How CommodityNode Works

Impact Score Calculation

Each node on a CommodityNode Impact Map displays an impact percentage — the estimated price response of that asset to a 10% move in the underlying commodity.

• Level 1 (Direct Impact): Calculated from 5-year rolling regression of asset returns against commodity price changes, using weekly data. We use the beta coefficient scaled to a 10% commodity move.
• Level 2 (Secondary Impact): Derived from supply-chain input-output relationships. If Company A sources 30% of its COGS from a Level 1 commodity, and the commodity moves 10%, we estimate the margin impact after typical hedging and pass-through rates.
• Level 3 (Tertiary Impact): Estimated using industry-level sensitivity models. These capture second-order effects like demand destruction, substitution, and capex cycle changes.
• Level 4 (Macro Factors): Based on historical macro-commodity relationships — e.g., the oil-CPI transmission coefficient, USD-commodity inverse correlation, and freight cost indices.

Impact scores are directional estimates, not precise predictions. They represent the most likely magnitude and direction of response based on historical patterns.

Data Sources

CommodityNode draws on multiple data sources to build each Signal Report:

• Price data: Yahoo Finance API for real-time and historical commodity, equity, and ETF prices
• Fundamental data: SEC filings (10-K, 10-Q) for revenue breakdowns, cost structures, and hedging disclosures
• Industry reports: IEA, USDA, World Bureau of Metal Statistics, LME, and other commodity-specific reporting bodies
• Supply chain data: Corporate supply chain disclosures, Bloomberg supply chain mapping, and industry association data
• Macro data: Federal Reserve (FRED), BLS, and central bank publications for inflation, employment, and monetary policy context

Data Types

CommodityNode uses three types of price data:

Important: when an upstream feed becomes unreliable because of futures rollover, unit discontinuity, frozen prints, or proxy distortion, we may temporarily suppress the displayed day-over-day change instead of showing a misleading figure. We treat missing data as preferable to false precision.

Proxy prices reflect the performance of the proxy asset, not the underlying commodity spot/futures price directly. They may diverge from commodity prices due to company-specific factors, currency effects, or fund composition. All proxy assets are clearly labeled on their respective hub pages.

Correlation Methodology

Correlation values displayed on each node represent the Pearson correlation coefficient between the asset's weekly returns and the commodity's weekly price changes over a rolling 3-year window.

• Range: -1.0 (perfect inverse) to +1.0 (perfect positive)
• Update frequency: Recalculated quarterly using the most recent 156 weeks of data
• Significance: We only display correlations where p-value < 0.05 (95% confidence)
• Lag adjustment: For secondary and tertiary impacts, we test correlations at 0, 1, 2, and 4-week lags and report the strongest statistically significant relationship

Sensitivity Analysis

Sensitivity ratings classify how responsive an asset or sector is to commodity price movements:

• Very High: Impact > 10%, Correlation > 0.80 — asset moves almost in lockstep with the commodity
• High: Impact 5–10%, Correlation 0.60–0.80 — strong, reliable transmission
• Medium: Impact 2–5%, Correlation 0.40–0.60 — meaningful but moderated by other factors
• Low: Impact < 2%, Correlation < 0.40 — commodity is one of many drivers

Sensitivity accounts for hedging ratios (from 10-K disclosures), contractual pass-through mechanisms, and inventory buffer effects that dampen or amplify raw price transmission.

Confidence Levels

Each Signal Report carries a confidence assessment based on data quality and model reliability:

• High: 5+ years of clean price data, well-documented supply chain, multiple confirming data sources, statistically significant correlations (p < 0.01)
• Medium-High: 3–5 years of data, established commodity-equity relationship, some hedging uncertainty
• Medium: 2–3 years of data, or newer commodity/equity with less historical context. Correlations may be regime-dependent.
• Low-Medium: Limited historical data, emerging commodity (e.g., lithium pre-2020), or structural market changes that may invalidate historical patterns

Node Types

Impact Map nodes are classified by their relationship to the commodity:

• Commodity (gold center) — the primary commodity being analyzed
• ETF (purple) — exchange-traded funds providing direct or sector exposure
• Producer (orange) — companies that extract or produce the commodity
• Processor (cyan) — companies that refine or transform the commodity
• Consumer (orange-light) — companies or sectors that consume the commodity as input
• Supplier (green-light) — upstream suppliers to the commodity's production chain
• Substitute (yellow) — alternative commodities or technologies
• Regional (blue-light) — geographic or country-level exposure factors
• Macro (purple-light) — macroeconomic factors like FX, inflation, policy
• Policy (pink) — regulatory, tariff, or government policy factors

Data Sources — Complete List

CommodityNode integrates data from the following sources, updated on the schedules noted:

Correlation Calculation — Technical Detail

Our Pearson correlation coefficients are computed as follows:

1. Data preparation: Weekly closing prices for both the commodity and the related asset, aligned to Friday close. Missing data points are forward-filled (last observation carried forward). Weeks with holidays in either market are excluded.
2. Return calculation: Log returns — ln(P_t/P_t-1) — to normalize for compounding effects and ensure statistical stationarity.
3. Window: Rolling 156-week (3-year) window, recalculated quarterly. We use a 3-year window as a balance between statistical significance (minimum ~100 observations) and recency (capturing recent structural shifts).
4. Significance test: T-test with H₀: ρ = 0. We display correlations only when p-value < 0.05. The t-statistic is computed as t = r√(n-2) / √(1-r²), where r is the sample correlation and n is the number of observations.
5. Lag testing: For Level 2+ nodes, we test correlations at 0, 1, 2, and 4-week lags and report the lag with the strongest statistically significant relationship. This captures delayed transmission effects (e.g., oil → refiner margins have a 1-2 week lag).

Node Graph Layout Logic

Impact maps use D3.js force-directed graph simulation with the following configuration:

• Center force: The commodity node is pinned at the canvas center with a strong gravitational pull (forceCenter).
• Radial force: Nodes are assigned to concentric rings based on their level (1-6). A radial force pushes each node toward its assigned ring radius, creating the ripple-ring visual structure.
• Link force: Edges between connected nodes exert spring forces that keep parent-child relationships visually proximate.
• Collision force: Nodes repel each other at short range to prevent label overlap.
• Charge force: A mild electrostatic repulsion (forceManyBody) ensures even distribution within each ring.
• Node sizing: Node radius is proportional to |impact|, so high-impact nodes are visually larger.
• Color encoding: Node type determines color (see Node Types above). Edge opacity is proportional to |correlation|.

The simulation runs for 300 ticks on page load. Users can drag nodes to adjust layout; dragged nodes are pinned to their new position.

Ripple Chain Methodology

Ripple Chains trace multi-hop impact paths from a commodity shock to downstream effects. Each hop represents a documented transmission mechanism:

• Hop 1: Direct price sensitivity (commodity → producer/consumer)
• Hop 2: Cost pass-through (producer → downstream industry)
• Hop 3: Second-order demand effects (industry → end consumer/macro)
• Hop 4: Macro feedback loops (inflation → policy → commodity)

Ripple Chain strength diminishes with each hop. We typically see 60–80% of the signal transmitted at Hop 1, 30–50% at Hop 2, and 10–25% at Hop 3+.