Risk Update: January 2021 – Uncapitalized Tails by design.

Like many of you, we could not help but have our attention grabbed by the activities in a previously little-known US stock by the name of GameStop. As most will be aware, it witnessed an epic short squeeze as, allegedly, retail investors sniffed out excessive hedge fund shorts and proceeded to exact significant punishment on them.

Figure 1: GameStop (GME) Share Price

Source: Bloomberg

The news around the GameStop squeeze dominated market chatter in the last days of January, even rising to the lofty heights of being raised as an issue at the presser of Fed Chair Jay Powell. (https://www.reuters.com/article/us-usa-fed-powell-gamestop-idUKKBN29W32J). As noted in the linked article, Chair Powell brushed aside any premise that somehow their policies of zero interest rates and unlimited QE could have anything to do with frothiness of asset values, or with the odd behaviour as seen with GameStop.

We, of course, would beg to differ. Low interest rates and relentless injection of liquidity/removal of the supply of carry securities, on their own, may not necessarily account for the frothiness, though we suspect it does have some connection. We would argue that the question, per the way we look at the world, should revolve around how the Fed’s actions lead to fragility in the financial system. They are the principal drivers of a fragile unstable system and, in many respects, Wall Street is the nuclear core of that fragility.

This fragility, as we often discuss, revolves around what we refer to as “uncapitalized tails”; risks taken predominantly by fiduciary managers that they neither adequately capitalize nor represent. The Fed’s role in this boils down to three interconnected and reflexive drivers:

  1. Moral Hazard. By perennially bailing out the levered risk takers (banks, investment funds, NBFIs), the Fed encourages maximization around the flawed asymmetric incentive structure of fiduciaries (ie. participate on the upside but not the downside).
  2. Low cost of leverage. Incentivize the levered search for return throughout the system.
  3. Regulatory construct. A reporting mechanism that acts as cover to asymmetric risk taking with other people’s money (and eventually taxpayer’s money).

The final point on the regulatory construct is the area that we spend most of our time focused upon. How are the accounting, capital requirements, risk measuring and reporting methodologies being used to hide risks by those that do not bear the downside themselves? The GameStop situation is a great example of just the type of uncapitalized, mis-reported risk that we are always harping on about. Two major shortcomings of risk reporting in the levered fund manager space (be it PQR in the US, Annex IV in Europe, or similar elsewhere) are the total lack of anything related to non-linear risk (there is a big difference between the risk of being long or short an option) and, secondly, a seeming lack of any knowledge of the risks of short equity positions (ie. if you own an equity the most you can lose is your capital but if you are short your loss potential is infinite). These two shortcomings obviously intersect where a fund manager, for example, is short call options on equities, likely something relevant to what was going on in GameStop.

At a, supposedly, more sophisticated portfolio level using a market standard like Value at Risk (VaR), a short position in GameStop, an equity that had historically been generally positively correlated to the overall market, would be considered to reduce the riskiness of an otherwise net long overall portfolio. We can fire up our simple VaR calculator tool to show how this would look over an 18mth data series up till late August 2020 (a period where one might say GameStop behaved normally). Below we compare a portfolio that is 70% long of SPX and 30% short of GameStop to a portfolio that is 100% SPX. You can see the portfolio that includes the GameStop short, by these measures, is significantly less “risky”, with lower VaR, lower cVar, and better skew and kurtosis dynamics.

Figure 2: Long 70% SPX and Short 30% GME vs Long 100% SPX (18mth data to 28Aug’20)

Source: Convex Strategies, Bloomberg

We can, in a sense, normalize the risk by reducing the risk in the 100% SPX portfolio down to a 70% long in SPX and placing the 30% into US Treasury bonds. Looking at this comparison, the VaR and cVar numbers look similar, but the portfolio with the short in GameStop still expresses less risky skew and kurtosis factors. The short GameStop portfolio manager, utilizing this sort of market standard risk tool, could rightly argue that their portfolio is less risky (and likely justify applying leverage to it!).

Figure 3: Long 70% SPX and Short 30% GME vs Long 70% SPX and Long 30% US Tsy (18mth data to 28Aug’20)

Source: Convex Strategies, Bloomberg

Now, we can take the 70% long SPX and 30% short GameStop portfolio and look at it during the previously stated 18mth period ending late August 2020 and compare it to the most recent 18mth period terminating in January 2021. Obviously with the sharp short squeeze, the GameStop position was clearly not a risk reducing component of the portfolio! The 99%tile daily VaR of -6.95% as of August, turns into a single actual realized worst day of -42.25%, and remember the GameStop short was just 30% of the overall initial portfolio allocation.

Figure 4: Long 70% SPX and Short 30% GME Portfolio (18mth data to 28Aug’20 and to 20Jan’21)

Source: Convex Strategies, Bloomberg

We can more clearly see the portfolio destruction using our scattergram and compounding line views over the two data series, again compared to the 70% long SPX and 30% long US Tsy portfolio. In the initial set of pictures, up till late August 2020, the two portfolios seem somewhat similar as represented by the previous VaR risk calculations.

