Ironically (the markets do have an uncomfortable habit of throwing this stuff right back in your face), this is precisely what we saw in February in terms of the relationship between equities and gold. GLD finished the month down slightly. However, the monthly number obscures a sharp selloff that occurred in the last couple days of the month. Why did it happen?
I don’t know that anyone knows for sure (if any of you are traders or market makers with special insight, please leave a comment!). My personal hypothesis is that this was a function of investors rebalancing portfolios, taking down gross exposure and getting margin calls. Gold in particular has had a tremendous run over the last 8 months or so. Since I started running this portfolio, GLD is up 23%, while the S&P 500 has returned 2.77% and the BBgBarc Aggregate Bond Index nearly 9%.
Portfolios with a static allocation to gold are probably overweight it. And if you need to sell something for whatever reason, what makes the most sense to sell?
This is the kind of “real-world” trading activity that takes correlations to 1 in a crisis environment. And this portfolio is certainly vulnerable to it, as we saw in February (albeit to a relatively mild degree). It’s why volatility and trend are used as overlays for risk management.
Incidentally, the one-year lookback I use didn’t flag a need to add cash to the portfolio.* A one-month lookback would have, with trailing one-month volatility of about 15%. Equity exposure would have been trimmed to add the cash, with most equity market segments having crashed through their 200-day moving averages in February.
Why do I not use a one-month lookback? Originally, I did. However, I became concerned that such a short lookback period might be too sensitive to very short-term shocks, whereas the strategy is intended as a strategic allocation more geared toward navigating changes in market regimes.
Candidly, I’m not sure this is the right decision.
But we’ll see.
* Astute readers may notice that the portfolio weights in this update differ slightly from those in the previous update. There are two reasons for this. First, I trimmed and rebalanced an overweight to ex-US equity exposure that had crept in. Second, yesterday I sold some of the gold and NTSX exposure to make some purchases in my individual stock portfolio. If you’ve been following these updates since the beginning you may recall that I pair this strategy with a concentrated, high-risk, 10(ish) stock equity portfolio.
With January over I ran the latest leveraged permanent portfolio rebalance check. Still in good shape from 12% volatility limit perspective. Relative performance versus the S&P 500 has diminished since 4Q18 rolled off the lookback period. But it remains quite strong versus a Global 60/40 comp.
Actual realized performance from my implementation:
January is an interesting month because it demonstrates the diversifying power of uncorrelated assets (gold, Treasury futures in NTSX) in the face of macroeconomic event risk. In this case, coronavirus.
Recall that the whole purpose of this approach is to be insulated from unexpected macroeconomic or geopolitical shocks without having to predict anything. So far we’ve had two out-of-backtest opportunities to test this: 1) trade war anxieties in August 2019, and 2) coronavirus. In both instances, the strategy has performed as expected.
Which I suppose raises an interesting question: how would you “break” this strategy?
The strategy breaks if equities, Treasuries and gold become highly correlated in a period of sharply negative performance. It is difficult to imagine what would cause correlations to change in this way. I tend to believe it would be some kind of end of the world scenario such as nuclear war, suspension of private property ownership, or zombie apocalypse. I’m not sure portfolios can or should be built with such extreme scenarios in mind.
But anything is possible.
And that is why we set a volatility threshold for the portfolio. If pairwise correlations between equities, Treasuries and gold go to one, and the world has not ended, we would almost certainly see a sharp spike to overall portfolio volatility. At 12% portfolio volatility, we would effectively begin to be “stopped out” of risk assets, and would have to add cash to bring the portfolio back below the 12% max volatility threshold. In theory, if the world were truly turned upside down, this would give us the opportunity to re-allocate to other asset classes that are “working” in the new regime.
Most nation states are fairly rational actors. Iran. North Korea. Even Libya under Qaddafi and Iraq under Saddam. These are not the kinds of regimes I would like to live under. But they are not irrational in their foreign policy agendas, either. Don’t mistake what a head of state says for what that state will actually do. Even the most vile, tyrannical regimes are first and foremost concerned with self-preservation. It isn’t in their interest to start wars they can’t win.
What they are solving for is the optimal mix of shenanigans to maximize geopolitical power and domestic prestige while minimizing existential risk. If you are the kind of person who thinks only OTHER countries operate this way, you have a lot to learn about geopolitics.
