Revisiting Beyond 60/40: Five Decades of Risk-Weighted Allocation
In Beyond 60/40 I argued that the classic balanced portfolio rests on an assumption — that stocks and bonds will hedge each other — and that the assumption fails when the macroeconomic regime changes. The argument was built on the post-2005 ETF era, the only window where clean real-price data exists for the three assets needed to test it. Twenty years made the case. Fifty-eight years sharpens it.
This article extends the same backtest to 1968 using a synthetic data construction: London Bullion Market gold from 1968, CRSP-derived total-return equity series back to 1926, and a Treasury long-bond series reconstructed from FRED yields. Each pre-2005 series is anchored multiplicatively to its modern ETF level so the levels are scale-consistent with what an investor sees on a brokerage screen today.
Two findings emerge from the longer record that the shorter window cannot show. The stock-bond hedge that gives 60/40 its appeal is not a property of the assets — it is a feature of one specific regime. And volatility, the one signal that does survive the longer window, can be exploited at every level of construction.
The Headline Result
A portfolio of stocks, bonds, and gold, weighted inversely by trailing volatility and rebalanced monthly, with a single 8% portfolio volatility target overlay, delivered the following over 1968 to today.
The 60/40 returns more in absolute terms; the IVOL portfolio takes half the worst-case loss and runs at half the volatility. The point of this construction is not to maximize CAGR. It is to harvest the underlying risk premium — the compensation paid for bearing systematic risk in equities, duration, and inflation hedging — under conditions where its persistence is most defensible. Everything in the construction serves that single goal.
Stocks and Bonds Are Not Negatively Correlated
This is the result that changes the most when the window extends. The case for 60/40 rests on a single load-bearing claim: bonds rise when stocks fall. Across 58 years, that claim does not hold.
The 120-day rolling correlation between SPY and TLT averaged +0.06 over the full period. The two assets were positively correlated 61% of the time and strongly correlated above +0.3 for 31% of the period. The 1970s and 1980s ran consistently positive at +0.3 to +0.5. The 2022 episode that surprised many investors was a return to the historical baseline.
The hedge that gives the 60/40 portfolio its appeal worked for one generation. The next thirty years will not necessarily resemble the last twenty.
Volatility Is the Signal That Holds
The single quantity that does survive the longer window is volatility persistence. Trailing 63-day realized volatility explains roughly 40% of the variance in forward 21-day realized volatility across the full sample, and the relationship holds in every decade.
This is the empirical foundation that makes a vol-target overlay function. It is not a forecast about returns — only about risk. The strength of the signal becomes obvious when the same rule is applied at the single-asset level. If we target each asset to the same 5% volatility budget, allowing leverage where required so that bonds can compete with equities and gold on equal risk terms, the picture clarifies what the rule actually does.
At equal risk, equities deliver roughly twice the Sharpe of bonds and a third more than gold over 58 years. The Sharpe ranking is exactly what the buy-and-hold ranking would predict, but now visible at the same volatility scale, so the ranking is real and not an artifact of leverage. The TLT line is also instructive: its 2022 drawdown is sharp because the leveraged-up position sized during the multi-decade low-vol regime carried that leverage into the inflation shock. This is exactly why the portfolio version uses risk-weighted allocation across all three rather than levering any single one.
The Construction
For the portfolio version, the rules are simple enough to state in two paragraphs.
Asset weights, set monthly. On the last trading day of each month, measure each asset’s 63-day realized volatility. Weight inversely. An asset with twice the volatility gets half the weight. The portfolio is balanced by risk contribution, not by dollar exposure. Across 58 years, target weights settled near 34% stocks, 35% bonds, 31% gold — close to equal, because all three assets have similar long-run vol once the 1970s gold and the 2022 bond shocks are included.
Vol target overlay, checked daily. If 63-day realized portfolio vol exceeds 8.5%, scale total exposure down so expected forward vol returns to 8.0%. Otherwise stay fully invested. No leverage. Cash earns the prevailing Fed Funds rate. Two triggers, one time-based and one event-based — that is the entire strategy.