Figure 5: Long 70% SPX and Short 30% GME vs Long 70% SPX and Long 30% US Tsy (Mar’09-Aug’20)

Source: Convex Strategies, Bloomberg

In the revised data series, terminating at the end of January 2021, you see the total wipe-out of the portfolio, and then some! So a portfolio that would report itself as a mere 100% gross and 40% net long, with a 99%tile daily VaR of circa -7%, just lost in the region of 10x its AUM. Pretty useless disclosure of the risk, one might say, and quite definitively an uncapitalized tail.

Figure 6: Long 70% SPX and Short 30% GME vs Long 70% SPX and Long 30% US Tsy (Mar’09-Jan’21)

Source: Convex Strategies, Bloomberg

This is precisely how the funds, that were alleged to have the shorts, were on the verge of total wipe-out. Of course, these funds could have constructed their “short” through buying a put option on GameStop, or alternatively by selling a call option. If hypothetically they had chosen to do so by buying a 30delta put, or by selling a 30delta call, on the equivalent notional of their full AUM, they would have, for regulatory reporting purposes, reported it just the same as having taken the 30% outright short position, and both the put buyer and the call seller would report the same 30% net short delta. In this case, however, the put buyer in the end would have been exposed to a maximum loss of the option premium paid. The call seller on the other hand, as opposed to losing the 10x of AUM that the outright seller lost, would have lost something more in the region of 30x their AUM. All reported the same ‘risk’ to their regulators.

Not all that surprisingly, given the scale of capital destruction, the squeeze in this little stock, along with a few other names caught up in the excitement, was sufficient to trigger a general risk off environment in the final three days of January in the broader market. In our simple example, as the manager with the 30% short position in GameStop quickly runs out of capital, they also have to liquidate the 70% long SPX part of the portfolio. All these little tremors impact the volatility and correlation that is feeding into the Var, RiskMetrics, CAPM, Vol Optimization, Risk Parity, Balanced Portfolio methodologies across the investment universe and risks triggering a general risk reduction throughout the system. Naturally, the powers-that-be would not want that to happen, so manoeuvres will no doubt be undertaken to try to put a stop to the whole thing. We have already seen Robinhood, the broker through which the retail buyers were most active (and a few others as well), in the midst of the hottest of the squeezing frenzy, put a halt to purchasing of GameStop on their platform.

We have discussed frequently over the last several months (https://convex-strategies.com/2020/12/21/risk-update-november-2020/) the Fed’s (and their followers) explicit purchase of “carry assets” commencing on March 23rd as a direct bailout of uncapitalized risks on the books of levered investment managers/NBFIs. There is no reason to doubt that those in charge will not continue to practice a policy of averting market dislocations resulting from the fragility that they themselves have cultivated. One wonders if the Fed, who is clearly willing to protect funds by buying the assets that they have levered beyond their capital’s ability to absorb the loss, would go so far as to sell stocks that funds have likewise shorted beyond their loss absorbing capacity? We suspect not, but in the modern central banking world of “whatever it takes”, one never knows.

Figure 7: GameStop vs VIX Index

Source: Bloomberg

The system is awash with fragility. Some may have seen us draw pictures of trees with great big leafy canopies, but skinny trunks and shallow roots, thus very fragile to big gusts of wind. That’s just one of our analogies (we’ve got a few others!) of a system that doesn’t have sufficient capital to support the risk it is underwriting. The game, if you will, for the central banks is keeping the whole thing afloat, keeping more and more plates spinning. They will keep incentivizing the use of leverage to sustain the accumulated imbalances, making the imbalances increasingly unsustainable.

We regularly get asked our thoughts on potential exogenous risks to the system. As a general rule, we slap such questions aside and focus our discussions on what matters, ie. fragility of the system and portfolio risk construction. We do not know, any better than anybody else, what or when something is going to happen that triggers the next correlated deleveraging of the system. We are, however, human and enjoy a little friendly banter every now again, so when recently asked we posed a response that fits within our framework. We proclaimed that we could perceive two big potential event risk shocks looking forward this year:

  1. The vaccine does not work. It certainly appears as if pricing in asset markets, across basically the entire swath of asset classes and geographies, continues to factor in strong recoveries as economies reopen. Despite continued start/stop lockdowns across much of the world, there is seemingly a strong belief that as the rollout of the vaccine extends things will quickly return to normal. If the vaccines turn out to be a disappointment in their efficacy, it could prove rather disappointing relative to current valuations.
  2. The vaccine does work. Policy settings, both for central banks and governments, are set on maximum potential support for economies. We are already seeing reflationary frictions around the globe. If the vaccine is a storming success, will policy makers pull back their unprecedented support before even greater dangers of inflation materialize? Will they be able to pull back said support as asset markets have already absorbed the past support into bubble-like levels?

