How financial players keep day-ahead and real-time markets honest.
Every ISO/RTO in the United States operates a two-settlement system: a day-ahead market that clears 24 hours before delivery, and a real-time market that balances supply and demand as electricity actually flows. In theory, prices in both markets should be similar — the day-ahead price for 3 PM tomorrow should approximate what the real-time price actually turns out to be at 3 PM.
In practice, they often diverge. Weather forecasts miss. Generators trip offline. Demand spikes unexpectedly. Solar output exceeds predictions. The gap between day-ahead and real-time prices creates both a problem and an opportunity.
The problem: persistent divergence means the day-ahead market isn't doing its job of price discovery. If day-ahead prices consistently underestimate real-time prices, generators may withhold capacity from the day-ahead market, waiting for higher real-time prices — a strategy that can undermine market efficiency and reliability.
The opportunity: if you can predict the direction of the gap, you can profit from it. This is virtual trading, also called convergence bidding.
Virtual traders submit bids and offers in the day-ahead market without any intention of physically generating or consuming electricity. There are two basic positions:
Virtual supply (INC): A trader offers to "sell" power at a location in the day-ahead market. If cleared, the trader receives the day-ahead price. The position is automatically liquidated in real-time at the real-time price. If the day-ahead price exceeds the real-time price, the trader profits from the difference.
Virtual demand (DEC): A trader bids to "buy" power at a location in the day-ahead market. If cleared, the trader pays the day-ahead price and is credited the real-time price. If the real-time price exceeds the day-ahead price, the trader profits.
In neither case does any physical power change hands. Virtual trades are purely financial positions that settle on the price difference between the two markets.
Virtual trading serves a critical market function: price convergence. Consider what happens when day-ahead prices are systematically too low:
The reverse works too. If day-ahead prices are too high, virtual supply (INCs) pushes them down. Through their profit-seeking activity, virtual traders compress the spread between day-ahead and real-time prices, making the day-ahead market a better predictor of actual conditions.
This convergence has real economic value. When day-ahead prices accurately reflect expected real-time conditions, generators make better commitment decisions (which units to start up), load-serving entities make better procurement decisions, and the overall system operates more efficiently.
Successful virtual traders combine several analytical capabilities:
Weather forecasting: Since weather drives both demand (heating, cooling) and supply (wind, solar), accurate weather models are essential. Some trading firms employ PhD meteorologists and run proprietary weather models that outperform public forecasts.
Load forecasting: Predicting electricity demand based on weather, day of week, holidays, economic activity, and historical patterns. Even small improvements in load forecasting accuracy can generate significant trading profits.
Grid topology analysis: Understanding which transmission constraints are likely to bind, and how that affects locational price differences. Maintenance schedules, planned outages, and new transmission projects all matter.
Statistical models: Machine learning and time-series models that identify systematic patterns in the day-ahead/real-time spread. Many virtual traders rely on algorithms that process gigabytes of historical market data to generate trading signals.
Virtual trading has been controversial since its introduction. Critics raise several concerns:
Market manipulation: In 2013, JPMorgan agreed to pay $410 million to settle FERC allegations that its traders had used virtual bids to manipulate energy markets in California and the Midwest. The traders had allegedly used money-losing virtual positions to trigger favorable payments on related physical positions — a strategy that violated market rules.
Volatility: Some market participants argue that virtual trading amplifies price volatility, particularly at illiquid nodes. When a large virtual position clears at a small bus, it can distort prices in ways that don't reflect physical fundamentals.
Barrier to entry: The analytical infrastructure required for profitable virtual trading — weather models, grid models, data pipelines, execution systems — creates high barriers to entry. Critics argue this concentrates profits among a small number of sophisticated firms.
Defenders counter that virtual trading provides essential price discipline, improves market efficiency, and reduces the ability of physical generators to exercise market power. Without virtual traders patrolling the spread between day-ahead and real-time prices, generators would have more room to manipulate.
Virtual trading volumes are substantial. In PJM, virtual transactions often represent 10-15% of total day-ahead cleared volume. In CAISO, convergence bidding (introduced in 2011) quickly grew to represent a significant share of day-ahead activity.
The profits — and losses — can be significant. A 2019 analysis found that virtual traders in PJM earned aggregate net revenues of approximately $80 million per year, though this was distributed very unevenly across firms. A handful of top traders captured most of the gains; many others lost money consistently.
Virtual trading illustrates a broader truth about electricity market design: financial participants, however controversial, perform a valuable economic function. By putting capital at risk to correct price discrepancies, virtual traders improve the informational efficiency of markets — making prices better signals for investment, operations, and consumption decisions.
The challenge for regulators is maintaining the benefits of financial participation while preventing manipulation — a balance that requires sophisticated market surveillance, clear rules, and vigorous enforcement. FERC's Office of Enforcement monitors trading patterns across all ISOs/RTOs, using data analytics to detect suspicious activity. The cat-and-mouse game between traders and regulators is one of the defining dynamics of modern electricity markets.
The Rise of Trafigura's Power Desk
When Trafigura, the giant commodity trading house, expanded aggressively into U.S. power markets in the mid-2010s, it was part of a broader trend: the financialization of electricity. Trafigura joined firms like Citadel, Jane Street, and Tower Research in applying quantitative trading techniques developed in equity and commodity markets to the peculiarities of electricity.
These firms brought something new to power markets: a relentless focus on data, speed, and statistical rigor. Where traditional energy traders relied on fundamental analysis — checking weather forecasts and generation outage reports — quantitative firms built automated systems that processed millions of data points, identified statistical patterns, and executed trades algorithmically.
The results were transformative. Day-ahead/real-time price convergence improved measurably across ISOs that allowed virtual trading. But the profits flowed increasingly to firms with the deepest pockets and the most sophisticated technology, raising familiar questions about whether financial innovation serves the public interest or merely redistributes wealth to those with informational advantages.
By the early 2020s, virtual trading had become an established — if still debated — feature of American power markets. Whether you viewed virtual traders as essential market plumbers keeping prices honest, or as sophisticated speculators extracting rents from a public good, depended largely on where you sat in the market.