Managing It: Hold, Trail, or Invalidate

Three legal moves and what each costs

Holding to the written plan means the target and the stop stay where they were written, and the trade ends when one of them is reached. It is the simplest rule and the easiest to audit, it captures the full payoff when the target is hit, and its cost is that it gives back open profit on the way to a stop: a trade that reaches 2.5R and then reverses to the original stop is a normal outcome rather than a management failure. Trailing the stop means moving it behind the trade by a stated rule — a fixed R given back from the highest favourable price, or a structure-based trail at the last higher low from TR7. It converts open profit into protected profit at the cost of being stopped out of trades that would have continued, and its arithmetic is entirely determined by how much room the trail leaves: a tight trail banks more small wins and loses the large ones, a loose trail does the reverse. Exiting on invalidation is the third move, and it is the only one that is not about profit at all. If the retest from TR13 failed, or the last higher low broke, the reason the position exists has gone and the position should not. This is the move that has to be pre-written, because in the moment the structure breaks, the alternative explanations are plentiful — the market is volatile, the level is being shaken out, the thesis needs more time — and none of them is checkable while the loss is live. Written first, it is a rule; decided live, it is a story. What makes all three work is that they are decided before the trade, and what makes them fail is scale and story. Adding to a loser, averaging down, doubling after a stop-out, holding an event position on a widened stop — each is a fourth move that is not in the plan, and each of them increases size at the moment the evidence has deteriorated. The rule of thumb that follows is blunt: if a management action increases risk after the trade has moved against you, it is not management. The same trade under three rules — Hold to plan: stop 1R below, target 4R above: Ends at the stop or the target; gives back open profit on reversal · Trail 1R from the highest favourable price: Protects after 2R, exits on a 1R give-back — truncates the tail ← · Structure trail at the last higher low: Stays in while the sequence holds; exits when it breaks · Add to a losing position: Not a management move: size increases as evidence deteriorates ← A trailing rule needs a starting condition. The common version is to trail only once the trade is 1R or 2R in profit, because a trail active from the entry is a tighter stop in disguise and will be hit by the noise from TR3.

Scaling, and why partial exits are an arithmetic decision

Scaling out — selling part of the position at 1R, part at the target — is popular because it makes the experience of holding easier, and that is a real benefit with a measurable cost. Expectancy is the sum of each share’s outcome, so taking a third off at 1R caps a third of the trade’s payoff at 1R; on a setup whose profits come from the occasional 6R winner, that can turn a positive expectancy into a negative one. The honest version is to compute both: the full-size expectancy and the scaled expectancy, using the same probabilities, and check whether the improvement in behaviour is worth the arithmetic. Sometimes it is — an investor who can hold a scaled position through volatility but not a full one is better off with the scaled one, because the comparison is against their own alternatives rather than against an ideal. There is a second scaling question that gets less attention and matters more: whether the original size was right. Scaling out is often a workaround for a position that was too large to hold comfortably, which is a sizing problem from TR3 dressed as a management strategy. If the plan repeatedly produces trades that can only be held in thirds, the diagnosis is the risk fraction rather than the exit rule, and the fix — smaller size, same rule — is cheaper than the workaround. The last piece is time. A plan written for a swing trade has an expected holding period, and a position that has been open four times longer than planned is not being patient, it is being unmanaged: the setup that justified the entry has been replaced by a hope that the market will return to the entry. Time stops — “if the setup has not resolved in ten sessions, the idea is wrong” — belong in the plan for exactly this reason, and they are the one exit condition that is about the thesis rather than the price. • Hold to plan: full payoff when hit, gives back open profit on reversal. • Trail by a stated rule: protects profit, truncates the tail, needs a starting condition. • Invalidate on structure: the only move that is about the thesis rather than the profit. • Scaling out is arithmetic — compute both expectancies rather than assuming it helps. • If every trade can only be held in thirds, the size is the problem, not the exit rule. • A time stop is what separates patience from an unmanaged position. The most expensive management habit is the one that feels most rational: deciding, while a position is open, that the target should be extended because the news is good. That is not a plan, it is a new trade with the same position size — and if it is worth taking, it is worth writing as an addition with its own stop and its own risk budget rather than smuggling itself in as management.

