Antifragile Investing

by: Kevin Wilson


Markets have been flat for 18 months, showing signs of topping; risks to both the global economy and markets appear to be asymmetric; a surprise (Black Swan event) is possible.

If central bank intervention fails and Modern Portfolio Theory fails to provide adequate protection as it did in 2008, we are forced to seek what author Nassim Taleb calls "antifragility".

A portfolio with convex or antifragile behavior is actually stronger the worse things get; although long-term bonds (TLT) seem to be antifragile on first glance, in fact they are not.

A portfolio that is 90% cash and 10% thematic stocks (e.g., ORA, AWK, FANUY, TSLA, LNKD) is actually antifragile because downside risk is limited and big upside is theoretically possible.

Nassim Nicholas Taleb


In an uncertain world, many seek to offset various identified risks by using models of expected behavior. For this reason we have seen the rise and widespread acceptance of Modern Portfolio Theory ("MPT") over the last 30 years, which has, in turn, led to the rise and increasing popularity of passive investing vehicles like index funds and passive ETFs. In spite of the massive damage done to passive investors in the Great Financial Crisis, which saw markets drop by 56%, there has been a recent resurgence in the popularity of passive investing. This seems to be due to four different influences on investor sentiment: 1) one of the longest bull rallies in history, driven by incessant interventions in the markets by central banks, and characterized by consistent and substantial underperformance by actively managed funds of all types; 2) the long period of ever lower yields on bonds, making higher yielding stock ETFs more appealing to investors; 3) the recent, 18-month-long period of essentially flat markets, characterized by high volatility and continued underperformance by actively managed funds and 4) the admittedly much-reduced costs associated with many passive investing vehicles. The last of these reasons is a strong argument for passive investment vehicles, but not necessarily for passive investing per se. By this I mean that asset allocation (see example below) is still critical to success, and there are cogent reasons to suspect that a passive approach to allocation may not work out very well in the coming months and years, as I have attempted to show in a previous article.

Example of a Defensive Tactical Asset Allocation:

Source: Author

In addition, as the chart below shows, the markets have grown so dependent on central bank intervention that the bull market seemed resilient to almost anything that happened, that is until it completely stalled out once the Federal Reserve decided to end its Quantitative Easing ("QE") program in October of 2014. Now it seems to be forming a top similar to the one in 2000, and it does not seem so resilient any more. Thus the question arises: if markets are now dependent on central bank interventions, in what way has risk evaluation been affected, and how might this spill over into how risk management operates going forward? Is it reasonable to assume that central bank interventions will always boost the market? Answer: no, of course not; as we saw in both 2000 and 2008 (see second chart below), once investors are risk-averse, nothing the Fed does makes any difference.


Federal Funds Rate Cuts Fail to Stop Market Panic in 2008:

Source: John Hussman;

Recall that "MPT" includes amongst its basic assumptions the demonstrably false idea that markets are efficient, the problematic postulate that reward is proportional to risk in the long term, and the equally dubious idea that investors are generally rational. I have discussed the evidence undermining MPT elsewhere, but I think it is important to note here that there have been extended periods when markets have mispriced equities (e.g., the tech bubble's monstrous overvaluation in 1996-2000, and then its aftermath of much lower valuations, especially in certain sectors). Famous investor Warren Buffett has roundly criticized the Efficient Markets Hypothesis ("EMH") component of "MPT" for its violations of common sense regarding security valuations. There is also clear evidence that actual data do not support a proportionality between reward (returns) and risk (Beta or Sigma) as envisaged by the Capital Asset Pricing Model ("CAPM") component of "MPT," at least for long periods of time (such as the 20-year period ending in 2009, according to data from JPMorgan; see also the chart below from Societe Generale). This has been clearly demonstrated by data published in 2003 by the same Nobel laureate, Eugene Fama, who developed the "CAPM" in the first place. Finally, the economic theory that investors generally act rationally has been completely undermined by over 25 years of research on behavioral finance, which now includes amongst its laurels two Nobel prizes (to Kahneman and Shiller) in economics.

