Natural Gas.
Henry Hub natural gas — extreme volatility meets seasonal demand
$
NYMEX Henry Hub (dollars per MMBtu)
52-Week Range
1.614 — 6.954
543 trading days
Ann. Volatility
37.6%
MLE λ=0.900 (N=542)
VaR (95%)
3.93%
FHS
Regime
Calm
P(turb.) = 20%
Price History.
Daily closing prices with Bollinger Band overlay
Regime Detection.
Hamilton (1989) two-state Markov switching model
Calm State
σ = 46.5% ann.
Avg. duration: 3 days
Turbulent State
σ = 131.7% ann.
Avg. duration: 2 days
Regime Timeline
Transition Probabilities
P(calm → calm) = 67.8%
P(calm → turb.) = 32.2%
P(turb. → turb.) = 50.0%
P(turb. → calm) = 50.0%
Model Diagnostics
Converged: Yes
Observations: 542
Log-likelihood: 888.6
Vol ratio (turb/calm): 2.8x
Returns Distribution.
Log return histogram with normal overlay
Mean
0.0799%
Std Dev
5.3930%
Skewness
-2.134
Excess Kurtosis
34.319
Jarque-Bera
27010.43
JB p-value
0.0000
Normal?
No
Observations
542
Why this matters
Negative skewness indicates more frequent sharp drops than rallies — standard VaR understates downside risk. Excess kurtosis (fat tails) means extreme moves occur more often than a normal distribution predicts. Gaussian risk models systematically understate tail risk for this asset.
Volatility.
EWMA volatility (λ = 0.900) — annualized
Seasonality.
Monthly return patterns across available history
| Year | Jan | Feb | Mar | Apr | May | Jun | Jul | Aug | Sep | Oct | Nov | Dec |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2024 | — | — | — | -45.2% | +29.9% | +0.5% | -21.7% | +4.5% | +37.4% | — | +24.2% | +8.0% |
| 2025 | +4.1% | +4.0% | +7.4% | -19.3% | +3.6% | +0.3% | -10.1% | -3.5% | +10.2% | +35.5% | +17.1% | -23.7% |
| 2026 | +31.8% | -34.3% | +0.9% | -7.4% | — | — | — | — | — | — | — | — |
Average Monthly Return
Seasonal Context
The storage cycle dominates: injection season (April–October) builds inventories, withdrawal season (November–March) depletes them. The storage vs. 5-year-average comparison is the most important fundamental indicator. Winter cold snaps can cause 10%+ single-day moves. Summer heat (power generation demand) creates a secondary peak.
Risk Metrics.
Value at Risk, Expected Shortfall, and drawdown analysis
Interpretation: On 95% of trading days, the loss is expected to be smaller than 3.93%. On the worst 5% of days, the average loss (CVaR) is estimated at 5.08%. The 99% VaR captures more extreme tail events at 5.63%.
Estimated from 542 daily returns. Tail risk estimates improve with longer history.
Maximum Drawdown: 62.63%
Related Markets.
Return correlations with economically linked assets
NG vs CL: 167 overlapping return observations
Note: Return correlations are unstable over time and do not imply causation. These pairs are shown because they share economic drivers (e.g., agricultural supply chains, energy complex), not because correlation alone is meaningful. Short-sample correlations (N < 250) should be treated as rough estimates.
Trend Analysis.
Hurst exponent and Bollinger Band bandwidth
Hurst Exponent
Estimated via R/S analysis from 542 return observations
Bollinger Bandwidth
Bandwidth is contracting (squeeze) — volatility is compressed. Whether this resolves up or down is not predictable from bandwidth alone.
About Natural Gas.
Fundamentals, catalysts, and Ethiopian trade relevance
Ethiopian Trade Relevance
While Ethiopia has limited direct natural gas exposure, Henry Hub pricing affects global energy costs, fertilizer prices (ammonia production), and LNG-linked contracts that impact Horn of Africa energy imports. Natural gas volatility is a benchmark for extreme commodity risk management.
Supply & Demand Fundamentals
US production ~100 Bcf/d, primarily from Marcellus/Utica and Permian associated gas. LNG export capacity adds structural demand. Storage levels vs. 5-year average is the key fundamental indicator. Power generation demand (coal-to-gas switching) creates price-sensitive demand. Weather (heating degree days, cooling degree days) is the primary short-term driver.
Key Reports & Catalysts
Quote Convention
NYMEX Henry Hub (dollars per MMBtu)
Unit
$/MMBtu
Trading Days/Year
252