7-Day vs 14-Day vs 30-Day Volatility Horizons: How Forecast Window Shapes Metals Risk Management

Why Forecast Horizon Is a Design Decision, Not a Default

Every volatility model embeds assumptions about the future through its forecast window. A 7-day horizon captures fast-moving event risk. A 30-day horizon smooths through noise to reflect regime-level shifts. The choice of window determines what kind of risk the model can detect, what kind it will miss, and which desk the output actually serves.

In critical minerals, this choice matters more than in deep, liquid equity markets. Metals like cobalt and lithium exhibit clustered volatility driven by supply-side shocks that propagate on different timescales depending on the mineral's geographic concentration, inventory levels, and contract structure. A single export ban can move cobalt within days; the downstream effect on nickel sulphate pricing may take weeks to fully express. Forecasting at only one horizon guarantees blind spots.

The Volterra model generates probability forecasts at all three windows — 7-day, 14-day, and 30-day — precisely because each serves a different consumer within the risk management chain. Understanding what each horizon captures, and where it degrades, is essential for interpreting the signals correctly.

7-Day Horizon: Event-Driven Risk and Tactical Positioning

The 7-day window is the most responsive to breaking information. It is where news flow from sources like GDELT exerts maximum predictive power — article tone shifts, theme spikes, and geographic event clusters often resolve into realized volatility within a week.

For options desks, the 7-day signal maps naturally to weekly expiry management and short-dated gamma exposure. A jump from MODERATE to HIGH at this horizon flags near-term tail risk that may not yet be priced into the vol surface. Systematic traders running mean-reversion or momentum overlays on metals futures use this window for signal confirmation before adjusting position sizing.

The tradeoff: 7-day forecasts are inherently noisier. False positive rates rise because the model must react to incomplete information. In Volterra's walk-forward validation framework, the 7-day horizon shows higher sensitivity but lower specificity compared to longer windows — a pattern consistent with how information decays in commodity markets.

14-Day Horizon: The Risk Manager's Sweet Spot

Two-week forecasts occupy a middle ground that aligns well with standard VaR reporting cycles, margin review windows, and hedging roll schedules. The 14-day horizon filters out much of the day-to-day noise that afflicts the 7-day window while still capturing event-driven volatility before it fully dissipates.

For risk managers at physical trading houses, the 14-day window maps to the typical lead time for adjusting hedge ratios on forward metal purchases. It also corresponds roughly to the periodicity at which supply concentration risk translates into observable price dispersion — a mine disruption in the DRC, for instance, typically propagates through cobalt spot markets over 7 to 14 trading days.

In Volterra's feature importance decomposition, the 14-day model draws more heavily on supply chain structure and HHI-based concentration features relative to the 7-day model, which leans more on real-time news velocity. This makes the 14-day output more stable across regimes but slightly slower to respond to sudden geopolitical breaks.

30-Day Horizon: Regime Detection and Strategic Hedging

The 30-day window is where volatility forecasting shades into regime classification. At this horizon, the model is less concerned with individual events and more with whether the structural environment — inventory trends, trade policy trajectories, seasonal demand patterns — supports elevated or suppressed volatility over the coming month.

Procurement teams and corporate treasury desks managing metals exposure across exchanges typically operate at this cadence. A persistent ELEVATED or HIGH signal at 30 days justifies widening hedge coverage ratios or shifting from spot to forward procurement. It also informs strategic decisions about supplier diversification and inventory buffer sizing.

The 30-day horizon produces the lowest false positive rates in Volterra's backtest but is the slowest to upgrade risk levels when shocks arrive. This is by design — the window is built for strategic awareness, not tactical reaction.

Combining Horizons for Layered Risk Intelligence

The operational value of multi-horizon forecasting lies in divergence, not agreement. When all three horizons align at HIGH, the signal is unambiguous. When the 7-day window spikes to ELEVATED while the 30-day remains LOW, the model is flagging a transient shock that the longer-term structure does not support — a pattern that often accompanies headline-driven selloffs that reverse within days.

The Volterra dataset delivers all three horizons simultaneously for 12 exchange-traded minerals, enabling systematic consumers to build layered risk frameworks calibrated to their own decision cycles. Figures from the Volterra daily pipeline. Full historical backfill available on AWS Data Exchange.

Matching forecast horizon to decision horizon is not optional — it is the difference between a signal that drives action and one that generates noise.