Supporting · portfolio construction

Risk parity

How should risk be balanced across asset classes?

This portfolio weights S&P 500 e-mini, gold and 10-year Treasury futures so each contributes similar risk, from 2015 to 2025, and compares the result with an equal-weight mix.

$1.61risk parity growth of $1

$2.36equal weight growth of $1

~16%2022 drawdown, both

3futures markets

How it works

Volatility. Estimate each asset’s volatility over a rolling 36-day window.
Inverse-volatility weights. Weight each asset by one over its volatility, so calmer assets such as bonds get more weight.
No look-ahead. Weights use only yesterday’s data; the risk-free rate comes from the 13-week T-bill.
Rebalancing and costs. Daily, weekly or monthly rebalancing, with turnover costs in basis points.

Results

Risk parity versus equal weight growth of $1 and drawdowns — risk_parity_backtest.ipynb · section 3

Lower growth, similar drawdowns

Growth of $1 for risk parity (blue) and equal weight (gold), with drawdowns below.

Both fell about 12% in March 2020 and about 16% in 2022.
2022 hit bonds as hard as stocks, the case risk parity handles worst; equal weight also held more of the gold and stock rally since 2023.

Values are read from the chart.

Daily inverse-volatility weights of S&P 500 e-mini, gold and 10-year Treasury futures — risk_parity_backtest.ipynb · weights

Bonds carry most of the weight

Because Treasury futures are the calmest of the three, they usually hold about 40–80% of the portfolio. Stocks range from roughly 10% to 40%, shrinking after volatility spikes such as early 2020 and growing in calm markets; gold fills the rest.

Values are read from the chart.

How I tested this

Data: Daily S&P 500 e-mini, gold and 10-year Treasury futures, 2015 to 2025.
Out-of-sample test: Weights use a trailing 36-day window and are applied the next day, so every trade uses only past data. The rule has no fitted parameters.
Costs: Zero in the reported run; the backtest can charge costs on turnover, and monthly rebalancing keeps them small.
What this does not show: Three assets and one decade, with no leverage or volatility target, so the result says little about a full multi-asset risk-parity fund.

Takeaways

Balance is not the same as return. Risk parity spreads risk evenly, but in a stock-and-gold bull market it gives up return.
Correlation regimes matter. When bonds and stocks fall together, as in 2022, the diversification it relies on disappears.
Next step. Add a volatility target so the lower-risk mix can be levered to a chosen risk level.

In the repository

View on GitHub: damion-m/Quantitative-Strategies

risk_parity_backtest.ipynbfunc_lib.py