What margin is

What happens when a position opens 

Nothing is bought. The value of the contract does not leave the account, and neither does anything else, because nothing leaves the account at all. What happens is a reclassification. An amount is taken from the pool of money that was free to support further positions, marked as committed to this one, and reported separately from that moment on. The total the account holds is the same figure it was a second before the position opened. Its composition changed. Its size did not.

The committed amount is calculated rather than negotiated, and it is calculated the same way every time. The contract's full value is worked out first, the price of the instrument multiplied by the number of units the contract covers, and a stated percentage of that value is set aside. The percentage is the margin requirement. It is set by the counterparty, it differs by instrument, and it is published in the instrument's contract specifications rather than settled at the moment of opening.

Two consequences follow, and the rest of this module is built on them. The committed amount is a function of the contract's size, so a larger contract commits more of the same account. And it is a fraction of a much larger number, so the contract being tracked, priced and settled is considerably larger than the amount held against it. That second fact is why this module comes before any chart.

Margin is not a payment 

The most durable confusion about margin is that it behaves like a price. A figure is quoted, the amount available falls by exactly that figure, and a position appears. But a payment leaves the payer's hands and belongs to someone else afterwards. Margin does not. It stays in the account, in the account holder's name, and it is released back into the available pool when the position closes. Nothing is charged for it.

• Not a fee. A fee is charged, is gone once charged, and is revenue to the party that charged it. Margin is set aside, remains the account holder's money throughout, and is released on close.
• Not the price of the contract. Nothing is purchased. A contract for difference is an agreement to settle a difference in cash, and it has no purchase price to pay.
• Not a deposit toward a later purchase. No later purchase exists. The contract is settled in cash and extinguished, never delivered.
• Not a maximum loss. This is the one that costs people money, and it has its own section below.

The last two rows are the ones to hold on to. The account is neither poorer nor richer for having opened the position. What changed is how much of it is free, which is a statement about capacity rather than about value.

What margin does not limit 

Margin is the amount set aside. It is not the amount at stake. Profit and loss on a contract for difference are calculated on the contract's full value, so an adverse move is measured against the whole contract and never against the fraction of it that was committed. A loss can therefore consume the margin held against a position completely and carry on past it, and losses are not limited to the amount deposited. A favourable move is measured on exactly the same basis and to exactly the same degree, because it is the identical multiplication with the sign reversed. Nothing in the structure of the calculation treats the two directions differently, and nothing in it refers to the margin at all.

Reading the rows in pairs makes the mechanism visible. The percentage move is a percentage of the contract's value, not of the margin, so the money it produces bears no relationship to the size of the amount set aside. Margin is the condition for holding the contract, not a boundary on what the arithmetic can produce.

Why the amount required changes 

A margin requirement is a percentage, and a percentage on its own is not an amount. What gets set aside is the percentage applied to the contract's value, and the contract's value is the price of the instrument multiplied by the number of units the contract covers. Both of those quantities move, so the amount required moves with them.

Size is the direct one. Doubling the units doubles the contract's value at the same price, and doubling the contract's value doubles the amount set aside at the same requirement. It scales exactly, with no threshold and no rounding, which is why position sizing and margin are one piece of arithmetic read from two ends.

Price is the less obvious one. The contract's value is denominated in money, so a rise in the instrument's price raises that value even though the number of units has not changed. Practice differs on what follows. Many counterparties calculate the requirement once, at the opening price, and leave the amount committed fixed for the life of the position. Others revalue it continuously, so the amount committed drifts as the market moves and a position that has run a long way can hold more of the account than it did on the day it opened. The convention is stated in the instrument's terms rather than inferred, and the two methods diverge most in the conditions where the difference matters.

The percentage itself is not fixed either. Counterparties set a different requirement per instrument, raise it for instruments whose prices travel further, raise it across a weekend or a scheduled announcement, and apply a higher one above stated size bands, so a single instrument can carry several requirements at once. None of these is a market rate. Each is a published term, and the lesson on how a margin requirement is calculated takes them in turn.

Held rather than spent 

Margin is released, not refunded. The distinction is small in words and large in what it explains. A refund implies the money went somewhere and came back. Release means only that a restriction was lifted: the amount was never anywhere other than the account, and closing the position removes the label rather than returning the money. What returns to the available pool is the same money that was committed, and what settles alongside it is the difference the contract produced, which is a separate figure carrying its own sign.

It follows that margin is not a reserve the account can draw on while a position is open. It is committed, and the same money cannot be committed twice. Every further position competes for what is left, so the uncommitted remainder decides whether another position can be opened at all. Several open positions therefore interact through one shared quantity even when they are in unrelated instruments, and a move against one of them changes what the others have room to do.

Building that panel out, one field at a time, is what the rest of this module does: the balance the account started with, the running profit or loss on positions still open, the total set aside across all of them, the account's value once that running figure is counted, what remains free, and the ratio between the value and the amount committed that decides when a position stops being the account holder's to keep open. Each field is arithmetic on the fields before it, and the close out that ends a position is the last line of the same sum rather than a separate event.

Where practitioners disagree 

Two disagreements about margin are genuinely unsettled, and both are usually met stated as fact. The first is what a requirement is for. One tradition treats it as a performance bond: collateral held against a contract that can move against the client, sized to cover a plausible move in the time it would take to close the position. Another treats it as an access threshold: the minimum an account must show to hold a contract of that size, with the closing rules rather than the collateral doing the real work of limiting exposure. Both descriptions fit the mechanics. They diverge on what a requirement is calibrated to, and therefore on what an unusually low requirement is evidence of.

The second is whether a requirement carries information about how far an instrument tends to move. Counterparties do set higher requirements where prices travel further, so a requirement is loosely a statement about observed volatility, and one tradition reads it as one. Another points out that the figure is a commercial and regulatory term, revised infrequently and shaped by capital rules and competition as much as by price behaviour, so reading it as a volatility estimate imports a precision it does not have. The position both sides can defend is narrower than either: a requirement bounds the arithmetic of an account, and it forecasts nothing.