Day-ahead planning meets real-time reality.
Electricity has a unique property that makes market design exceptionally difficult: it cannot be stored at scale (or at least, it couldn't be until recently). Every megawatt-hour generated must be consumed the instant it's produced. The grid operates at a precisely controlled frequency — 60 Hz in North America — and any deviation signals an imbalance between supply and demand.
This means the market can't simply let prices fluctuate and wait for the normal adjustment mechanisms of commodity markets. A generator that decides to wait for higher prices can't hold its megawatt-hours in a warehouse. A load that delays consumption until tomorrow can't always do so. The grid operator must balance supply and demand continuously, in real time, with no tolerance for error.
The solution is a two-settlement system. The day-ahead market runs the evening before actual operations: generators submit offers for each hour of the next day, load-serving entities submit demand bids, and the market-clearing algorithm solves for the generation schedule and prices that minimize total cost while respecting all transmission constraints.
The day-ahead market is a forward market — it creates financially binding commitments. If a generator clears in the day-ahead market for 100 MW at hour 14, it receives the day-ahead price for those 100 MW regardless of what happens in real time. This gives generators and load a predictable financial baseline — the ability to hedge against real-time price volatility.
Real-time operations deviate from the day-ahead schedule for countless reasons: generators trip offline, demand is higher or lower than forecast, transmission constraints emerge. The real-time market runs continuously — every 5 minutes in most RTOs — to re-dispatch generators and balance the system.
Participants who deviate from their day-ahead schedule are settled at the real-time price, not the day-ahead price. A generator that was committed to produce 100 MW day-ahead but only produces 80 MW real-time "buys back" the missing 20 MW at the real-time price. A load that scheduled 200 MW day-ahead but actually consumes 220 MW pays the real-time price for the extra 20 MW.
This two-settlement design creates a sophisticated financial architecture. Day-ahead prices provide the basis for long-term contracts. The spread between day-ahead and real-time prices creates opportunities — and risks — for traders who forecast them accurately. Virtual bids (financial-only bids with no physical delivery) allow market participants to arbitrage the day-ahead/real-time spread, improving price convergence.
The system is complex. It requires significant operational discipline from market participants, sophisticated risk management, and constant regulatory oversight. But it has proven remarkably robust at balancing the need for physical reliability with the goals of competitive pricing.
PJM, June 1, 2000. PJM's day-ahead energy market opened as the first fully operational large-scale two-settlement system in the United States. In the market's early months, day-ahead prices frequently diverged from real-time prices by wide margins as the system calibrated. Financial traders quickly noticed. Virtual increment bids — selling day-ahead, buying back in real time — and decrement bids — buying day-ahead, selling back in real time — began flowing through the market as participants arbitraged the gap. The spread between day-ahead and real-time prices narrowed measurably within the first year. PJM's 2000 State of the Market Report documented the convergence and treated it as evidence the design was working: financial participants, pursuing profit, were simultaneously helping the day-ahead market forecast real-time conditions. The pattern held, and PJM's two-settlement model became the architecture every subsequent U.S. market adopted.
PJM — State of the Market Report 2000 Iowa State — "Experience with PJM Market Operation, System Design, and Architecture"