enough to counter potential negative effects, I just need to learn more about how to do this efficiently.
Another key point brought up is the effect of silent evidence, which relates to history being written by the winners and not the losers. “It is so easy to avoid looking at the cemetery while concocting historical theories” (101) says Taleb, which reminds me of the problem of survivor ship bias in quantitative trading. Survivor ship bias occurs when we train models on existing companies and their stocks, while forgetting the companies which no longer exist (bankruptcies, acquisitions). Oftentimes, this omitted data (the silent data) is most important, as we can learn what events leads up to a bankruptcy or sudden stock collapse. This relates to the idea of blind risk usually leading to better short term rewards, whilst completely backfiring later on down the road: “The fools, the Casanovas, and the blind risk takers are often the ones who win in the short term” (117).
Image result for bankrupt companies
A third interesting point which really made me think was Taleb’s take on information, and more importantly, misinformation. He writes: “The more information you give someone, the more hypotheses they will form along the way, and the worse off they will be. They see more random noise and mistake it for information” (144). This really hit close to home, as someone who is naturally paranoid and always thinking that the solution to my problems lies in all the many books, studies, and reports I haven’t read. It also makes me think about not mentally constructing narratives (which, more often than not, become red-herrings) based off of information I have.
Finally, despite being so apprehensive on the idea of predicting the future, Taleb also claims that “In the end we are being driven by history, all the while thinking that we are the ones doing the driving.” There are many ways to interpret this great quote, but I see it as history being contextual. In other words, we don’t control history, rather, events happen and build off of one another, whether this be in random fashion or through similarly repeating patterns (the latter being my theory, though this might just be me mentally constructing a narrative based on information I have).
If Mr. Taleb ran a hedge fund (which he previously did, under the name Empirica Capital), I’m guessing his strategy would be to benefit off short term, consistent profits while always hedging against potentially huge losses. This way, he would be able to see gradual returns safeguarded from catastrophic events. Likewise, he could make his bigger bets on positive Black Swans, while pursuing a small-scale trading strategy in the time between these rare occurrences. This is where the options trading, which my previous interviewee brought up, comes into play. A strategy founded on single, greatly dispersed events which are not even guaranteed to happen may sound like an overly passive, boring, and minimally profitable way to run a hedge fund. However, in recent years, many asset management companies have gone bust by neglecting Taleb’s advice and discrediting the power of Black Swans. One such instance which comes to mind are the number of hedge funds which invested in cryptocurrencies like Bitcoin, not expecting the price to drop from ~$20,000 all the way down to sub-$4,000.
Bitcoin Graph
Another more salient example is with optionstraders.com, a Florida-based hedge fund which recently went bust after unexpected volatility caused his positions to collapse completely. The hedge fund’s manager, who I don’t need to name because he is receiving enough bad publicity already, founded this fund on the premise of ‘naked options’ (as opposed to covered options) — essentially buying either puts or calls without hedging potential losses as a method to save money. The problem with this strategy is that if you sell a call for a stock at a strike price of $50, you are hoping that the stock will remain    <= $50 up to a certain date. However, if the stock rises above 50$, the call seller loses money, since he has to buy the stock at this price which is more than what he paid for. In a very unlikely, and very unfortunate, scenario, the price of the stock could grow an incredible amount (say it goes up to $500), and the resulting losses would be even more incredible. This is what happened to optionstraders.com, which lost all of its clients’ investments (+ more) because of an unpredictably volatile period for crude oil prices.
 

