Risk parity is a portfolio-allocation method that tries to balance each holding's contribution to portfolio risk. Equal cash weights do not guarantee equal risk: an asset that moves more, or moves with the rest of the portfolio in a particular way, can dominate the result.
What risk parity balances
Risk parity balances estimated risk contributions rather than the dollars assigned to each asset. A common version uses volatility as its risk measure; other formulations use downside measures such as conditional value at risk. The chosen measure is part of the method, not a cosmetic setting (Frontiers study of minimum-risk portfolios; Choi's reward-risk parity paper).
Why the cash weights can look uneven
The allocation depends on how each asset's returns vary and how those returns relate to the others. If one component is more volatile, a risk-parity portfolio may assign it less capital; correlations can change the contribution too. The weights therefore reflect a risk model, not a forecast that the assets will earn the same return.
A useful distinction:
| Allocation question | Equal-weight approach | Risk-parity approach |
|---|---|---|
| What is made equal? | Capital weights | Modelled risk contributions |
| What drives the weights? | Number of holdings | Risk measure and estimated relationships |
| What can drift? | Risk contributions | Contributions when estimates or market relationships change |
This comparison describes the allocation rule; it does not establish that either portfolio will outperform. The arXiv paper studies variants that pair risk contributions with different reward-risk measures, while the Frontiers paper compares risk-parity forms using standard deviation and conditional value at risk.
The model has to be maintained
Risk parity is not a one-time calculation. When estimates change, the weights that once balanced the chosen risk measure may no longer do so. Rebalancing can restore the target, but trading costs and turnover matter: the Frontiers study explicitly includes transaction costs in its minimum-risk portfolio analysis and reports that their effect varies with trading frequency and market conditions. Its estimates come from its own stock and cryptocurrency portfolio samples; they are not a universal cost forecast.
There is a second limit: balancing historical risk does not remove concentration or guarantee diversification when relationships change. The output is only as credible as the return data, risk measure, estimation window and constraints used to build it. A portfolio can satisfy its mathematical target while remaining exposed to a shared market shock.
How you could test the allocation rule
Write down the risk measure, return history, estimation window, constraints and rebalance trigger before comparing portfolios. Then hold the investable universe and test period constant and compare risk parity with a simple equal-weight baseline. Include commissions, spread and turnover in the comparison; report both the target-risk contributions and what they became between rebalances.
Do not tune the choices against the final test period. Change one assumption at a time, and keep a separate period untouched until the rules are fixed. For a related example of testing portfolio interaction rather than trusting a headline, see how correlated trades can concentrate prop-account risk. The same discipline underpins realbacktesting's reproducible backtest approach: define the rule, include costs and make the result checkable.
Frequently asked
Is risk parity the same as equal weighting?
No. Equal weighting assigns the same capital share to each asset; risk parity targets equal contributions under a selected risk model. The resulting capital weights can differ.
Does risk parity guarantee a safer portfolio?
No. It balances estimated contributions, not future losses. Results depend on the risk measure and estimates, and rebalancing introduces costs.
Can a risk-parity portfolio lose money?
Yes. Risk parity is an allocation method, not a return guarantee or a loss limit. A balanced model can still decline when its holdings lose value together.
Takeaway
Risk parity can redistribute a portfolio's measured risk; it cannot make the estimate—or the market—stand still.