Sometimes rationality does not prevail in international relations. Or, for idiosyncratic reasons (see WWI), what appears to be a series of rational actions when taken in isolation ultimately leads to an irrational outcome.
Hence, in my view we tend to overestimate the frequency of severe geopolitical shocks and underestimate the severity of the shocks that will inevitably occur. In other words, we are really, really bad at handicapping expected values related to geopolitical risk.
The obvious lesson?
QUIT TRYING TO PREDICT GEOPOLITICAL SHOCKS AND IGNORE ANYONE WHO CLAIMS TO BE ABLE TO PREDICT THEM.
Now, I enjoy reading foreign policy think-pieces as much as the next guy. Maybe even more than the next guy. (Fun Fact: Many moons ago I took the FSOT. At the time, you took the exam and wrote the essays separately. I passed the exam but promptly failed the essay section. In retrospect, I was almost certainly disqualified for ideological unreliability) Anyway, the way to read a foreign policy think-piece is as a scenario–a scenario that might play out in a grand strategy game like Hearts of Iron. This is essentially a speculative activity (the parallels with investment research should be obvious). The fact of the matter is that this stuff is useful for getting you to think about a range of possible futures and strategic options. It is NOT useful for actually predicting the future.
In practical terms, the way you manage geopolitical risk is with hedges and rules-based guardrails. Personally, I’d rather ignore geopolitical risk all together than make predictions based on subjective probabilities.
This is relatively simple, process-over-outcome and OODA loop stuff. There are dozens, if not hundreds of ways to adapt a portfolio to geopolitical risk without predicting anything. And 99.99% of what’s written about geopolitics is trash, anyway. Per the above, that’s kind of the point.
The challenge here isn’t identifying the right tools, or even understanding the tradeoffs they entail.
I deliberately held off on the 12/2019 rebalance for the leveraged permanent portfolio in order to sync the rebalancing checks with calendar month end dates going forward. However, ultimately it was another boring month with the portfolio falling well below its 12% risk target over the one-year lookback period. You can check the one-year trailing data here.
Realized performance for the live strategy since inception is below.
Despite its brief length this is an interesting time period to examine the live strategy as it demonstrates exactly the performance profile you’d hope to see over longer time periods: limited downside capture with solid upside participation. Since the inception of NTSX in September 2018, this portfolio has actually outperformed the S&P 500 on both an absolute and risk-adjusted basis (you can play with the dates in Portfolio Visualizer if you click the above link). I would not expect that type of outperformance going forward. However, over very long time periods I suspect realized performance will compare favorably with a diversified equity portfolio, due to the strong downside protection.
I have updated the (corrected) S&P 500 expected return model for the recent 3Q19 Z.1 data release. The good news: it shows a modest increase in the forward 10-year return estimate, to 4.18%. The bad news: this is almost certainly lower today given how US equities have rallied over the last quarter. (12/19/19 EDIT: I hacked together an estimate as of today and it’s about 3.03%)
Now, I don’t think this model is at all useful as a market timing tool. But it is definitely arguing for lower forward-looking return expectations. This is partly why I’ve implemented the leveraged permanent portfolio with a significant portion of my net personal net worth. Make no mistake: there will come a day to be all-in on equities again. You’ll know it because people will be screaming bloody murder and trumpeting the death of buy-and-hold like they did from 2009 to 2012. (Remember this when your friends and/or financial advisor are pitching you on expensive liquid alternatives some day)
I’ve mentioned before that one of the weaknesses of this model is that it isn’t macro-aware. It doesn’t “know” anything about credit or interest rates. The underlying intuition is simply that as an increasing proportion of assets are “financed” by equity, expected equity returns decrease. In a world of very low or even negative interest rates, it’s possible we’ll see a structural shift in investor preferences for equities. In a regime where interest rates stay very low for a very long time, it makes sense for equity valuations to remain elevated. One should not underestimate the persuasive power of No Good Alternatives (I have been guilty of this, personally). Recall that we tried the whole “normalize interest rates” thing in 2018. We didn’t even get to 3% on Fed funds before the Fed backed off.
