Build a risk-parity mix of equities (volatility 18%), bonds (6%) and gold (15%) ignoring correlations: the inverse-volatility weights, the portfolio volatility at zero correlation, and the leverage needed to reach 10% volatility.

Inverse-volatility weights are equities 19.2%, bonds 57.7% and gold 23.1%, each contributing 3.46 points of volatility. At zero correlation the portfolio volatility is that figure times the square root of 3, about 6.0%, so reaching the 10% target needs 1.67 times leverage, borrowing about 67% of capital. Real correlations and funding costs then move the number: under the desk's correlations the risk shares drift apart and the unlevered volatility shifts, and every rupee borrowed costs the funding rate.

Give each sleeve capital in proportion to one over its volatility. One over 18%, one over 6% and one over 15% are 5.56, 16.67 and 6.67, summing to 28.89; divide through and the weights are 19.2%, 57.7% and 23.1%. Each weight times its volatility is then the same 3.46 percentage points, which is what equal risk means when the sleeves are uncorrelated. Three people carrying a sofa share the weight equally by standing where the load is equal, not by standing an equal distance apart; the bonds sleeve stands closest to the heavy end because it is the weakest mover per rupee. Compare an equal-weight book: a third each gives equities 55% of the risk and bonds 6%, which is an equity fund with a bond garnish.

With zero correlation the variances add. Each sleeve contributes 3.46 points, so the portfolio volatility is 3.46 times the square root of 3, about 6.0%. That is lower than every sleeve on its own, including bonds at 6%, because three uncorrelated sources of risk partly cancel; it is also well below the 10% target, which is the normal state of a risk-parity book and the reason the strategy is nearly always levered. Say the limit: zero correlation is an assumption, not a measurement. With a desk estimate of minus 0.2 between equities and bonds, plus 0.1 between equities and gold and plus 0.2 between bonds and gold, the same weights give about 6.2% volatility, and the risk contributions are no longer equal: equities 28%, bonds 31%, gold 41%. True risk parity solves for weights that equalise contributions under the full covariance matrix; inverse volatility is the first cut the head of desk asked for.

Volatility scales with leverage, so the multiple is the target over the unlevered figure: 10% over 6.0% is 1.67 times. For every Rs 100 of capital the book holds Rs 167 of positions and borrows Rs 67, or uses futures with the same economic effect. Leverage is not free. If the desk's funding rate were 7%, borrowing 67% of capital would cost about 4.7% of capital a year before any return, and the desk must confirm its own rate rather than take that figure as given. The 1.67 times also assumes the 6% bond volatility holds; if bond volatility doubles in a rates shock, the bond sleeve that holds 58% of the capital carries far more than a third of the risk and the levered book's volatility jumps with it. That is the failure that hurts risk-parity books: the leverage was sized on the calm sleeve.

Close with the view. The first cut is right as far as it goes: bond-heavy capital, equal risk, about 6% volatility, levered about 1.7 times. What the desk should do before trading it is re-solve with the full covariance matrix, size the leverage on a stressed bond volatility rather than today's 6%, and put the funding cost next to the expected return of each sleeve, because a levered book of low-return assets can be equal in risk and still not worth running.