Member-only story
I Backtested the Simplest Portfolio I Could Think Of: One-Third SPY, One-Third Gold, One-Third Long Treasuries, Rebalanced Every Month
A no-signal portfolio beat SPY on drawdown and Sharpe for 21 years. Then 2022 showed where diversification stops working.
The idea
Most backtests start with a clever signal. This one has none.
The whole strategy is one sentence: hold equal parts SPY, GLD and TLT, and on the last trading day of every month, rebalance back to one-third each. There are no indicators, no parameter grid and no forecasts. It is stocks, gold and long-term Treasuries, kept in balance.
The data starts at GLD’s inception (November 2004), and the first rebalance is on 31 May 2005. Every rebalance pays 0.03% in fees plus 0.03% in slippage on whatever turnover it creates. The code values the portfolio daily, so the drawdown figures come from daily equity rather than month-end snapshots.
The cost of balance
Because the three assets drift apart during the month, the rebalance has real turnover to trade. It averages 2.65% of equity per month. At 0.06% per unit of turnover, that is roughly 0.02% a year in total friction. For this…