The Great Jihad. This is a situation where the transition back to a multi-polar world order, combined with domestic political divisions, results in wars and violent revolutions. Scary, but not worth thinking about all that much from an investment POV. In this future just focus on staying alive. Don’t sweat the markets. In fact, you might as well go all-in, because you’ll be scooping up assets at steep discounts.
Muddle Through. Here everything just kind of works out. Rates and returns stay low, but policymakers effect a “soft landing” and everything works out. In a world where economies can be run with mechanical precision, this can probably be engineered just fine. That’s not the world we live in, however. We live in a world where economic reality must be made politically palatable. Politically, we seem to be headed to a world that is more hostile toward trade, and where there is strong pushback against policymaking elites. I therefore assign a relatively low probability to muddling through.
Stagflation. This is a situation where we have lower economic growth but higher inflation. This is quite frightening from a financial perspective as you have to invest very differently from what is now conventional wisdom to come through stagflation okay. Avoid bonds and cash, as well as equities without pricing power. Real assets are pretty much the only game in town here. Maybe some alts. Personally, I believe we are close to stagflation today. I am one of those loony Inflation Truthers who believes “real” inflation (as experienced by real people in their daily lives) is higher than the CPI numbers trumpeted in the news, because CPI is restrained by things like hedonic adjustments for the improving quality of goods. But I’ll leave the details for macro wonks to fight over.
You will note that I have omitted an inflationary boom from the list. The reason for this is that developed world demographics do not appear to support much of an inflationary boom. What could change this? Well, obviously population growth could suddenly increase. Or, we could start encouraging lots of immigration (not going to happen in the current political climate–and this can’t work for every country in the world simultaneously, anyway). I don’t think either of those things is particularly likely. But, there is always technology. Historically, it has not been great positioning to be short human ingenuity. Maybe Elon gets us to Mars or somewhere else in the solar system and we start colonizing other planets. Who knows.
So anyway, what’s an investor to do?
I am more and more convinced that the average person or institution’s asset allocation should be managed with a trend following and/or volatility targeting overlay (note that this stuff can also work as a risk management tool in more idiosyncratic portfolios). The point here is not to market time (that is impossible to do profitably as far as I’m concerned). The point is to detect regime changes, and to make sure you end up more or less on the right side of them.
Do not be the guy who is short equities for 10 years into a bull market.
Do not be the gal who goes all-in on equities at the top.
Do not be a permabear, or a permabull.
Be biased toward being long, and biased toward bullishness, but with some sense of proportionality and a framework for risk management. As a saver, or an institution that is more or less a saver, you don’t have to catch every market move to make money. You just have to be roundabout, directionally correct about the relationships between economic growth, inflation and valuations.
Identify the regime you’re in. Then make sure you own the right stuff.
No changes this month. The portfolio is still below its risk target on a trailing one-year basis. Again, in an ideal world I’d be leveraging it to 12% annualized volatility. But given the limited toolbox at my disposal, remaining fully invested will have to suffice.
You can geek out with some data here. Below is actual performance data for the strategy since I implemented it.
This post marks the second rebalance check for my leveraged permanent portfolio. Based on some feedback from Twitter, I am making a small tweak to the volatility targeting overlay, and increasing the lookback period from 1 month to 1 year. The intention here is to make the portfolio less sensitive to sharp, short drawdowns in the underlying assets. The purpose of the volatility and trend overlays is not to avoid these types of drawdowns, but rather to adapt to regime changes.
Here is the current portfolio:
On a 1-year lookback this gives us a 9.2% return and 10.33% volatility.* Below the 12% target for the portfolio, despite being fully invested. You can nerd out on the lookback data versus a global 60/40 portfolio and SPY here. In an ideal world, if I had access to the full investment toolbox, I would actually leverage the portfolio to reach the risk target. But, as a small investor, being fully invested will have to suffice.
So, no changes this month.
Below is net performance since inception versus the S&P 500 (my actual allocation differs from target slightly due to transactional frictions, but not in a material way). Again, I wouldn’t normally expect the portfolio to perform this well against a 100% equity allocation over any arbitrary time period. But I think this time period offers an excellent out of sample test of the strategy’s efficacy and in particular its ability to tamp down risk.
* Fun Fact: 10.33% volatility for the portfolio in spite of the fact that individually, each asset in the allocation had a volatility above 12%. This is the magic of true diversification.
