The Players

At 09:58 on a summer morning, somebody in a room in Houston is looking at a screen showing the next hour, deciding whether to leave a unit online at a loss because the afternoon looks tight. Two desks away, somebody else is quoting a calendar 2029 block to a utility. Down the corridor, a scheduler is making sure that what the first two people agreed actually gets nominated to the system operator before the gate closes.

Those three jobs are usually described as one job (power trader), and they are not. This chapter is the cast list: who is on the other side of a power trade, what each of them is trying to do, and why the roles are divided the way they are. It sits at the front of Part Five because the instruments in Chapter 24 make much more sense once it is clear who wants them.

Why the desk is split by tenor

Almost every power trading operation of any size divides along the same line: time.

The reason is Chapter 1. Electricity has to be produced and consumed in the same instant, so the closer a trade sits to delivery, the more it is about the physical machine and the less it is about a view on price. An hour before delivery, the only facts that matter are which units are running, what the weather is doing, and what the operator has just posted. Three years before delivery, none of those are knowable, and the trade is about fuel curves, capacity auctions and credit. The two require different information, different temperaments and different risk limits, so they are staffed separately everywhere.

Two of those seats (day ahead and congestion) exist only because of the market design established earlier. The day-ahead seat exists because there are two settlements (day ahead and real time) rather than one, which is Chapter 9. The congestion seat exists because prices differ by location, which is Chapter 10, and because somebody had to be given something to hedge that with, which is the transmission rights of Chapter 24. A market without nodal pricing has no congestion desk, which is one of the more visible practical differences between the designs in Chapter 15.

The financial side

The third seat, term trading, goes beyond real time and day ahead. Term trading desks can go out months or years into the future. These traders take price views that are less linked to physical power flows and more connected to utility and power consumer hedging. As these trades can go out years, margining and credit lines are involved. Bank term trading desks were once large in power and are now selective. The term trading commodity desks that ran physical power books through the 2000s mostly withdrew after the financial crisis, under a combination of capital rules and a supervisory preference for banks not owning power plants. What remains is structured financing, hedges written against a client’s own asset, and high net worth investor index products. The physical optimisation moved to firms that were happy to own the asset.

Hedge funds arrive in two distinct flavours. Some trade term (months or years ahead) power as one liquid instrument among many, taking directional views on the curve and relative value between hubs, and can leave the sector entirely in a week. Others build genuine specialist capability in congestion and weather, employ meteorologists and grid modellers, and compete directly with utility congestion desks for transmission rights. The second group is small, technically serious, and one of the few places where a power-specific edge is durable, because the underlying data is public but expensive to interpret.

The physical trading houses are a smaller presence in power than in oil, nat gas, ags and metals. A merchant house makes money by owning logistics between a producer and a consumer, however electricity transmission logistics is instant: no ships, no tanks, no blending. Where the houses do participate they do it by owning generation, retail books and flexible assets, which makes them look more like an independent producer than like a trader. Commodities 101 develops the physical merchant model at length; power is the market where that model has the least to grip.

The unglamorous seat that keeps it real

The scheduler, sometimes called a real-time operator or a power marketer depending on the firm, is the role an outsider never hears about and an insider will tell you is the one that ends careers.

Agreeing a price is the easy half of a power trade. It becomes real only once the correct volume, at the correct location, for the correct hours, has been submitted to the correct system operator through the correct interface before that operator’s time gate closes, and confirmed against the counterparty. Deadlines are absolute and measured in minutes. Location codes are long and similar to one another. Daylight saving produces a 23-hour day and a 25-hour day once each per year, and both have to be scheduled correctly.

An oil or nat gas trader reading this has an analogue, and it is the nomination process on a pipeline, with one difference that matters: a missed pipeline nomination usually costs a fee and a delay, while a missed power schedule means the system operator did not know a resource existed and the position settles at whatever the imbalance price turned out to be. Chapter 12 explains why that number can be very large.

