Why adding gold can improve an all equity book

Sunday morning. Friend over chai. Says he is 100 pct equity. Asks why he would add gold to the book if gold compounds slower than equity.

Asked myself the same question the first time I read Markowitz. The honest answer is the one nobody believes until they see the math.

Adding a lower return asset can leave the portfolio better off.

Take the toy. Indian equity at 13 pct expected return and 19 pct annualised vol. Gold at 8 pct return and 15 pct vol. Correlation around 0.15 in calm regimes. Pure equity book sits at 13 / 19. Drop a 20 pct gold sleeve in. The portfolio now sits at 12.0 pct return and 15.9 pct vol.

One pp of return surrendered. Three pp of volatility absorbed. The brochure number went down. The portfolio went up.

The reason is in the variance formula. Portfolio variance is not a weighted average. It carries a cross term. Two times weight one, weight two, vol one, vol two, correlation. When correlation is low the cross term shrinks, and the whole variance shrinks with it. That is the line of math seventy years old that still trips most people on first reading.

Run the mix lower. Seventy thirty lands at 11.5 / 14.7. Sixty forty at 11.0 / 13.7. The minimum variance point in this toy sits at 36 / 64 equity gold, with portfolio vol of 12.6 pct. That number is below pure equity vol of 19, and below pure gold vol of 15. The mix is calmer than either ingredient. That is not a bug in arithmetic. That is the cross term doing its job.

The judgment that follows. Gold compounds slower than equity. So does cash. So do most hedges. If you reject every diversifier on its return number alone, you reject the frontier. The question is not what an asset earns alone. The question is what it does to the portfolio when the rest of the book is bleeding.

The frontier, drawn

Plot every mix of the two assets on one chart, risk across and return up, and you get a curve, not a straight line. The bend is the cross term. Every point on the upper part of the curve is a mix no other mix beats on both return and risk. That upper part is the efficient frontier.

The all gold point sits below and to the right of the 36/64 point. That means a portfolio of only gold is worse on both counts than a mix with 36 percent equity. Nobody would choose it once the curve is drawn.

Run the mix yourself

The calculator below uses the opening's inputs. Change the gold weight and watch return fall in a straight line while volatility falls along a curve.

The same idea on one real Indian year

The toy uses tidy numbers. A real year is messier, and in the year to 30 Sep 2026 gold was not the calm asset at all. An NSE listed gold ETF swung about 26.6 percent a year on its monthly returns, against about 16.9 percent for the Nifty 50. Gold was the more volatile of the two. The correlation of their monthly returns was about 0.20.

Work it through. Equity variance at 80 percent is 0.64 x 16.9 squared, about 182.8. Gold at 20 percent adds 0.04 x 26.6 squared, about 28.3. The cross term is 2 x 0.8 x 0.2 x 16.9 x 26.6 x 0.20, about 28.8. The total is about 239.9, and its square root is 15.5 percent. A sleeve that was far more volatile than equity still lowered the portfolio's swing, because it moved differently. Over the same year, the gold ETF rose about 26.9 percent while the Nifty 50 fell about 8.1 percent.

One year is not a law. Correlation moves, as the diversification cliff shows for crisis weeks. The structure of the arithmetic does not move.

The free Learn module on modern portfolio theory builds the frontier step by step, covers where its inputs go wrong, and ends with a quiz and a certificate. The asset allocation calculator turns a horizon and a risk level into a starting mix.