My latest note for Epsilon Theory is about possible futures. And Dune.
One of the recurring images in the book is what we in finance know as a probability tree. In the world of Dune, if you are at least a little bit psychic, and you amplify that psychic ability with a generous helping of hallucinogenic “spice,” you can catch a glimpse of the branching probability tree that is the as-yet-unrealized future.
Here in the investment and financial advice businesses, we, too, seem to have reached an evolutionary crossroads. I don’t claim to know exactly what the industry will look like in ten or twenty years. But like Dune‘s protagonist, Paul Atreides, I think I can peer through the haze of a spice trance to glimpse some of the branching possibilities.
I got a lot of great feedback on this note. In reflecting on it, there are a couple points I wish I’d articulated or emphasized more explicitly.
Non-Linearity In Causal Relationships
The imagery of a probability tree used in the note is oversimplified. In reality, discrete paths do not lead inevitably to particular futures with such-and-such probabilities. Rather, events exert a kind of gravity on one another. (See: The Three Body Problem) For example, if MMT were to become the fiscal policy paradigm adopted by our fiscal policymakers, it wouldn’t “automatically” mean that X, Y, and Z would follow as consequences. Rather, the “gravity” of this event would shift the positioning of events in probability space.
A “better,” but more conceptually challenging way of thinking about this is in terms of the light cone used in special and general relativity.
A detailed exploration of the light cone concept is beyond the scope of this post (A Brief History of Time by Stephen Hawking offers a good, in-depth introduction if this topic piques your interest). For simplicity’s sake I’ll rip the relevant principles regarding causality straight from the light cone Wiki:
Because signals and other causal influences cannot travel faster than light (see special relativity), the light cone plays an essential role in defining the concept of causality: for a given event E, the set of events that lie on or inside the past light cone of E would also be the set of all events that could send a signal that would have time to reach E and influence it in some way. For example, at a time ten years before E, if we consider the set of all events in the past light cone of E which occur at that time, the result would be a sphere (2D: disk) with a radius of ten light-years centered on the position where E will occur. So, any point on or inside the sphere could send a signal moving at the speed of light or slower that would have time to influence the event E, while points outside the sphere at that moment would not be able to have any causal influence on E. Likewise, the set of events that lie on or inside the future light cone of E would also be the set of events that could receive a signal sent out from the position and time of E, so the future light cone contains all the events that could potentially be causally influenced by E. Events which lie neither in the past or future light cone of E cannot influence or be influenced by E in relativity.
As events “fire” in space-time, they dynamically shape the geometry of possible futures. Of course, when we think about this in the context of politics, geopolitics, or economics, it is important to acknowledge that events/signals “fire” with different levels of intensity–they create proportionally greater or lesser perturbations in probability space.
If someone were to shoot me dead tomorrow it would not even cause a ripple in global probability space. The event would really only impact probability space in a way that is localized to me and my immediate personal connections. Maybe my local community.
If the President of the United States were to be shot dead, however, the event would “shock” global probability space. A much wider range of possible futures are impacted, distorted, and/or brought into play, and on a much larger scale.
The concept of “blowback” is interesting to consider in this context. The term is used in the intelligence community to describe unintended consequences resulting from covert ops. For example, you arm and train some Islamic fundamentalist religious groups to fight communism during the Cold War. Decades later, the same fundamentalists are using their arms and training to commit terrorist attacks against you. Blowback results from our inability to precisely forecast changes in the geometry of probability space.
We are not Paul Atreides.
And for what it’s worth, if you’ve read Dune: Messiah, you know that even Paul’s prescient vision lets him down in the end.
Some Thoughts On Permabearishness
On a completely unrelated note, I think it’s worth making a few comments on bearishness and permabearishness in particular. If you are not familiar with the term “permabear,” it refers to someone who is constantly calling for the end of the world and therefore refuses to put capital at risk in the equity markets, or is chronically short equities. Sometimes people mistake me for a permabear because I spend a lot of time thinking and writing about economic and investment risks.
There is an important difference between spending a lot of time and energy thinking about risk and refusing to put capital at risk, or being chronically short equities.
