Day-Ahead vs Real-Time Power Markets

You cannot put electricity in a tank. Every megawatt-hour generated must be consumed at essentially the same instant, or the grid itself becomes unstable. That single physical fact is why power trades through two separate markets instead of one: a day-ahead market, which schedules generation a day in advance based on forecasts, and a real-time market, which corrects that schedule every five to fifteen minutes as actual conditions replace forecasts.

Two markets, one physical constraint

In the day-ahead market, generators and buyers submit bids for each hour of the next day; the grid operator clears an auction and locks in a schedule — which plants run, at what output, at what price — based on forecast demand and forecast renewable output. This gives generators time to plan fuel purchases and buyers price certainty.

But forecasts are never exact. A heatwave arrives hotter than modeled, a wind farm underperforms, a plant trips offline. The real-time market exists to reconcile the gap: every five minutes, the grid operator re-clears based on what is actually happening, and any generator or buyer whose real position differs from its day-ahead schedule is settled at the real-time price for the difference.

Worked example

A gas plant clears the day-ahead auction to produce 100 MWh in hour 14 tomorrow at a day-ahead price of $40/MWh, so it locks in revenue of $100 \times 40 = 4{,}000$, or $4,000 for that hour. When hour 14 arrives, a heatwave has pushed real demand well above forecast, and the real-time price for that hour spikes to $90/MWh.

1. If the plant produces exactly its scheduled 100 MWh, it is paid the day-ahead price of $40/MWh on its schedule — $4,000 — and has no further real-time exposure, because its output matches its schedule exactly.
2. If the plant is unexpectedly forced offline and produces only 60 MWh, it is short 40 MWh against its day-ahead commitment. It must buy that 40 MWh back in the real-time market at $90/MWh to cover the shortfall: $40 \times 90 = 3{,}600$, or $3,600 — a cost that eats directly into the $4,000 it was paid for the schedule.
3. Net outcome of the outage. $4,000 day-ahead revenue minus $3,600 real-time buyback cost leaves only $400 for 60 MWh actually generated, far below what the plant would have earned running as scheduled.

What this means in practice

The day-ahead/real-time spread is where power traders make or lose most of their money, because it isolates forecast error — weather surprises, plant outages, renewable intermittency — from the underlying commodity price itself. A trader can be flat on the level of power prices and still take a large position purely on whether real-time will settle above or below the day-ahead clear. This structure is also why renewable generators, whose output is genuinely hard to forecast a day ahead, face real-time price risk that a predictable gas plant does not, and why battery storage — which can shift output between the two windows — has become a valuable participant in both markets at once.