How a power generator with below-average efficiency can still turn a profit – not by predicting prices, but by trading the spread between fuel and electricity.
Energy Trading & Risk Management Insights
Ask most traders how to make money in energy markets and the instinct is to try to predict where prices are headed. It’s the wrong question. Commodity and power markets listed on liquid exchanges are close enough to efficient that no participant can reliably buy fuel cheaper or sell electricity dearer than everyone else, no matter how much research goes into it. Prices already reflect the information available, and whatever new information arrives tends to hit the market independently of what came before. Chasing directional forecasts in a market like that is a losing game over time.
What is not efficient in the same way is the relationship between two related prices. The spread between the cost of fuel and the price of the power generated from it turns out to be a far more tradeable, more stationary signal than either price on its own – and that distinction is the foundation of two practical trading strategies used by power generation businesses: spot-futures spread arbitrage and statistical arbitrage.
The Core Idea: Trade the Spread, Not the Price
A gas-fired power plant has one dominant variable cost: the natural gas it burns. That relationship can be expressed simply – unit generation cost is a linear function of the gas price plus a fixed cost component. The gross margin between the cost of the gas needed to produce a unit of electricity and the price that electricity sells for is what traders call the spark spread.
Because gas and power futures prices are both non-stationary but cointegrated – meaning they wander individually but are anchored to a stable long-run relationship – the spread between them behaves far more predictably than either series alone. That stable long-term relationship is what both trading strategies below are built to exploit.
Strategy One: Spot-Futures Spread Arbitrage
The first strategy compares a generator’s own physical economics against the market’s futures pricing. If the generator’s expected spot spread – what it actually costs the business to convert gas into power – comes in narrower than the equivalent spread currently priced into the futures market, there’s an arbitrage available. The trader takes a long position in gas futures and a short position in power futures, sized to the volume of electricity the plant plans to generate.
Done correctly, this locks in a margin regardless of which direction gas and power prices subsequently move, because the position is hedged against the underlying commodity risk and only captures the mispricing between the futures spread and the operator’s real generation economics.
Strategy Two: Statistical Arbitrage on the Futures Spread
The second approach doesn’t reference the generator’s own cost structure at all – it trades purely off the statistical relationship between the two futures markets. By estimating a long-term equilibrium equation between gas futures and power futures using historical price data (re-estimated on a rolling basis to capture shifts in seasonality and market structure), a trader can identify when today’s futures spread has drifted wider or narrower than the equilibrium relationship predicts.
When the spread is unusually narrow relative to its long-run equilibrium, the position is a long power / short gas trade, closed once the spread reverts and the trade clears at a profit. When the spread is unusually wide, the position flips. Because this is a mean-reversion strategy built on a cointegrating relationship rather than a physical hedge, it can be run at any position size – including leverage well beyond the operator’s actual physical generation volume.
What the Numbers Actually Showed
Tested against a full year of historical Henry Hub natural gas and PJM Western Hub power futures and spot data, spot trading alone – simply buying gas and selling the generated power at market prices – produced a loss, because the assumed plant’s generation costs sat above the average spot power price for the period. That’s the disadvantageous starting position every under-efficient generator actually faces.
Layering in spot-futures arbitrage alone turned that same period from a loss into a solidly profitable one. Adding statistical arbitrage on top – even sized conservatively relative to the physical position – improved results further, and the analysis found that scaling up the statistical arbitrage leverage to roughly three times the physical position was enough, on its own, to offset the entire spot trading loss. Profit and maximum drawdown both scaled proportionally with leverage, underscoring that this isn’t a free lunch: bigger statistical arbitrage positions mean bigger unrealized losses to manage before the spread reverts, not just bigger eventual gains.
The Real Lesson for Trading Desks
The headline takeaway isn’t the specific profit figures – it’s the structural insight underneath them. A power generation company with genuinely inferior conversion efficiency, competing in a market where it cannot win on physical economics alone, can still turn a structurally disadvantaged position into a profitable one by systematically trading the statistical relationship between its input and output markets. That requires disciplined cointegration modeling, monthly re-estimation of the equilibrium relationship, and, critically, rigorous position sizing, since leverage that improves statistical arbitrage profit also directly increases unrealized loss exposure during the holding period.
From Strategy Design to Live Execution
Modeling a spread strategy in a spreadsheet is one exercise. Running it live – tracking rolling cointegration estimates, managing simultaneous long/short futures and spot positions across multiple commodities, and monitoring unrealized P&L in real time against risk limits – is a different job entirely, and it’s exactly what a production ETRM platform is built to handle.
If your team is moving from strategy theory to actually configuring and running these positions inside a live trading system, Apollo Skill Labs’ Endur training covers the practical side: position management, deal capture, and market risk workflows inside the platform most energy trading desks already rely on day to day.
Key Takeaways
- Don’t trade the price, trade the spread – individual fuel and power prices are close to unpredictable in efficient markets, but the spread between related, cointegrated prices is a far more stable, tradeable signal.
- Spot-futures arbitrage is a hedge against your own cost structure – it locks in the gap between a generator’s real economics and what the futures market is currently pricing.
- Statistical arbitrage is a pure mean-reversion play – it trades deviations from a rolling long-term equilibrium relationship, independent of the operator’s physical position.
- Combining both strategies compounds the benefit – historical simulation showed spot-futures arbitrage alone could turn a loss into a profit, and statistical arbitrage layered on top improved results further.
- Leverage cuts both ways – larger statistical arbitrage positions scale both profits and maximum unrealized losses, so position sizing has to be matched to the capital a business can actually risk.
0 Comments