What risk per trade describes

Three numbers sit around one position 

A single open position carries three different money figures, and they are routinely spoken about as though they were one. The notional value, price multiplied by units, is the figure profit and loss is calculated on, and it follows from the contract size. The margin held against it is collateral, and the counterparty sets it. The amount the position would lose if it closed at the level chosen in advance is set by neither: it falls out of the size and the distance to that level, both decided before anything is opened. Of the three, it is the only one a trader determines directly, which is why the discipline organises itself around it.

Risk per trade states that third figure as a fraction of account equity rather than as a money amount. Equity, not balance: the figure that already carries the unrealised result of everything currently open. It is computed before anything happens, which makes it a plan rather than a measurement. The loss eventually recorded is whatever the market delivers, and the two are not the same figure.

Stating it as a fraction is what makes positions comparable. Two positions in different instruments, at different sizes, with exit levels at different distances, can carry the same risk per trade, and two positions of identical size can carry wildly different ones. Lot size on its own says nothing about exposure.

The fixed fraction convention 

The convention that gives the topic its name is fixed fractional position sizing, and the rule it states is very small. The same fraction of current equity is placed at risk on every position, and because equity changes, the money amount changes with it: fixed in the fraction, variable in the money. The arrangement it is conventionally contrasted with is a fixed money amount per position, the same two quantities with the constant on the other side.

Recomputing against current equity has one mechanical consequence, and it happens without anybody deciding anything. After a position closes at a loss, equity is lower, so the money placed at risk on the next one is smaller. After a gain, it is larger. The rule does not change. The base it is applied to does.

The last row is the comparison in one line. A fixed money amount does not hold exposure constant. It lets the fraction drift, upward after a loss and downward after a gain, and a fixed fraction does the reverse. Neither is a claim about results. They are two answers to the question of which quantity moves.

What a sequence does to the arithmetic 

The property the convention is usually defended on only becomes visible across a run of positions. Because each fraction is taken from whatever remains, equity is multiplied by the same number repeatedly rather than reduced by the same money amount repeatedly. Repeated multiplication is not repeated subtraction: each step down is smaller in money than the one before it, and the sequence decays rather than running in a straight line.

Two things in that block are worth reading slowly. The cumulative fall is smaller than the fraction multiplied by the number of losses, and the gap widens as the run lengthens. The same base shifting property produces a larger rise on the way up, which is why the gain rows are not the percentage mirrors of the loss rows. Each step is measured against what the previous step left behind.

The consequence the module on margin established sits on top of that: a fall of a given size requires a larger rise to return to the starting figure, and the requirement grows faster than the fall does. Taken together, the two say what the convention rests on. The size of the fraction determines how quickly a run of losses carries an account into the region where recovery is arithmetically steep.

It is also frequently overstated. A fixed fraction taken from a shrinking base cannot arithmetically reach zero, which is sometimes presented as a form of protection. That is a property of a geometric series, not of an account. Costs are charged in money rather than in fractions, positions cannot be sized below the smallest quantity an instrument trades in, and the loss recorded is not always the loss intended.

Where the convention comes from 

The structure did not originate in trading. It came from the older literature on staking a bankroll across a long run of repeated wagers, where a stake set as a fraction of the current bankroll behaves quite differently from a fixed stake, for the reasons the block above makes visible. Trading writers adopted the form and renamed it. Nothing in the arithmetic is specific to markets, which is both its strength and the source of most arguments about it.

What kept it in use is narrower than the arithmetic. It converts a question that cannot be answered into one that can. How much a position will make or lose is unknowable before the fact. How much it would lose at a level already chosen is a multiplication, available in advance. Three further properties follow from working in fractions rather than money amounts.

• Results across instruments become commensurable: positions that risked the same fraction contribute on the same scale.
• Size leaves the set of things decided while a position is being opened, the moment least insulated from the last result.
• Exposure is stated in the account's own unit, so a run of results reads as one series rather than a list of money amounts.

What the convention does not do 

A fraction describes an intended loss, and an intended loss is not a cap on a realised one. A market that gaps, or that moves faster than an instruction can be filled, does not stop at the level the instruction names, so a position can close worse than specified, and losses on a leveraged contract are not limited to the amount deposited. The fraction is arithmetic performed on an assumption about the exit, and the assumption is the part that fails.

The fraction is also silent on everything outside the single position it describes. Several positions open at once in instruments that move together are not several independent exposures, and a per position fraction contains no information about the total. Frequency is missing in the same way: the same fraction taken many times in a day and taken twice in a month are different exposures over any period longer than a trade.

And no fraction makes a method work. Position sizing governs the size of losses relative to each other and to the account. It has no influence on whether the positions themselves are any good, and a method that loses will lose at every fraction, more slowly at a smaller one. It is a statement about arithmetic and cannot be asked to carry a claim about outcomes.

Where practitioners disagree 

The first argument is about the denominator. Some traditions take the fraction of equity, some of balance, some of a running high water mark, some of closed equity only. Each produces a different money amount from the same fraction, and the differences widen precisely when several positions are open and unrealised results are large. Equity is the honest current figure, which is the case for it. The case against is that it makes a new position's size depend on the floating result of old ones, so a position taken while others are deeply underwater is sized by their condition rather than its own.

The second is about whether the fraction should ever change. One tradition holds it constant on principle, reasoning that a rule which responds to recent results is not a rule but a mood, and that recomputing against current equity already reduces the money at risk after a loss. Another reduces the fraction itself after a run of losses. The standing objection to that is that it shrinks the amount at stake exactly where the arithmetic of recovery demands more of the account, so the drawdown deepens in duration what it saves in depth.

The third is about whether a fraction of equity is the right unit at all. A fixed fraction treats every instrument identically, and instruments do not move identically: the same fraction on a quiet instrument and on a violently moving one produces two positions with very different sensitivities to an ordinary day. Volatility scaled approaches argue the fraction means nothing until it is adjusted by how much the instrument actually moves. Both camps agree on the part that matters here: the fraction alone is not enough information to size a position.

The fourth runs in the opposite direction from the others. A smaller fraction reduces the money at stake and everything else with it, while cost does not shrink in proportion: commission is charged per lot and per side, and the spread is collected on entry whatever the size. Below a certain size a position is dominated by its own cost, and below another the instrument does not trade in a quantity that small at all. A fraction can be too small as well as too large, which is why the argument does not resolve into a number.

Why no figure appears here 

The obvious question after all of this is which fraction, and it has no answer on this page. That is not evasion. A fraction depends on facts a page cannot see: how many positions are typically open at once and how related they are, how often positions are opened, and what a long sequence of losses would mean in circumstances the page knows nothing about. Naming a figure would be a recommendation issued with none of that information, to everyone at once.

What can be said without knowing anything about a reader is everything above. The fractions that recur in the trading literature are small ones, and the second worked block uses two of them as labelled assumptions to show what smaller and larger do to the same sequence. No fraction removes the risk of loss, none guarantees the survival of a sequence of losses, and none of the disagreements above is settled by picking a number instead of understanding what the number does.