Why do permabears exist? Some are cynical charlatans who are permabears because they make a living as permabears. Other permabears get one bearish call right and it leads them down a path of perpetual bearishness as a result of overconfidence in their own prescient vision (there is a Dune reference for everything).
In my view, the core failing of permabears is confirmation bias. They become so myopically focused on justifying their perpetually bearish stance that they lose sight of the fact that you don’t actually make much money (any money?) as a permabear.
The core tenets of my personal investment philosophy these days are the following:
Minimax Regret > Utility Maximization
My affinity for barbell-type portfolios is the result. Rather than create “muddy” blends of fixed income and equity, strive for a convex risk/return profile. Use some method (simple annuity, permanent portfolio, T-bills) to create a kind of “floor” for a portfolio. Then take the remainder of your capital and place your high risk, high reward, high convexity bets. The goal is to create and maintain an asymmetrical risk/reward profile. Skewed to the upside, obviously.
It is not okay to be a permabear. In fact it is dumb to be a permabear.
It is okay to be a nervous bull.
It is okay to view the world through the lens of minimax regret instead of utility maximization (though you must acknowledge potential opportunity costs).
Over the last couple days I’ve had the pleasure of corresponding with David Merkel of The Aleph Blog over differences in our S&P 500 expected returns modes. (Mine was much higher than his). Upon comparing models, I discovered I’d made a huge mistake. I’d essentially included only corporate debt in my calculation, excluding a huge swath of government liabilities from the total figure.
After adjusting my numbers to correct for this, and updating the model, I get a 3.74% expected return for the next 10 years. This is consistent with David’s 3.61% estimate. The small difference that remains is likely down to some minor differences in the time periods we used to estimate our models, as well as the type of S&P 500 return we use in the calculation (I believe David uses the price return and then adjusts for dividends, whereas I simply regress the S&P 500 total return against the “allocation” data).
Previously I’d been referring to my results as “A World of Meh.” I think I’m now comfortable revising that down to “A World of Bleh.” (“Meh” is kind of like an indifferent shrug, while with a “bleh” you are maybe throwing up in your mouth a bit)
I’ll give David the last word here, since I think his take on all this is a nice summary of the quandary investors face these days:
Not knowing what inflation or deflation will be like, it would be difficult to tell whether the bond or stock would be riskier, even if I expected 3.39% from each on average. Given the large debts of our world, I lean to deflation, favoring the bond in this case.
Still, it’s a tough call because with forecast returns being so low, many entities will perversely go for the stocks because it gives them some chance of hitting their overly high return targets. If this is the case, there could be some more room to run for now, but with nasty falls after that. The stock market is a weighing machine ultimately, and it is impossible to change the total returns of the economy. Even if an entity takes more risk, the economy as a whole’s risk profile doesn’t change in the long run.
In the short run it can be different if strongly capitalized entities are taking less risk and and weakly capitalized entities are taking more risk — that’s usually bearish. Vice-versa is usually bullish.
Anyway, give this some thought. Maybe things have to be crazier to put in the top. At least in this situation, bonds and stocks are telling the same story, unlike 1987 or 2000, where bonds were more attractive. Now, alternatives are few.
Here is a question I get all the time, either directly or from colleagues on behalf of clients:
“I bought [INSERT RANDOM STOCK]. It is down 50%. Should I sell it or hold it?”
The answer should be obvious and it is that you sell ASAP. I kind of hate to say it (okay, not really), but if you need to ask someone like me this question, you had no business buying the damn thing in the first place. Note that the fact the stock is down 50% is basically irrelevant here. You should not take a position in a security if you have no framework in place for updating your views on the basis of new information.
There are lots of different ways to play the game. You can immerse yourself in 10-Ks and 10-Qs and try to find great businesses selling at reasonable prices. You can take the view that “price is all that matters” and trend follow or whatever. There are all kinds of sensible strategies for making money.
Hope is not one of them.
If you own something that has halved and the only reason you have for holding it is “gee, I really hate the idea of locking in a loss” then you are in trouble. No one will be able to begin helping you until you first help yourself by exiting the position. Hopefully it is at least in a taxable account and you can write it off. You can think of the slightly milder after-tax loss as discounted tuition.
When people talk about “dumb money” or “the suckers at the poker table” they are talking about hope-based investing.