The two exits nobody writes down

Hold, trail, invalidate — the three moves are about what price does. Two other exits decide the same position and get written down far less often. The first is a **time stop**: the thesis had an expected timeline, and if the timeline passes without the move, the reason for the trade has quietly expired even though no price level was touched. A breakout that has not gone anywhere in three weeks is not a breakout you are still right about; it is capital sitting in a spot with no edge. The second is an **invalidation exit** — leaving because the condition that justified the position is no longer true, at any price. If the trade existed because earnings would beat, and the guidance is withdrawn, the position is wrong at a profit as surely as at a loss. Profit is a reason to reconsider the size, never a reason to keep a thesis that has been falsified. Research on the disposition effect shows the opposite tendency is the default: people close winners early and hold losers, which means the invalidation exit is exactly the one that requires a written rule. Both exits share a property the price-based ones lack: they fire on *new information* rather than on a level. That makes them harder to automate and more valuable, because the whole point of a plan is to act on reasons rather than on feelings about the current price. A position with no time stop can be held indefinitely on the strength of having once been a good idea; a position with no invalidation exit can be held after the idea is gone. Neither is a price decision, so neither appears on a chart.

Adding to a winner, and the arithmetic that keeps it legal

The three legal moves cover what to do with a position that is working, and one of them deserves its own arithmetic: increasing the size. **Pyramiding** is adding only in the direction the trade has already moved, at levels written into the plan, with the total risk budget unchanged. The mechanics sound like a contradiction — how can a larger position carry the same risk? — and the resolution is that the stop moves up with the average entry. Worked through plainly: a position entered at 100 with a stop at 97 carries 3 points of risk. When price reaches 106, the original stop moves to 103, so the original shares now risk nothing measured from the entry, and the trade is up 3R on the first unit. A second tranche can now be bought at 106 with a stop for the whole position at 103, which is 3 points of risk on the second tranche and zero on the first — so a second unit can be added while the trade’s risk stays at 3 points per unit, which is the original 1R. The average entry rises to 103, the shared stop sits 3 points under it, and the position is twice the size with the same open risk as the day it was opened. Two numbers have to be recomputed every time, and they are the two people get wrong. The first is the **average entry**, because the stop has to be re-derived from it rather than left where it was: adding a tranche at 106 while leaving the stop at 97 on the whole position turns a 1R trade into a 3R trade in one action. The second is **total open risk**, measured from the average entry to the shared stop on the entire size, and it is the number that decides whether the second tranche is an add or a new trade. If total risk after the add exceeds the original budget, what happened is not pyramiding; it is the risk of a fresh position smuggled in under an existing one, which is exactly what the heat cap in R11 exists to catch. The reason adding to a winner is not the mirror image of averaging down is the direction of the evidence, and the shape of the risk. Averaging down increases size as the price evidence against the thesis accumulates, and it moves the average entry *closer* to a stop that does not move — so open risk rises while the reason for the position weakens. Pyramiding increases size as the price evidence confirms the thesis, and it moves the shared stop *up* toward the average entry, so the risk to the account’s capital falls while the size grows. The distinguishing test is not whether the trade is winning; it is which way risk is moving. An add that raises total risk after the trade has moved in your favour is still an add that raises risk, and it belongs in the plan as a decision with its own budget rather than as a reward for being right. The cost is real and worth stating in advance. Every add raises the average entry and therefore tightens the stop in percentage terms relative to the position, so ordinary noise now closes a trade that is twice the size — the win rate on the final tranche is worse than the win rate on the first, and the payoff has to be larger to compensate. And the geometry limits the technique: if each tranche is only added when the trade has moved a full R, the number of adds is capped by how far the trade runs and by how much room the structure offers between the shared stop and the current price. A trade that never reaches the second level simply never gets a second unit, which is the correct outcome rather than a missed opportunity. • Add only in the direction of the move, at pre-written levels, with the stop raised to the new average entry. • Recompute the average entry and the total open risk on the whole size every time you add. • If total risk after the add exceeds the original 1R budget, it is a new position, not an add. • The test is the direction risk moves, not whether the trade is winning. • Each add tightens the stop relative to the position, so the last tranche has a worse win rate and must earn a bigger payoff. Pyramiding is optional, not obligatory. A plan that never adds is a complete plan — the three legal moves are hold, trail and invalidate, and a fourth move is only permitted when it has its own written arithmetic. The version to avoid is the one that is described as “letting the winner run” while the position is quietly doubled at a price nobody wrote down.

Sources

• Trailing stops and exit rulesTharp, “Trade Your Way to Financial Freedom”
• Expected value of early exitsStandard trade-management literature; expectancy arithmetic
• The disposition effect in exitsShefrin & Statman (1985); behavioural finance literature