One aspect that has always fascinated? attracted? fooled? me about financial portfolios are Tactical Asset Allocation strategies. Rebalancing 10.0, a smarter approach to “buy low, sell high”, factors applied to asset classes, call it as you prefer.
Strategic (static) Asset Allocation strategies are cool but there should be a better way to invest than buying stocks with P/Bs higher than the IQ of Stephen Hawking, no? Or buying bonds with negative yields.
After 30 years in the game, I would say no whatever. If there were, it would stop working in 5 years anyway 😉 Maybe it is in front of us but everyone is as lazy as I am to do a proper backtest to validate it? There are well-advertised and profitable (for the publisher) trading rules based on astrology, why no one launch an ETF based on Verdad research?
Somehow, it seems that the market has settled on the “sell signals for a fee” model. WifeyAlpha, AllocateSmartly, PrometheusResearch are the few that I stumbled upon. I understand that launching a financial product is a capital-intensive endeavour, especially compared to a landing page and a MailChimp account, but…the upside potential is also completely different.
To be honest, I never fully understood the financial newsletter business model. If your research adds value, trading based on it should generate more gains than any subscription fee and it is easier to scale. Qualitative pieces might not be a black/white deal: the conclusions I reach and actions I take by reading a newspaper are different than Soros; technical analysis might fall in the same category. Investors might need to aggregate and include different sources/data points in their process. But these are exceptions to the rule, innit?
MacroAlf (Alfonso Peccatiello) started a newsletter research firm and is now launching his hedge fund (with a cringy name if you ask me. Are we at that point, where all the good names have already been taken? I guess so, Latin/Greek mythology has a fixed capacity. I wonder when it would be fine to launch Lannister Capital Ltd…ah ok, there is a trademark issue there…). That’s how it makes sense to me. But if you are stuck at the newsletter level, not so much.
[TBH, WifeyAlpha said that an ETF/HF might be in the pipeline]
Prometheus ETF Portfolio
Prometheus Research published a lean doc to describe their process, let’s have a look.
“We think that by using a rigorous and systematic approach to macroeconomic analysis, we can avoid particularly poor periods of asset returns. Furthermore, using a quantitative approach to assessing risk and return, we can improve the return characteristics of both assets within a beta portfolio and the overall portfolio.“
There are at least two conceptual ways to improve an SAA à la Permanent Portfolio:
- identify in which quadrant we are and overweight the asset(s) that perform better in that scenario (i.e. stocks when econ growth is up and inflation down)
- identify the expected returns for each asset given today’s valuations and rebalance accordingly
In short, win by market timing.
“We think it is important for investors to recognize that achieving these objectives is consistent with a near doubling of equity market return to risk characteristics, which typically have Sharpe Ratios in the 0.2-0.4 range. Furthermore, consistently achieving the higher ends of these return-to-risk characteristics (Sharpe 0.70, Sortino 1.00) would put an investor somewhere amongst the ranks of the top 10% of asset managers of all time.“
Just your friendly reminder that what you lived in the recent decade+ investing only in stocks (maybe only US stocks) is an exception, not the rule. If their stated goal can be achieved their way or “my” way, a different mix of, albeit static, assets, is debatable. That’s why I am interested in this space, I am looking for possible low-hanging fruits.
They have a peculiar way of defining the interrelation between the Risk-Free Rate and risk premia. We all know that (unfortunately) risk premia are not static; what I read from other sources points in a different direction, meaning that risk premia tend to increase when the risk-free rate decreases and vice-versa. Might be just a matter of different windows of observation.
“Every asset is exposed to liquidity risk. The less liquidity in the system, the more the drag on assets. The cash rate and risk premiums are not entirely uncorrelated and often tend to cascade in a similar direction as liquidity tends to dry up in financial markets. As such, falling liquidity can often cause asset prices to underperform cash individually and in aggregate. There is no macroeconomic beta exposure that directly benefits from this environment, and therefore, an active management approach is required where we choose to rotate into cash or short assets (outright or relative to one another) in order for our portfolio to continue to have positive expected returns during periods of tightening liquidity.“
Advanced (?) SAA’s reply to this issue is to include convex, uncorrelated correlated assets like LongVol and Trend Following. None of these approaches is bulletproof, they both have pros and cons.
