Adding to a losing position, as a risk topic

What the addition changes, and what it does not 

Start with the thing that does not change. The open loss on the original position at the moment a second tranche is added is exactly the loss it was a second earlier. Adding does not reduce it, repair it, offset it or average it away. The money is gone from equity in both cases, and the account statement records the same figure either way. Nothing about the first position is altered by the existence of a second one.

What is created is a second position at a lower price, and the two together carry a weighted average entry that sits between them. That average is the price at which the combined position is flat, so it is true, and checkable, that the price required for the whole holding to return to zero is now closer to where price currently is. This is the entire mechanical case for the practice, and it is worth stating accurately rather than dismissing, because a reader who is told the arithmetic does not work will discover for themselves that it does, and will then discount everything else on the page.

The second change is to size, and size is the multiplier on every point of movement that has not happened yet. A holding of twice the original size records twice the loss for every further point in the adverse direction, and twice the gain for every point in the other. The break even price moved closer. The cost of each further step away from it doubled at the same moment. Those two facts are produced by one action and they point in opposite directions, which is why the practice cannot be assessed by looking at the average entry alone.

The third change is the one that decides the outcome. A second tranche is a new position, and it requires its own margin, calculated on its own notional value. Used margin therefore rises at an instant when equity has not risen at all, because equity fell with the loss that prompted the addition in the first place. Free margin, which is equity less used margin, is consumed twice over: once by the new requirement, and again by every further point of adverse movement across a position that is now larger.

The last two rows are the practice in one pair of lines. In the favourable case the addition is the reason the account is flat rather than down. In the adverse case it is the reason the account is down by a third more than it would otherwise have been, over the same move, on the same instrument, at the same moment. The rows are the same decision read twice.

The distance to close out 

The margin module established the ratio that governs an open account: margin level is equity divided by used margin, expressed as a percentage, and once it falls to a stated level the counterparty closes positions automatically. The mechanics of that process, the order in which positions go and what a margin call is, are set out in the guide on stop out mechanics. What matters here is only the shape of the fraction. Equity sits on top and falls with the loss. Used margin sits underneath and rose when the tranche was added. An addition attacks the ratio from both ends at once, and that is the mechanism this lesson exists to describe.

The calculation runs on the full contract value of the combined position rather than on the margin held against it, so an adverse move is measured against the whole of it, a loss can exhaust the margin posted, 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. On a YAL account, open positions are closed automatically once the margin level reaches 50%, a level set by the counterparty rather than by the trader. It is arithmetic on equity, and it does not consult the reasoning behind any position it closes.

Two rows carry the lesson. The room the account had before the addition and the room it had afterwards are measured from the same price, on the same instrument, at the same moment, and the second is less than half the first. Nothing about the market changed between those two rows. The only thing that changed was the size the account was carrying and the margin held against it, both of which are decided entirely inside the account.

The row where nothing happens is the more instructive one. A second addition, sized to keep pulling the average down at the same rate, requires margin the account no longer has, because the loss that made the addition look attractive is the same loss that consumed the free margin needed to fund it. The requirement grows as the account's capacity to meet it shrinks, and the two curves cross well before price reaches any particular level. That crossing is a structural feature of the arithmetic and not an unlucky outcome.

Why the sequence is attractive 

Behavioural finance has a name for the underlying asymmetry. The disposition effect describes the observed tendency to hold losing positions and close winning ones, and the usual explanation is that a loss remains provisional until the position is closed. Closing crystallises it into a fact, and into a judgement about the decision that opened it. An addition postpones that moment while appearing to address it, which is a combination very few other actions offer.

The practice also has a payoff shape that flatters itself. Prices that move adversely frequently retrace some part of that move, so a sequence of additions resolves without incident on most of the occasions it is attempted, and each of those resolutions produces a small recovery that reads in a record as a decision that worked. The occasions that do not resolve are rare and take a disproportionate share of the account with them. A record dominated by frequent small recoveries and punctuated by an occasional total one is exactly what this shape produces, and it is not evidence about the practice in either direction.

There is a quieter reason the sequence appeals, and it is the one this module is positioned to notice. Every sizing method in the earlier lessons derives size from something decided in advance: a stop distance, a fraction of equity, a measure of volatility, a total across open positions. An addition prompted by an adverse move derives size from the adverse move itself, which appears in none of those calculations as an input. The account's risk fraction has been reset, and it was reset by circumstance rather than by a decision anyone made or wrote down.

The doubling family 

The formal version of the pattern predates markets. A martingale is a betting system in which the stake is doubled after each loss so that a single win recovers the whole sequence, and it is one of the oldest results in probability that the system requires two conditions to work. The first is capital without limit. The second is no ceiling on the stake. Casinos removed the second condition with table limits, and the arithmetic of the first condition was never available to anybody.

A margined trading account fails both conditions in the same place, and the worked block above shows where. Equity is finite, so the first condition is unavailable. Free margin functions as the table limit, and it is not a fixed ceiling but a falling one, because each addition consumes margin and each further adverse point consumes equity. The stake required and the resources available move in opposite directions on the same input. That is why the practice is described in this module as a mechanism with a terminal condition rather than as a technique with a failure rate.

Where practitioners disagree 

The sharpest disagreement is over whether a planned entry in tranches is the same thing at all. One tradition draws a firm line: a position broken into parts before it is opened, where the parts sum to the intended size, where a stop covers the whole position, and where the total amount at risk is fixed before the first tranche exists, is a sizing method that happens to be executed in stages, and the fact that the later tranches fill at worse prices is incidental. An addition decided after the loss has appeared is a different construction, because the total was never bounded and the trigger is the loss itself. The counter argument is that the two are indistinguishable in their effect once price keeps moving, that a plan drafted calmly is extended readily under pressure, and that the presence of a plan changes the account's arithmetic not at all.

The second disagreement concerns whether the practice imported from unleveraged investing survives the journey. Accumulating more of a holding bought outright, at lower prices, with cash, is a recognised approach in long horizon investing, and its defenders point out that it has a property a margined position does not: nobody can close it. The holding can be carried indefinitely, through any drawdown, because no counterparty has a claim that forces the sale. A margined position has a terminal condition written into its arithmetic, and the position is closed by that arithmetic regardless of what the holder believes about the instrument. Opponents of the analogy stop there. Its defenders reply that the difference is one of degree and that a sufficiently small position is not meaningfully margined. The disagreement is real, it turns on how much margin is in use rather than on the behaviour itself, and no version of it is settled.

Why this sits in a risk module 

This lesson describes a mechanism. It does not put forward a tranche size, a sequence, a trigger, a level, a limit or a circumstance, it recommends nothing, and nothing on this page describes a situation in which the approach is advisable. It sits in the module about risk for the same reason the lessons on close outs and loss limits do: it is a thing that happens to accounts, its consequences are arithmetic, and the arithmetic is checkable before the fact rather than only after it.

Five quantities are what actually changed, and every one of them is on the screen of any platform before a second tranche is opened. Average entry. Total size. Used margin. Free margin. The price at which the margin level reaches the close out level. Four of the five moved adversely in the worked block above and the fifth, the average entry, is the one the practice is usually described by. A description that reports one of five moving parts is not a description of the mechanism, and the reason this curriculum treats the topic at length is that the other four are rarely stated.