Every energy trading desk eventually runs into the same question: if a company buys crude oil or natural gas from three different regional benchmarks instead of one, is it actually reducing risk, or just spreading the same risk across a wider surface? The intuitive answer is “diversification always helps.” The mathematics of market connectedness says otherwise – and the difference matters enormously for how procurement desks, refiners, and utilities structure their fuel portfolios.
What “Connectedness” Actually Measures
Connectedness analysis, built on the vector autoregression (VAR) framework popularized by Diebold and Yilmaz, doesn’t just ask whether two markets move together. It quantifies how much of the forecast-error variance in one market’s returns or volatility can be attributed to shocks originating in another. Put simply: when Brent crude jumps, how much of WTI’s subsequent movement is really just Brent’s shock passing through the system, rather than an independent event?
This distinction is the difference between correlation and true risk transmission. Two indices can be highly correlated in price level while transmitting very little actual shock energy to each other – or the reverse. A portfolio manager who only checks correlation coefficients can badly misjudge how insulated a “diversified” fuel book really is.
Crude Oil: A Tightly Wired Global Market
Research comparing Brent, WTI, and Dubai-Oman crude benchmarks across 2018–2019 found total connectedness in returns above 40%, and connectedness in volatility above 60%. In other words, more than half of the volatility behavior in these three markets is shared rather than independent. Brent and WTI in particular showed strong bidirectional spillover, and even Dubai-Oman – despite trading at meaningfully lower volatility – absorbed a substantial share of risk transmitted from the Atlantic Basin benchmarks.
The practical implication: a fuel-procurement portfolio built purely across Brent, WTI, and Dubai-Oman does not diversify away much of the systemic risk. The three markets behave, to a significant degree, as one interconnected pool. Firms holding long positions priced off any single crude benchmark should not assume that switching indices, on its own, materially changes their risk exposure.
Natural Gas: A Fragmented, Regional Story
Natural gas tells almost the opposite story. Comparing the European TTF, North American Henry Hub, and Asian JKM benchmarks, total return connectedness came in under 2% – a striking contrast to crude oil. Volatility connectedness was somewhat higher, around 17%, concentrated mostly in longer-horizon spillovers rather than short-term shocks. This reflects the physical reality of gas markets: pipeline and shipping constraints, regional demand seasonality, and LNG cargo routing still segment these markets in ways that global seaborne crude oil no longer experiences.
That segmentation is exactly what gives a multi-region natural gas portfolio real diversification value – even though, counterintuitively, the risk of the combined natural gas portfolio can still come out higher in absolute terms than a crude oil portfolio, simply because each individual gas benchmark carries more standalone volatility to begin with.
Time Matters: Short-Term vs. Long-Term Spillover
Spectral decomposition of these connectedness indices – separating spillover into short-term (within a trading week), medium-term, and long-term bands – reveals a consistent pattern across both fuel types: return spillovers are dominated by short-term factors, while volatility spillovers are dominated by long-term factors. Risk transmission in price levels happens fast and fades fast; risk transmission in volatility builds and persists over a month or more. A risk desk that only monitors daily return correlations may be blind to a slow-building volatility contagion developing beneath the surface.
From Connectedness to Capital: Measuring the Actual Risk
Connectedness tells you how much risk moves between markets. Value-at-risk (VaR) and expected shortfall, estimated through copula models fitted to each portfolio’s return distribution, tell you how much it could cost. Using Gaussian, t, Clayton, and Gumbel copulas calibrated on historical data, the crude oil portfolio (Brent, WTI, Dubai-Oman) produced a one-day 99% VaR near 4.8% with an expected shortfall around 6.1%. The natural gas portfolio (TTF, Henry Hub, JKM) came in considerably riskier – VaR near 8.4% and expected shortfall near 10.7% – driven by the higher standalone volatility of each regional gas benchmark, even with weaker return connectedness working in the portfolio’s favor.
The takeaway for a trading or procurement desk isn’t a single number – it’s a workflow: measure connectedness to understand structural exposure, decompose it spectrally to know your time horizon, then run copula-based VaR and expected shortfall to translate that structure into a capital number finance can actually use.
Where This Analysis Meets Daily Trading Operations
Understanding connectedness and portfolio risk mathematically is one thing. Operationalizing it – feeding live market curves into a VaR engine, managing multi-commodity position books, and running scenario and stress tests across regional benchmarks in real time – is a different challenge entirely. That’s the layer most energy trading and risk management (ETRM) platforms exist to solve, and it’s why so many risk teams build their practical skills on systems like OpenLink Endur.
If you want to go from understanding fuel portfolio risk on paper to actually configuring, running, and interpreting it inside a production ETRM environment, OpenLink Endur training walks through exactly that – position management, market risk, and curve configuration inside the platform most large energy trading desks already run on.
Practical Takeaways for Portfolio Design
- Don’t assume regional diversification equals risk reduction – crude oil benchmarks are tightly wired globally; the diversification benefit is smaller than intuition suggests.
- Natural gas still rewards geographic diversification – but each regional index brings enough standalone volatility that the combined portfolio can still carry meaningful risk.
- Match your monitoring horizon to the risk type – return shocks are a short-term phenomenon; volatility contagion is a long-term one.
- Translate structure into capital – connectedness analysis identifies exposure; copula-based VaR and expected shortfall put a number on it that risk committees can act on.
- Systems execution is the final mile – the best risk model is only as useful as the platform that operationalizes it across live positions, and that’s where practical ETRM training closes the gap between theory and the trading floor.