Gold Earned Its Place
Across the IVOL portfolio over 58 years, the cumulative dollar contributions broke down as: equities 44%, gold 38%, bonds 15%, cash 3%. Equities and gold contributed almost equally.
This balance is invisible in any backtest starting after 1985. The 1970s stagflation and the 2003–2011 inflation cycle each rewarded gold heavily during periods when both stocks and bonds struggled. Investors whose mental models were formed during the 1985–2020 disinflation regime treat gold as a sentiment trade. The longer record argues that gold is structural — the third leg that absorbs the regime where the other two cannot.
The Real Question Is Drawdown
The question raised by the headline table is not whether 60/40’s higher CAGR is worth pursuing in the abstract. It is whether the path that produces that CAGR is one a real investor can actually walk.
For a retiree drawing capital from the portfolio, a 40% drawdown is not a bookkeeping entry. The withdrawals continue while the assets fall, locking in losses that compound against the remaining capital for the rest of life. Sequence-of-returns risk, which I covered in detail in the decumulation Monte Carlo work, dominates the math when capital is being withdrawn rather than added. A retiree who retired in October 2007 into a 60/40 portfolio and continued the standard 4% withdrawal saw their wealth at the bottom of 2009 sit roughly 50% below where it started, with two more years of withdrawals already taken out. That investor is now fighting an uphill battle they did not have to fight.
The accumulation case is similar but less obvious. The relevant time window for most accumulators is not the textbook 30 years. It is the 5 to 10 years during which they are actually adding meaningful capital — the years when career income peaks, before retirement constraints take over. A 40% drawdown three years into that window is not a drawdown; it is a structural setback that no recovery rally can fully repair, because the contributions made before the drawdown were the largest, and their compounding base was permanently impaired.
Where Higher Returns Come From
The natural objection to a 7% portfolio is that 7% is not enough. The right response is not to lever the same risk premium harder — that just buys a worse version of 60/40’s regime exposure. It is to add return streams that are orthogonal to the underlying risk premium. The seasonal trades I have written about previously — Friday Gold, Turnaround Tuesday, the Turn-of-Month effect, the FOMC drift — exploit calendar-based microstructure features that do not share the regime sensitivity of the base portfolio. Each is sized to its own risk budget and stacks onto the foundation without adding correlation to it.
Limits
The synthetic pre-2005 series is real data, not fabricated. London Bullion Market PM fix for gold; CRSP-style total-return reconstruction for equities, daily back to 1926; Treasury long-bond series from FRED converted to a TLT-equivalent total return. Each splice is anchored to today’s ETF level. The post-2005 numbers stand on real ETF prices and replicate the prior article’s reported figures within rounding.
Two forward caveats. The 4.94% average Fed Funds rate over the historical window will not repeat — forward, the cash-yield contribution shrinks by roughly half. And the 8% volatility target is not optimal in any objective sense; it was chosen to roughly match a balanced retiree’s risk tolerance. The framework matters more than the specific number.
Beyond Passive
The negative correlation that 60/40 depends on is regime-specific, not structural. Building a portfolio that requires this regime to continue is a forecast, whether the investor recognizes it as one or not.
Risk-weighted allocation with a simple volatility target produces a portfolio whose behavior is consistent across the regimes that actually appear in the historical record. It is the cleanest possible expression of the underlying risk premium harvested under conditions of maximum structural confidence. The point is not high returns. The point is to estimate and minimize the variance of that harvest as tightly as possible, so the raw edge of the risk premium comes through with the least possible noise.
That foundation is what makes everything else possible. Real plans in real lives need a base that carries the risk premium without breaking when the regime shifts. The rest is built on top.
Another very well-written and relevant article. It would be quite interesting to see what an overlay of the conditional Friday GLD, TA Tuesday and the EOM SPY/TLT rebalancing trade (described by Kris L) would add to the return and risk metrics over the same extended period. A basic (no Bayesian) implementation with 10% allocation to the overlay (first come/first serve - most overlay positions don’t overlap anyway) is something a retiree could maintain with an Excel sheet and an IB account.