One benchmark, deep liquidity.

Why markets grow deepest when participants meet at a single reference price.

Carry Research6 min read

Liquidity is the quality that makes a market useful. In a liquid market you can buy or sell when you need to, in the size you need, at a price close to the last one. In an illiquid market every trade is a negotiation, and the cost of getting in or out can outweigh the reason for trading at all.

Liquidity does not spread evenly. It gathers. Again and again, in market after market, it gathers around a single reference price. This article looks at why that happens, what it means for the people who use a market, and why it shapes how Carry builds its indices and markets.

What liquidity looks like

Liquidity is easiest to see in three measures.

  • Tightness. The gap between the best price at which someone will buy and the best price at which someone will sell. The narrower the gap, the less it costs to trade.
  • Depth. How much can be traded near the current price before the price moves. A deep market absorbs a large order without lurching.
  • Resilience. How quickly prices recover after a large trade. In a resilient market, new buyers and sellers step in quickly.

For someone hedging a cost, all three matter. A wide gap is a cost paid on every trade. A shallow market makes large hedges expensive to put on and take off. A market that does not recover leaves a hedger exposed when they most need to adjust.

Why liquidity concentrates

Every participant wants to trade where everyone else is trading. A producer looking to sell forward wants the most buyers. A buyer looking to hedge wants the most sellers. A market maker wants the most flow to offset positions against. Each new participant makes the market more attractive to the next, so the busiest market tends to get busier.

A shared reference price is what lets all of them meet. When everyone quotes against the same benchmark, offers and bids can be compared directly, contracts are interchangeable, and positions taken for very different reasons can offset each other.

That last point is easy to overlook. A data center selling future capacity and an AI company buying protection against rising prices have opposite exposures. If both use the same reference, their needs can meet in the same market. If they use different references, each needs someone else to take the other side, and both pay more for it.

The role of market makers

Natural buyers and sellers rarely arrive at the same moment. Market makers fill the gap: they quote prices to buy and to sell at all times, and earn the small difference between the two. They are willing to do this only if they can manage the positions they build up, usually by trading again in the same market or a closely related one.

One deep benchmark makes that possible. A market maker who can always offset a position in a liquid reference market can quote tighter prices, in larger size, in every product linked to it. Spread the same activity across many unrelated references, and each one becomes too thin for anyone to make a market in it with confidence.

The cost of many prices

Imagine the opposite. Every supplier publishes its own index, every contract references a slightly different measure, and every hedge is tailored to a single relationship. Each of those markets is thin. Spreads are wide, trades are slow, and a participant who needs to exit a position may find no one on the other side.

Fragmentation also weakens the prices themselves. A benchmark with little trading around it is easier to dispute and harder to rely on in a contract. Depth and trust reinforce each other: trusted prices attract trading, and trading makes prices more trusted.

What established markets show

Energy and metals markets have many grades, locations and specifications, yet trading gathers around a small number of benchmarks. Crude oil from many fields is priced with reference to a few widely followed benchmarks such as WTI and Brent. Metals produced and consumed all over the world are priced with reference to official exchange prices. Physical deals then adjust for the difference between a particular cargo and the benchmark, rather than inventing a new price each time.

That adjustment is called a differential. A cargo of a heavier crude might be priced at the benchmark minus a few dollars a barrel, reflecting its lower value to refiners. The two parties only need to agree the differential; the benchmark does the rest. Hedging works the same way. The benchmark can be hedged in a deep market, and the smaller differential is managed separately or simply accepted.

A small premium or discount to a deep benchmark is cheaper to agree, and easier to hedge, than a price nobody else uses.

When more than one benchmark survives

Concentration does not always mean a single number. WTI and Brent have coexisted for decades because they describe crude in different places, used by different refiners, and each is deep enough to stand on its own. Where exposures genuinely differ, more than one benchmark can thrive.

What rarely survives is a crowd of near-identical references competing for the same users. In those cases, liquidity tends to settle on one, and the others fade.

Why benchmarks are hard to move

Once liquidity has gathered around a benchmark, it tends to stay. Contracts, systems, risk models and habits are all built around it. The replacement of LIBOR shows how much effort a change takes: regulators, banks and borrowers spent years moving trillions of dollars of contracts to new reference rates, and US dollar LIBOR panels ended only in June 2023, nearly six years after the UK's financial regulator said, in July 2017, that it would no longer require banks to keep LIBOR going after 2021.

For a new market, the lesson cuts both ways. A benchmark chosen early, if it is well designed and well governed, can become the market's reference for a long time. A poorly chosen one can be very costly to replace.

One flagship, differentiated indices

This is the thinking behind Carry's structure. CICI is the flagship benchmark for the price of AI inference, built to be the reference the whole market can meet at. Around it, Carry publishes differentiated indices, such as CRTI for the realized cost of inference, for participants whose exposure is different. Each serves its purpose, and the flagship gives liquidity a place to gather.

The same applies across the value chain. Inference, compute and the infrastructure behind them are linked: the price of GPU time feeds into the cost of inference, and both depend on the data centers and materials that make them possible. Bringing these markets together in one place lets participants manage related exposures side by side.

Global by design

Liquidity also deepens with reach. The demand for intelligence is global, and so is the exposure to its price, yet access to AI-linked markets has so far been concentrated in a few places. A benchmark that participants anywhere can reference, and a market they can reach, brings more buyers and sellers to the same price.

Trusted benchmarks, globally accessible, and one place to meet at them: that is how deep markets have always been built, and how Carry is building the market for intelligence. What grain, metals and oil teach us traces the same pattern through history.

This article is for information only and is not investment advice.

Newsletter

New research, in your inbox.