Click to enlarge

Source: Dylan Grice, Societe Generale;

It has long puzzled those who are critical of "MPT" how they should proceed on the subject of managing risk. In spite of all of the evidence cited above to the contrary, and in spite of their massive losses in 2007-2009 that were not predicted by their models, most big firms still use some version of Value at Risk ("VaR") modeling to monitor and control the risk of their in-house portfolios. Famous author and thinker Nassim Nicholas Taleb has comprehensively criticized such models as unrealistic and prone to all sorts of errors. His books include Fooled by Randomness (2004, 2nd Ed.; Random House, New York, 316p), The Black Swan (2007; Random House, New York, 366p), and Antifragile (2012; Random House, New York, 519p). One of Taleb's chief contributions has been his rigorous evaluation of asymmetric or "fat tail" risk. He has made a compelling argument in favor of the idea that most change comes from huge events that are intrinsically unpredictable, i.e., "Black Swans." A classic example of a Black Swan might be the massive 22% drop in the markets that occurred on October 19, 1987. This was completely unexpected and, although caused by a variety of stressors in the system, was at least in part caused by program trading.

The Record One-Day Crash of October 19, 1987:


Now notice the striking similarity between the 1987 stock crash and Taleb's metaphor of a turkey approaching and then arriving at Thanksgiving Day (see chart below). Obviously, from the turkey's point of view there is no warning of what is coming, and the result is catastrophic (for the turkey at least). Applying Taleb's metaphor to the 1987 crash, it seems clear that fat-tail risk was not sufficiently discounted by pretty much everyone. Of course, there were a few people who actually benefited from the crash, either because of dumb luck or what Taleb calls "antifragile" characteristics in their asset allocations. I personally did very well out of that catastrophic market collapse, as I sold a month before the drop and then bought back in during the week it bottomed. However, I was a geologist at the time, and it was clearly an example of once-in-a-lifetime dumb luck. Antifragile (as used by Taleb) means that something has an intrinsic quality of thriving under or benefiting from conditions of volatility, randomness, disorder, and/or powerful stressors.

Click to enlarge

Source: N. Taleb;

Another example of fragility in the face of a Black Swan event involves the catastrophic losses incurred by Parisian bank Societe Generale (OTCPK:SCGLF) (OTCPK:SCGLY) in January 2008, when they were forced to sell $70 billion in stock positions in a "fire sale" triggered by illegal trading activity hidden from them by one of their traders (Jerome Kerviel). Other such events might include the Flash Crash of May 2010, and the Subprime Mortgage Crisis of 2007-2009. Black Swans are often observed in nature or in the interaction of mankind with nature: e.g., the Tohoku earthquake, tsunami and Fukushima nuclear meltdown of March 2011 in Japan. There is good reason to suspect that new Black Swan events could hit the markets again at any point in time. By definition this would not be an expected event, which automatically might exclude current risks that are at least partially discounted, i.e., the failure of Abenomics in Japan, NIRP in Japan and Europe, a bank crisis in Italy, a credit collapse in China, etc. However, some unpredictable chain of events that involves a combination of the aforementioned threats with new (unknown) factors might easily produce a new Black Swan. So although we can't predict Black Swans, we can take note of conditions of asymmetric risk or imbalances that suggest risk is not adequately modeled.

In his most recent book, Taleb describes how fragility and antifragility work mathematically, and in practical examples. Let's use your car as an example of fragility. If you drove it into a wall at 50 mph it would cause far more damage than ten times the amount caused by running it into a wall 10 times at 5 mph. Likewise, if you jump off a building 30 feet tall, the damage done to you will be far more than that resulting from jumping three feet ten times. In fact, as Taleb points out, you most likely will be dead. If you were to chart this as a time series you would see a feature called concavity, as shown in the chart below. The chart indicates that as the variable in question ("x") increases, you experience (moving downward on the curve from the spot labeled "you are here") losses that increase at a faster and faster rate, producing a line that curves inward. If you move in the opposite direction on the curve towards a lower level of the variable "x," you experience negative asymmetry, i.e., the gains are smaller than the losses incurred when moving the same distance negatively.

Concavity Produced by Fragile Behavior:

Source: Nassim Taleb;

If instead we set out seeking the opposite of fragility or concavity, i.e., antifragility or convexity (respectively), we would look for things that do better and better the worse things are. An investing example of this would at first glance seem to be the use of long-term Treasuries in a balanced portfolio during a recession and bear market. The worse things get as the economy worsens and the stock market collapses, the better the Treasuries do. The example of this from the 2008 market crash (using the iShares 20+ yr. Treasury Bond ETF (NYSEARCA:TLT), as the bond asset) is shown in the first chart below. We will come back to discuss this example again later because it may not actually turn out to be a very good example of antifragile behavior. The theoretical depiction of an antifragile effect is shown as convexity in the second chart below; i.e., as the variable "x" increases (moving upward from the spot labeled "you are here") the gains increase at a faster and faster rate, producing a line that curves outwards. Under convexity (antifragility), if you move in the opposite direction on the curve towards lower values of "x," you experience negative asymmetry, i.e., the losses are smaller than the gains incurred when moving the same distance positively.