Obviously, we were moderately tongue in cheek, and merely rolled the answer back to our basic premise – the system is fragile and there is a very clear disconnect between valuations and underlying fundamentals (the difference being sustained by the policy enhanced, fragility extending, leverage in the system). We showed last month https://convex-strategies.com/2021/01/22/risk-update-december-2020/ how the principles behind Jensen’s Inequality can help us construct resilient, superior compounding, portfolios whether we get inflation or deflation, up or down markets. We are continually frustrated hearing (usually fiduciary) portfolio managers balk at the thought of “paying” for protection, even as they sit with massive dead capital allocations to supposed “defensive” investments that just flat will not accrete to compounded returns.

We can break out our good ole coin toss game to help visualize the power of insurance. Please refer back to our September 2020 update for a lengthier breakdown of the coin toss game and the whole ergodic vs non-ergodic stuff. Using our standard bet where if the toss comes up heads you win 50% of your capital, and if it comes up tails you lose 40% of your capital, over 40 tosses (think of a toss as a year in the cycle of your retirement investment portfolio) the median outcome for 200 players of $100 starting capital is to end up with terminal capital at $12. Many of the 200 players are bankrupt. Most have lost the bulk of their capital. A few lucky ones have done really well, thus the “average” terminal capital, in this simulation, turns out to be $240.

Figure 8: Coin Toss Simulation +50% vs -40%. 200 Players. 40 Tosses

Source: Convex Strategies

As we have discussed before, the solution to most people losing most of their money, courtesy of the Kelly Criterion, is to deleverage your at-risk capital, aka “The Balanced Portfolio”. So only put 60% of your capital into the game, and hold the other 40% in a risk-free asset, hopefully with some sort of yield, aka “The 60/40”. We can punch these allocations into the simulator and have another go at it. So now we put 60% in the game and hold 40% in risk-free bonds at a 1% annual yield. This generates a median terminal capital of $67 and, in this particular run, an average terminal capital of $283 (note, the average fluctuates a lot from one simulation to another, wealth is very much not Gaussian!). Generally good news, as no participants have gone bankrupt, and most people have at least some money to retire on!

Figure 9: Balanced Portfolio: 60% in Game/40% in 1% yield Risk-Free Bonds

Source: Convex Strategies

Now for the interesting part. Regular readers are likely familiar with our simple rule of thumb to transform from a Balanced Portfolio to a Barbell Portfolio: take your fixed income allocation and split it 50/50 between more equity exposure and long volatility protection, then 2x lever the protection. This leaves us with our standard Barbell weightings of 80/40. So, for the coin toss game, we rebuild it to an 80% allocation to the game, and 40% in insurance. The insurance costs 2% of the entire portfolio on each toss (so 5% on the 40% allocation) and cuts off 100% of the loss on a third consecutive result of tails. Note, the insurance does nothing on the first and the second consecutive 40% losses from realizing a tails on a toss, it is only wing insurance, you have to lose 40% twice before it kicks in at all. All our regular readers will know what is coming, but amazingly so few people out in the real world seem the least bit aware of it. The Barbell Portfolio generates a median terminal capital of $758, nobody is going bankrupt, and the average terminal capital of the run is $2,111. Everybody is getting rich!

Figure 10: Barbell Portfolio: 80% in Game/40% in Tail Protection

Source: Convex Strategies

The value of cutting off the extreme negative compound cannot be overstated. It is just math. Makes one wonder, what if we paid more and got even more protection? What if we doubled the cost of protection to 4% of the portfolio (so 10% of the 40% allocation to protection) such that in the third consecutive tails we cover not just 100% of the loss, but 120% of the loss? A quick simulation of that generates a median terminal capital of $2,349 and average terminal capital of $8,660.

Figure 11: Barbell Portfolio: 80% in Game/40% in Juiced Tail Protection

Source: Convex Strategies

Before you think that we have moved into Crazytown here, this is, give or take, what happened with our Always Good Weather Portfolio in 2020 (30% SPX, 30% NDX, 80% CBOE Long Vol). The aggressive allocation to downside protection, through the CBOE Long Vol Index, meant that the otherwise negative compound of March 2020 became a positive profit event. Then the additional topside risk in Nasdaq 100 captured the series of tosses that came up heads, one after another.

Figure 12: Always Good Weather Portfolio vs Balanced Portfolio

Source: Convex Strategies, Bloomberg

That is a pretty clear real world example of which is more effective, deleveraging or structuring protection. Not only does the “insured” portfolio, just like in our simulations above, generate much greater compounded returns, but it is also much less “risky”. That is the point of insurance! Poor defensive strategies impair the return of a portfolio by reducing exposure to the upside while not adding true protection to the downside. Good risk mitigating strategies allow more participation in the upside and, if structured well, can turn fat loss tails into fat profit tails. A goalkeeper should not be evaluated by the same metric as that used for a goalscorer. Negative carry does not mean negative expected return, just as much as positive carry does not mean positive expected return. If you think otherwise, you aren’t looking at the right time horizon.

Read our Disclaimer by clicking here