Risk Parity with Trend Filters: Combining Equal Risk with Tactical Signals
Traditional portfolio construction weights assets by capital — a 60/40 portfolio puts 60% of dollars in stocks and 40% in bonds. But stocks are roughly three to four times as volatile as bonds, which means a 60/40 portfolio derives approximately 85-90% of its total risk from equities. The "balanced" label is misleading: the portfolio is overwhelmingly dominated by equity risk.
Risk parity addresses this imbalance by weighting assets inversely to their volatility, so that each contributes an equal share of total portfolio risk. The strategies covered here — RP_US and RP_GLOBAL — add a tactical overlay to this risk-balanced foundation: a moving average trend filter that removes assets in downtrends, replacing their allocation with cash. The result is a portfolio that is both risk-balanced and trend-aware.
Inverse Volatility Weighting
The core idea is straightforward: if asset A has an annualized volatility of 15% and asset B has 5%, asset B should receive three times the weight of asset A to contribute equal risk. In mathematical terms, each asset's raw weight is proportional to 1/σi, where σi is the asset's realized volatility. The raw weights are then normalized to sum to 100%.
This approach does not require estimates of expected returns or correlations — only volatility. This is a significant practical advantage because volatility estimates are far more stable and predictable than return or correlation estimates. An asset's volatility tends to persist: if it was volatile last month, it will likely be volatile this month. This persistence makes inverse-volatility weighting robust out-of-sample in a way that mean-variance optimization often is not.
RP_US: US-Focused Risk Parity
| Parameter | Value |
|---|---|
| Universe | SPY, EEM, IEF, DBC, VNQ (5 assets) |
| Risk-Off | Cash |
| Lookback | 10 months (SMA window) |
| Protection | SMA_FILTER (P₀ > SMA of 12 months, daily prices) |
| Allocation | INV_VOL (inverse volatility, normalized) |
Step-by-Step Process
- SMA trend filter: For each of the five assets, check whether the current price is above the 12-month simple moving average computed from daily prices. If an asset is trading below its SMA, it is removed from the portfolio and its allocation goes to cash.
- Compute inverse-volatility weights: For the assets that pass the SMA filter, calculate each asset's realized volatility. Assign a raw weight of 1/σi to each asset. Normalize so the risk-on weights sum to the total non-cash allocation.
- Allocate: Assets above their SMA receive their inverse-volatility weight. Assets below their SMA are replaced with cash. The portfolio is rebalanced monthly.
The Five-Asset Universe
RP_US's five-asset universe is designed to span the major risk factors in investable markets:
| ETF | Asset Class | Typical Volatility | Typical RP Weight |
|---|---|---|---|
| SPY | US Large Cap | ~15% | ~12% |
| EEM | Emerging Markets | ~20% | ~9% |
| IEF | Intermediate Treasuries | ~5% | ~36% |
| DBC | Commodities | ~18% | ~10% |
| VNQ | US REITs | ~18% | ~10% |
Notice how IEF, as the lowest-volatility asset, receives roughly 36% of the allocation under inverse-volatility weighting — far more than its equal-weight share of 20%. This is the risk parity effect in action: the strategy overweights low-volatility assets to ensure they contribute meaningfully to the portfolio's risk budget. Without this adjustment, IEF's contribution to portfolio risk would be negligible despite holding 20% of the capital.
RP_GLOBAL: Global Risk Parity
| Parameter | Value |
|---|---|
| Universe | VT, EEM, AGG, DBC, RWX (5 assets) |
| Risk-Off | Cash |
| Protection | SMA_FILTER (same mechanism as RP_US) |
| Allocation | INV_VOL (inverse volatility, normalized) |
RP_GLOBAL, developed by Andrew Clare, James Seaton, Peter Smith, and Stephen Thomas, applies the same inverse-volatility plus SMA-filter framework to a global universe. The key differences from RP_US are in the ETF selection: VT (total world stock market) replaces SPY, providing global equity exposure including both developed and emerging markets. AGG (aggregate bonds) replaces IEF, capturing the broader US bond market including corporates and mortgage-backed securities. RWX (international REITs) replaces VNQ, adding global real estate exposure.
The global universe provides broader diversification but introduces different dynamics. VT includes emerging market equities, which overlap with EEM — this correlation means the two equity holdings may receive reduced combined weights during periods of high equity correlation. AGG's inclusion of credit-sensitive bonds means it may behave differently from IEF during credit events.
Why Trend Filters Complement Risk Parity
Pure risk parity — without a trend filter — has a well-known vulnerability: it is always fully invested. During a broad-based bear market where every asset class is declining, the portfolio suffers losses proportional to its risk budget, which is distributed across all assets. The SMA trend filter addresses this by removing assets in downtrends and replacing them with cash.
The combination is particularly powerful because the two mechanisms address different risks. Inverse-volatility weighting protects against concentration risk — preventing any single volatile asset from dominating the portfolio. The SMA filter protects against sustained drawdown risk — preventing the portfolio from holding assets in established downtrends.
During the 2022 bear market, for example, both equities and bonds fell below their moving averages. A trend-filtered risk parity strategy would have progressively moved to cash as each asset violated its SMA, eventually holding a portfolio heavily weighted toward cash. A pure risk parity strategy, by contrast, would have remained fully invested throughout the decline, with its heavy bond weighting compounding the damage as both stocks and bonds fell simultaneously.
For strategies that use optimization rather than inverse-volatility weighting, see classical and adaptive asset allocation. For approaches that use momentum ranking instead of trend filters for their tactical overlay, see Mama Bear and Papa Bear.
Risk parity with trend filters represents a middle ground between passive and fully tactical approaches. The inverse-volatility core provides a structurally balanced portfolio that requires no views on expected returns. The trend filter adds a single, well-researched tactical rule that prevents the portfolio from holding assets in sustained decline. Together, they deliver a portfolio that is diversified by risk contribution and adapted to the current trend environment — without the complexity of multi-factor models or optimization-based approaches.