The Possibly Antifragile Behavior of Long Treasuries During the Panic of 2008-2009: Click to enlarge


Convexity Produced by Antifragile Behavior:

Source: Nassim Taleb;

Obviously, if it would be possible to find assets that are reliably antifragile or construct portfolios that have reliably antifragile characteristics, it would be to our benefit as investors, especially when new Black Swan events transpire or if highly asymmetric risks became prevalent and were recognized as such before an otherwise unpredictable event. There are a number of very good examples of Black Swan events in the markets that we could use as backtests of the antifragility of assets, including those already mentioned above. Of course, there are many problems with backtesting, such as survivorship bias, the occurrence of one-off events (the arrow of time effect) and of course the fact that "past performance is not indicative of future returns." Taleb points out that the way to further evaluate presumably antifragile characteristics is to use an "antifragile detection heuristic," i.e., to look for acceleration over time. Even a flawed model can prove useful if all we need to see is acceleration (i.e., it's not the level that counts, but the change in the level and its acceleration).

If, for example, we return to the idea of using long-term bonds in a balanced portfolio as we discussed in the example above, there may be a problem with labeling long-term bonds as an antifragile asset. If you think through the notion of the antifragility detection heuristic, it may not actually be true that bonds experience lower losses when rates or stock markets go up than the gains made when rates or stock markets go down. In the chart showing the behavior of the long-term bond in 2008 and thereafter, note that all of the relative gains made in 2008-2009 were completely gone by 2014. This is made more difficult to judge by the lingering impact of QE and the new impact of global NIRP. We must also make allowances for the secular bull market in bonds since 1982 (see chart below), which overprints long-term bond performance. The reversal of this secular trend would presumably push bonds into a secular bear market lasting decades and this would diminish their attractiveness as hedging assets, accordingly.

Click to enlarge

Source: Gary Shilling;

The antifragile behavior of a portfolio example used by Taleb seems much more likely to operate as expected. The chart below shows the model allocation, which is simply 90% cash and 10% risky thematic stocks. The stocks could be anything, but for purposes of argument let's say they include a clean energy stock (Ormat Technologies, Inc. (NYSE:ORA)), a water resources stock (American Water Works Company, Inc. (NYSE:AWK)), a robotics stock (Fanuc Corp. ADR (OTCPK:FANUY)), a driverless car stock (Tesla Motors Inc. (NASDAQ:TSLA)), and a social media stock (LinkedIn Corp. Class A (NYSE:LNKD)) in equal proportions. If you think through the antifragility detection heuristic, with a 90% cash allocation your downside is limited to 10%. But if one or two of the thematic stocks in the portfolio do well over time, your upside is asymmetrically higher, possibly much higher if one of them is theoretically a big winner.

Taleb's Generic Model Portfolio:

Source: Nassim Taleb; Author

This portfolio thus satisfies the requirements of an antifragile portfolio by placing tight constraints on losses and simultaneously leaving you free to take big risks and seek the potential for big gains with the 10% allocation to stocks. In other words, it delivers convexity in its results. Of course, explaining how this works to the average client would not be a picnic because many will ask why all that cash is necessary, and why isn't that a mistake to leave so much money idle? But if the goal is antifragility, this portfolio works. Given the increasing level of volatility in the markets, the increasingly artificial functioning of markets under constant central bank interventions, the extended nature of the bull markets in both stocks and bonds, and the extreme levels of economic and financial risk now visible around the world, it may just be worth considering.

Disclosure: I am/we are long TLT, AWK.

I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.

Additional disclosure: This article is intended to provide information to interested parties. As I have no knowledge of individual investor circumstances, goals, and/or portfolio concentration or diversification, readers are expected to complete their own due diligence before purchasing any stocks or other securities mentioned or recommended.

Editor's Note: This article discusses one or more securities that do not trade on a major U.S. exchange. Please be aware of the risks associated with these stocks.