The purpose of my article Seeking Alpha, The 60/40 Portfolio is a Risky Investment, from June 11, 2017, was to point out the high correlation of this strategic allocation with the stock market and the fact that Past performances have offered a false sense of security to investors.. Most readers received the article well, but there was also some reaction and denial of reality.
In 2022, my fears were confirmed as the correlation between stocks and bonds turned positive and 60/40 portfolios generated large losses depending on the bond duration mix. In 2017, but also now, it seemed to me that most investors think there are only two possibilities: a high positive correlation between stocks and bonds, like in the 1990s, or a high negative correlation, like at the beginning of the decade 2000 and early 2010. , and in both cases, the portfolio would win. However, there was one confounding factor: falling bond yields. As soon When the confounding factor due to high inflation and rising rates was removed, the 60/40 portfolio collapsed.
The correlation chart above shows that since 1990, there have been four different stock and bond correlation and performance regimes:
- High positive correlation with the rise of both stocks and bonds in the mid-1990s.
- High negative correlation with stocks falling but bonds rising during the dotcom bear market.
- High negative correlation with stock volatility but rising bonds due to quantitative easing period of early 2010s.
- High positive correlation with falling stocks and bonds during the 2022 inflation period.
The fourth possibility seemed to have a low probability for most investors in the 60/40 portfolio. However, even if the probability of an adverse event is low, given enough time it will occur, and it did.
The 60/40 portfolio in SPY and TLT ETFs plunged just over 23% in 2022 from the previous #4 ranking. Portfolios with shorter duration bond ETFs fell less. Given that before 2002 the largest loss was 8.5% in 2008, a 23.4% loss was a shock to investors. Can it happen again? In the quantitative space, some of us have a heuristic: the worst annual loss and drawdown in the future could be double that.. Therefore, it will not be a surprise to see annual losses and a reduction of around 50%.
Are there better strategic allocations than the 60/40 portfolio?
Investors should be wary of promises of strategic allocations that, in back-testing, have delivered strong risk-adjusted returns; most of them are curve-fitted based on historical data. Curve fitting suffers from the following biases:
- Post-hoc asset selection.
- Optimization of portfolio asset weights.
Today, with the availability of backtesting programs and even artificial intelligence, it is easy to optimize strategic allocations to eliminate large losses in 2022, for example. However, there seems to be a problem that can offer a heuristic to identify optimized strategic, but also tactical, allocations: When 2022 losses are minimized or even eliminated, 2008 losses emerge or increase.
Let’s look at two examples. First of all, the well-known All Seasons Portfolio. This is a portfolio with 40% in TLT, 15% in IEF, 30% in VTI and 7.5% in IAU and DBC ETFs. Sounds like a great assignment.
Everything was going well until the end of 2021. Someone could even leverage this portfolio for some serious alpha, as the largest annual loss was 3.3% and the maximum drawdown was around 15%. But then in 2022 there was even a tailback: the portfolio plummeted by more than 19%! What happened?
It’s easy to see that this portfolio has a large allocation to fixed income, 45% in total. It was natural that the portfolio would suffer a large loss and the maximum drawdown would increase from 15% to 24%. Note that in 2008 the portfolio gained 3.2%, which was an attractive feature heading into 2022.
In retrospect, someone might have realized that the bond allocation was too large and adjusted it as follows: 20% each in the SPY, AGG, GLD, DBC and IYT ETFs.
As shown in the backtest above, the 2022 loss falls in the “comfort zone” below 10%, but the 2008 loss increases to 19.1% and the drawdown increases to 37%. This seemingly “innocent” allocation fits the curve in retrospect. The heuristic about the balance between 2008 and 2022 works well in this case.
In the case of other strategic allocations, curve fitting may not be as evident as in this case. There is a problem with portfolio research and quantitative analysis in general: The goal should be to minimize type I errors (false positives), but the profit motive pushes developers to ignore this rule and optimize.
Investors should be cautious when reading reports about a new strategic allocation solution that appears strong in backtesting; There is nothing free in a post-quantitative easing world, and pressure to reduce deficits, which have fueled easy money in the past, could mean regime change. The era of simple, “do nothing” strategic investing that guaranteed future income may be coming to an end.with tactical investing being the only alternative left other than buy and hold, with all its problems and potential losses due to the possible use of unprofitable timing methods.