The people who own things

Chapter 14 covered the institutions that run the system. This is the commercial half of the same picture, where profit is the primary goal.

The vertically integrated utility owns generation, wires and customers inside an exclusive territory, recovers its costs through a regulated tariff, and trades mainly to manage what it already has rather than to take positions. Roughly a third of US load still sits with one, and most of the world outside the restructured markets does too. Its trading floor exists to buy what the fleet cannot cover and sell what the customers do not need, and its risk limits reflect that a loss is eventually a rate case rather than a bonus.

The independent power producer (merchant fleet) owns generation and no customers. Its entire commercial existence is the spread between a fuel it buys and a power price it sells into, which is Chapter 23 (the spark spread), and it is therefore the most exposed participant in the market and the most active hedger. The merchant fleets assembled after restructuring, and the private equity owners who bought them cheaply and sold them expensively into the load growth of Chapter 25, are the clearest example of a business that is a financial position with turbines attached.

The retailer, or competitive supplier, is the mirror image: customers and no generation. It sells a fixed price to households and businesses and buys a variable one, which is a short position in power as a service business. It is also the participant that fails most often, because the obligation to serve at a fixed price does not go away when the market moves, and a retailer with thin capital and an unhedged book is one cold snap from insolvency. Britain lost dozens of them in 2021 and 2022 for exactly this reason.

Two American-specific labels attach to two types of firms and are easy to confuse. A power marketer is an entity holding market-based rate authority from FERC, which lets it sell wholesale power at negotiated prices rather than under a cost-of-service tariff, and which it can hold while owning no generation and no wires whatever. That is how a bank came to sell electricity: Morgan Stanley Capital Group has held the authority since the mid 1990s and holds it still. The same two words are also used loosely on a trading floor for the scheduling job described in the next section, which is a person rather than a legal status.

A merchant generator is the asset-owning (usually power generation capacity) counterpart, a plant that sells its output into the wholesale market at whatever price clears rather than into a regulated tariff, and it is what makes the independent power producer above a merchant rather than a utility. Both categories are very much alive. FERC still runs the market-based rate programme and publishes the register of who holds the authority, and the large American merchant fleets, Constellation, Vistra, NRG and Talen among them, are the firms whose value the load growth of Chapter 25 transformed.

The buyers who were not supposed to be counterparties

The newest seat at the table belongs to firms that would not have described themselves as energy companies ten years ago.

A hyperscale data centre operator (for cloud storage or AI processing) signing a twenty-year power purchase agreement has taken a position on the power price for two decades, which is a longer view than most generators are willing to hold. Several now employ energy trading teams, hold portfolios of contracts across regions, participate in capacity auctions through their demand response, and negotiate directly with system operators about interconnection terms. Chapter 25 covers what their demand is doing to the system. These data centre hyperscalers became counterparties. Their core business lies outside electricity, and they behave differently on credit, on tenor, and on willingness to pay above market for certainty.

Industrial consumers have always been in this position and are becoming more active in it, for the reasons in Chapter 21. For example, a smelter is effectively a power trader that happens to produce metal.

What every seat is really doing

Underneath the job titles, the participants divide into three groups by what they are trying to achieve, and any counterparty can be placed in one of them.

Some are moving risk which is outside their control. A generator, a retailer, a smelter and a data centre all have exposure that arrives as a by-product of owning something or serving somebody, and their trading exists to reduce it. Some are taking risk deliberately in exchange for expected return, which is the funds and the proprietary books, and their presence is what makes it possible for the first group to lay risk off at all. And some are being paid to make the two meet, which is the market makers and brokers, earning a spread rather than a view.

Electricity has more compulsory participants than almost any other commodity: everybody who generates and everybody who consumes is in the market whether they trade or not, since the alternative to a price is exposure to whatever the imbalance settles at rather than abstention. There is no equivalent of declining to buy oil this month.

Which is the chapter’s point. In most commodity markets, participation is a decision. Here it is a condition of being connected, and the only real choice is how exposure is managed.