To Conclude…

I’ve always found that finishing a quality book is an overwhelming experience, due to the surplus of information you quickly gained (not to mention trying to remember all of this!) and also because it leaves you feeling a bit empty, as if you’ve lost something. Right now, I think of what Taleb’s work can teach me about my project. Some lessons I’ve learned from reading this book are:

  1. I need to think about what a Black Swan would look like in my context, and how I could protect against this.
  2. I need to familiarize yourself with statistical terminology. This problem was apparent to me before reading the book, but it’s now been solidified. I have never taken a formal statistics class (which is fine considering Taleb’s theory on harmful over-saturation of information), but it’s clear that I have to do some more focused learning on my own.
  3. “Note that a ‘history’ is just a series of numbers through time” (119). I was so happy when I read this, as it immediately made me think about how my project is essentially using ‘a series of numbers’ (right now, open/close prices + volatility) to map historical events.
  4. Don’t be quick to construct narratives, as this can greatly distort perception and information (you try to fit new information to the narrative you’ve established, without ever thinking if this makes sense).
  5. Don’t be too stressed about shortcomings, because failure is only failure if you are failing on your own established objectives, not some nebulous criteria set by outside sources.

 
Trying to cover every interesting topic raised in The Black Swan would be a senseless attempt, as you can simply read the book yourself considering that Taleb does a much better job explaining these issues than I do. The topics I mentioned above are the ones I find most relevant to my independent project on using Word2Vec to map similar patterns in market behavior, and also some which intrigued me most. I’m interested in hearing what other people who have read/know of this book think about its implications, and what some important lessons I might be missing?
I’m looking forward to my next read, which is a bit more focused and specific than the previous two: Advances in Financial Machine Learning by Marcos Lopez de Prado. From what I’ve heard (and read), this is more of a textbook-type piece which looks at actual solutions to common problems in ML and trading algorithms. Apparently, “Readers become active users who can test the proposed solutions in their particular setting”, which would be great in my case, as I’m moving into more actual programming and implementation.
 

“We no longer believe in papal infallibility; we seem to believe in the infallibility of the Nobel, though.” (Taleb, 291)

 
https://i1.wp.com/www.teninsider.com/wp-content/uploads/2017/05/Lehman-Brothers-Bankruptcy.png
https://www.welivesecurity.com/wp-content/uploads/2013/07/blackswan.jpg

The Shortcomings of Neural Networks for Trading Predictions

As someone who is devoting a large-portion of their senior year (and very likely time beyond that) to researching potential applications of deep learning in trading, I wasn’t thrilled to learn about the recent shortcomings of quantitative traders. Let’s begin with Marcos López de Prado, a frequently cited algorithmic trader who recently  published Advances in Financial Machine Learning. One thing that De Prado talks about is the idea of ‘red-herring patterns’ that are extrapolated by machine learning algorithms. These types of algorithms are, by design, created to analyze large bodies of data and identify patterns within this data. In fact, this idea of noticing patterns is one of the main assumptions I am basing my work on (using Word2Vec embeddings to identify past financial patterns and apply them to real-time data for more accurate predictions). But, what happens when these algorithms identify patterns that aren’t real? An aggressive neural network (In my case: One which adjusts vector weights heavily while learning from data) is prone to make these types of mistakes. Think of this example: A stock happens to go up a couple percent points every Thursday for three weeks in a row. A (poorly written) neural network would deduce that every Thursday in the future, this stock would go up by at least a percent point or two. Now, this is easily avoidable by training a trading algorithm on larger sets of data, but even large data sets are prone to these types of red-herrings. Once a trading algorithm clings on to a pattern, it could backfire horribly when that pattern eventually breaks.
Black-Swan-900
This brings the idea of Black Swans into light. The theory of Black Swans was popularized by Nasim Taleb in his accurately-titled book The Black Swan: The Impact of the Highly Improbable. The general gist of this theory is that the most profoundly impactful events oftentimes are the ones we least expect, due to our fallacious tendencies in analyzing statistics (I will go into more detail on these topics and more in a future blog post, once I am done reading the whole book). Taleb argues that one of our biggest shortcomings in analyzing data is creating ‘false narratives’, which are more convenient and easier to sell to clients. These false narratives oftentimes omit crucial data (silent data), which backfires once the narrative breaks.
But, on the other end, a more passive neural network (one which more slightly adjusts vector weights) can sometimes come to no meaningful conclusions, which means wasted time and computational energy. I want to create a Word2Vec model which can detect patterns, but I also don’t want it to actively follow patterns with no longevity.
So, what does one do? How aggressive/passive should I make my Word2Vec neural network? 