“To protect our edge in markets, we do not share what goes into our systematic process.“
This is understandable but it also represents the reason why I am sceptical about the whole proposition. LongVol and Trend Following funds are black boxes as well but I find it easier to understand when the process derails from target (and, more relevant, when it does not). I can also diversify “model risk” by investing in different funds in the same category.
Am I fooling myself?
I find the quarterly reports from the TF ETFs I invest informative, why I do not invest the same time to read Prometheus research? The main reason is that TF knowledge is ‘transferable’, what I learn about DBMF I can use it to better understand KMLM. Validating their approaches is easier. Prometheus active management is THEIR strategy, and it looks really complex. LongVol funds like CAOS are equally black box-ish but the ultimate test is simple: did it spike when stocks crashed? CAOS enhanced proposition is that it offers EV>0 but it is just a win-more condition for me, I would be happy even if their performance degrades in line with TAIL.
I can think of many fundamental reasons why TF should work: market trends, investors hate the return distribution of many small losses/few big gains, investors cannot stand tracking error vs their neighbour. On the other side, everyone understands active management, investors love the idea of successful market timing. They would pay big bucks for it (read, 2-and-20) and therefore many are going for it. Even if it is possible, competition is massive.
With an additional issue: it is not written in stone that a particular asset would thrive in a certain quadrant like it did in the past. Commodities and gold do not generate any income by themselves; stocks are different but they largely need other investors to validate your strategy by pushing prices up. While the quadrant-asset relationship is backed by fundamental reasons, it might break temporarily, meaning the market can stay irrational longer than your patience dealing with a black box. The timing overlays described in the Prometheus paper are comprehensible adjustments to prevent this but they also introduce further complexity.
“We can size up and down bets based on where our capital base is: at all-time highs in our capital base, we are comfortable taking more risk, while during drawdowns, we become more conservative.“
There is no clear consensus on dynamic risk calibration. If macro bets work 60% of the time but are independent, being in a streak should be meaningless. As always, we go back to trend vs mean reversion. This is more a risk management control to keep drawdowns within their limits but it does not come for free.
A Different Approach
Another way to approach a TAA strategy is to dynamically optimise for assets’ expected returns, volatility and correlations. This is what Risk Parity funds do. As for market timing, I have always considered this exercise possible in theory and impossible in practice (and good luck backtesting it).
Last week, BankerOnWheels included in their newsletter this report from Verdad that is quite insightful. I will list a few considerations.
The big bounty, as expected, is the ability to forecast short-term expected returns for each asset class. While long-term returns are quite stable, ST ones move a lot. There is no low-hanging fruit here because simple models are not able to produce reliable forecasts.
But…
Over a three-month horizon, volatility is stable.
Does is mean a static portfolio is fine? Kind of.
Volatility scaling has its merits, as described in this paper. It works for risky assets (stocks, credit) but has no impact on other assets (bonds, currencies, commodities). It works because it introduces some momentum and reduces the likelihood of extreme returns. Unfortunately, it also requires significantly more turnover and access to meaningful leverage, in particular for the 60/40 portfolio.
Would we achieve the same results if we apply volatility scaling to portfolios including TF and LongVol? Within Managed Futures, it is not established if vol-targeting adds value or not. Some models use it, others do not.
I am not even sure if you can, or if it even makes sense, to vol-scale a LongVol position. It might well be that including these assets in your portfolio is the poor man alternative to vol-scaling. On the other side, vol-scaling might be a great strategy for Europoors who do not have access to LongVol ETFs (I tried to do a quick test on PortfolioVisualizer but 10 years of data is meaningless).
The great news comes from correlations: they seem to be predictable even with a simple model. This means that regularly ‘refreshing’ the SAA to include the latest correlations forecast might yield better risk-adjusted returns than keeping weights always static.
At the end of the day, I found it funny that the dimension that is harder to forecast, returns, is the one that everyone is more than happy to label as under or overvalued in any type of conversation (even not financial ones!).
What I am reading now:
Follow me on Twitter @nprotasoni