Who is actually in the market

Two purposes, one price 

The importer's problem is not the euro. The importer's problem is that a bill denominated in euros has to be settled out of revenue earned in dirhams, and the size of that bill in dirhams is currently unknown. Buying the euros forward removes the unknown. Once the trade is on, the importer largely stops watching the rate, because the outcome has been fixed and watching it changes nothing.

The dealer has a different problem, which is inventory. Having taken the other side, the dealer now holds a position nobody asked for, and the work of the next few minutes is finding someone who wants it at a price that leaves something behind for having stood there. The dealer is not forecasting the euro. The dealer is managing what has just arrived on the book.

Neither of them is the market's intended customer and neither is an interloper. The price is the level at which those two unrelated problems could be settled against each other at that moment, and every quotation on every screen is that same thing repeated at scale. A market has no opinion of its own, and no member whose presence is more legitimate than anyone else's.

The categories, and what each is there for 

Sorting participants into categories is useful in roughly the way sorting traffic is useful. It is approximate, the boundaries leak, and it still explains most of what is observed. The conventional grouping runs as follows, and the second half of each line is the part that matters, because the purpose is what predicts the behaviour.

• Central banks, which act to implement monetary policy and manage official reserves. They are the only participants pursuing a policy objective rather than a commercial one, and the only ones who are sometimes indifferent to the price they get.
• Commercial and investment banks, which quote prices to their own clients, settle those clients' commercial payments, and run positions for their own account on separate desks under separate mandates.
• Asset managers, pension funds and insurers, which hold assets denominated in foreign currencies and deal in currency mostly as a consequence of decisions taken in other markets entirely.
• Corporates, which buy and sell currency because they buy and sell goods and services. Their activity follows an invoicing calendar rather than anything happening on a chart.
• Hedge funds and proprietary trading firms, whose positions exist for their own sake and whose holding periods run from several months to a fraction of a second.
• Non-bank electronic market makers, which quote continuously and in volume, hold what they accumulate very briefly, and now do much of the quoting in widely traded pairs.
• Brokers, which stand between retail clients and the wholesale market and are the counterparty to the contracts their clients hold.
• Retail clients, individuals dealing for their own account, in sizes that are small relative to every other line on this list.

A hedge and a position are not the same trade 

The most useful division in that list is not by size or by sophistication. It is by whether the exposure existed before the trade did.

A hedger arrives carrying an exposure created by something outside the market: an invoice, a foreign subsidiary, a shipment of copper already contracted for. The trade offsets that exposure. A speculator arrives carrying none, and the trade creates one that exists nowhere else. The instrument used can be identical. The reason it is on is not, and the reason governs how it is managed.

That single difference produces most of the observable differences in behaviour. A hedger works to a deadline set by a commercial calendar rather than by a market, is largely indifferent to the direction the rate then takes, and does not close a hedge early because it has moved favourably, since the favourable move is being paid for elsewhere in the same business. A speculator has no external deadline and no offsetting exposure. Hedging flow is therefore predictable in timing and insensitive to price, and speculative flow is the opposite of both.

The two cases together are the point of the block, and neither is a good or a bad outcome. In the first the hedge records a credit and the goods cost more. In the second it records a debit and the goods cost less. The total outlay is the same figure in both. A hedge does not produce a gain: it removes a variable, and it removes it in both directions with equal force. A hedger who judges a hedge by its own profit and loss, in isolation from the exposure it was placed against, has misread what the instrument is for.

Who quotes and who takes 

Cutting the same population a second way separates the participants who show prices from those who accept them. A market maker quotes two prices at once, one at which it will buy and one at which it will sell, and stands ready to deal on either at the moment it is asked. It does not need a view on direction to do that. Its business is the difference between the two prices and the management of whatever it ends up holding.

The obvious risk in that business is one sided flow. A maker that keeps being asked to buy, because everyone is selling, ends up holding a growing position it never wanted in a market moving against it. The prices it shows move to reflect that, and they move further apart the less certain it is of finding the other side. A quotation widening during a fast move is not a penalty applied to anybody. It is the cost of the risk of standing there at all, and participants under no obligation to stand there simply stop quoting.