Another theory which I encountered over the weekend is the idea of survivorship bias. In training neural networks, how do we treat data from companies which have failed? If we are analyzing the stock price data for various important stocks over time, what do we with data from once-important stocks which are now defunct, such as Lehman Brothers? I initially thought it would be best to throw this data out, since it is no longer applicable, but it turns out this strategy can have negative consequences. If we only train our network on stocks which have survived, then we will miss out on crucial data about when stocks go bankrupt. So, how do we properly treat this type of data?

lehman

All of these seemingly insignificant flaws in trading algorithms can evoke catastrophic mistakes. This concept is synthesized by quantitative investment officer Nigol Koulajian, saying: “You can have one little pindrop that can basically make you lose over 20 years of returns.” This ‘little pindrop’ which Koulajian mentions is the eventual divergence from the false patterns identified by neural networks. I personally think it would take more than a little pindrop to erase 20 years of returns, but the idea still stands. So, this warrants the question, how do we avoid the little pindrop? My (far-fetched?) theory is that you can use neural networks to estimate worst-case scenarios int the same way they are designed to estimate best-case scenarios, and then work to avoid this.
In broader terms, Bloomberg reports that the Eureka Hedge Fund Index, which tracks the returns of hedge funds which are known for using machine learning, has under performed yearly compared to the S&P 500. The harsh truth (right now) is that simply investing in the S&P500 will return ~13% yearly, while machine-learning based hedge funds return ~9% yearly.
Eureka Hedge Fund Index
(The keen observer will notice that despite all the noise, the index has been steadily going up over the past 7 years)
These are some of the questions I ask those few who read what I am writing, and are the types of questions I will ask through my personal research interviews (Good News! I have my first interview scheduled this upcoming Tuesday, and, interviewee permitting, I will post a summary of our talk later in the week).
In my personal opinion, the recent under performance of trading algorithms in general is not a bad sign. This is still a relatively new field, meaning that more research needs to be done and new discoveries need to be made. I think of it this way: If trading algorithms are working perfectly, then what’s the point of a newcomer (like me) coming in and doing research on them? If it ain’t broke, don’t fix it.

William Deresiewicz and Homework-Induced Stress

Last week I attended a lecture/Q&A hosted by William Deresiewicz, the famous ex Yale professor who criticizes schools and questions the efficacy of the stressful college application processes. Throughout the discussion, he wasn’t hesitant to remind the crowd full of future college students that they shouldn’t stress out about homework assignments and similar trifles. One statement regarding school-induced stress that stood out to most of the attendees was “The truth is that you guys are not going to end up sleeping under a bridge; it is much more likely that you’ll end up jumping off of one instead.”  This was a powerful assertion which is presumably correct, considering the ridiculous amounts of mental pressure that teenagers have to deal with nowadays. (Click here if you don’t trust my claim).
As a teenager who lives in America, I can confirm, because of interactions with peers and from personal experience as well, that teenagers in America are actually subject to large amounts of stress. Obviously, there are people who really don’t care about what’s happening in their classes and avoid doing homework at all costs, but if you want to be somewhat “successful” in your high school education, then you will have to wade through the copious amount of work.
Thanks to recent research and to books published by people like William Deresiewicz, most people in America are aware of this fact. But like with most situations in the nation, it seems as if we are spending too much time attempting to find out how to deal with the problem that we are facing rather than actually asking ourselves why the problem exists in the first place. We spend a considerable amount time talking to our kids about ways to cope with stress so to make sure that they don’t experience mental breakdowns, but we have never discussed why such pressure subsists.
So, why do American high schools assign so much homework? Why are college admission processes so demanding and time consuming?