In broker terminology, the firms that quote to a broker are its liquidity providers, and their prices are what a broker's clients ultimately deal against. How several such quotations are combined into the single price on a screen is the subject of a later lesson in this module, as is what happens to an order once it leaves the client's platform.

The wholesale layer where those firms deal with each other is still called the interbank market, though the name has outlived its accuracy: non-bank electronic firms now quote alongside banks in the most heavily traded pairs. There is no building, no bell and no central order book. What defines the layer is a network of bilateral credit agreements about who may deal with whom, and in what size.

Where retail activity sits in all of this 

Retail participation is the smallest activity on the list by a considerable distance, and it is the only one that is entirely discretionary. A corporate treasurer hedging a payable is doing a job that has to be done. A pension fund holding foreign assets carries currency exposure whether it wants it or not. An individual has no exposure at all until one is created.

This page states no figure for the proportion, and the reason is worth stating rather than hiding. Central banks and industry bodies publish periodic surveys of turnover by counterparty type, their definitions differ, the numbers are revised after publication, and a statistic quoted inside a lesson goes on being displayed long after it stopped being true. What matters here is the order of magnitude, and that can be shown as arithmetic over assumed order sizes instead.

The arithmetic is not an argument that retail activity is unimportant, and it says nothing whatever about anyone's results. It explains a structural fact the rest of this module depends on: retail orders are absorbed by the market rather than moving it. An order of ordinary retail size arrives at a price set by other flow and leaves it where it found it, which is why the interesting question is not what the order does to the market but what the market does to the order.

Why the mix changes through the day 

None of these participants is present continuously. Corporate treasuries and asset manager dealing desks are staffed during their own business hours, and their orders arrive then and not otherwise. Bank market making follows the same clock, because a desk quotes when its clients are awake. As one region's working day ends and another begins, the population quoting a given pair changes composition entirely, and there are hours in which the firms quoting continuously are almost entirely automated makers whose holding periods are measured in fractions of a second.

Two further concentrations are worth recognising. Scheduled benchmark fixings pull a large quantity of orders into a short window, because index funds and corporate treasuries are frequently required by their own mandates to deal at a published reference rate rather than at whatever is showing. Scheduled announcements do the reverse: participants under no obligation to quote often withdraw beforehand, so the population thins immediately before a release and repopulates afterwards.

What the composition does to a quotation 

How many participants are willing to deal at a given moment, and how varied their reasons are, is what produces the properties the next lessons in this module take apart. When many participants with unrelated purposes are present, an order of ordinary size meets a counterpart close to the price displayed, because someone in the crowd wanted the other side for reasons of their own. When the mix narrows to a few firms running similar models and seeing the same flow, the same order meets fewer willing counterparts, and the price it meets can sit further from the one displayed a moment earlier.

That is the mechanism behind an observation which otherwise looks arbitrary: the quotation for one instrument can be tight and steady for most of the day and wide and unsteady for part of it, with nothing having changed about the instrument. Nothing did change about the instrument. The population changed. Everything the rest of this module examines, where a price comes from, how liquidity is aggregated, how an order is routed and how the resulting execution is measured, sits downstream of that one fact.

Where practitioners disagree 

Two arguments about the market's population are genuinely unsettled. The first is whether the visible market is the market at all. A large bank matches many of its clients' orders against each other internally, and those trades never reach an external venue or appear in a public quotation. One tradition treats internalisation as the market functioning efficiently, since matching two clients directly is cheaper for both than sending each out to a wholesale venue. Another holds that a price formed on whatever fraction of flow does reach the outside is a partial picture of supply and demand, and that how partial it is varies with how much is being internalised at the time. Both are accurate about part of the same arrangement, which is why the argument persists rather than resolving.

The second concerns retail positioning. Some brokers publish the aggregate proportion of their own clients holding long against short positions, and a contrarian tradition treats a heavily one sided reading as informative about the opposite direction. The objections are practical rather than theoretical: each dataset covers one firm's clients rather than the market, publishers define an open position differently, the samples are not comparable with one another, and the claim has never been demonstrated on evidence available outside the firms holding the data. This curriculum reports the convention because a reader will meet it. It does not adopt it, and it puts forward no reading of any such figure.