A Trailing Stop Is Not the Absence of a Target. It Is Nine Numbers.

A stop and an exit rule sit on adjacent lines of the same trade plan and do two unrelated jobs. The stop fixes the largest loss the position is permitted to produce, and does nothing whatever to the top end of the outcome. What sets the top end is the rule that closes a position which is working. That is a separate specification with its own parameters, and it either exists before the order is sent or it gets invented while the position is open.

The shorthand for the second rule is no target. Leaving the field blank is not a rule; it is a decision postponed to the moment it is hardest to make. The alternative to a fixed target is a trailing rule, and a trailing rule is longer to write than a fixed one, not shorter.

What follows is arithmetic, not a market study. Three exit rules run over the same 20,000 simulated paths, same entry, same stop, under assumptions stated in full below. Nothing here ranks the rules or describes any real instrument.

What a Stop Order Limits, by Definition

The SEC Office of Investor Education and Advocacy publishes an investor bulletin on stop, stop-limit and trailing stop orders. The boundary of what a stop does is written into its definitions.

• Stop order. An order to buy or sell a stock once the price reaches a specified price, known as the stop price. When the stop price is reached, "a stop order becomes a market order."
• Stop-limit order. Once the stop price is reached, the order becomes a limit order to be executed at a specified price or better. The bulletin notes such an order "may not be executed if the stock's price moves away from the specified limit price."
• Trailing stop order. "A stop or stop limit order in which the stop price is not a specific price. Instead, the stop price is either a defined percentage or dollar amount, above or below the current market price of the security."

Two details in the trailing stop definition do most of the work. The first is that a trailing stop is a real order type with a written definition, not desk folklore. The second is its asymmetry: as the price moves favourably the stop price "adjusts or 'trails' the market price of the security by the specified amount," but if the price moves unfavourably "the trailing stop price remains fixed." The ratchet turns one way only.

Notice what none of the three definitions contains. There is no profit target anywhere in them. The entire specification of a stop concerns the side of the trade that loses money. A stop placed 2.00 points below entry has said everything it has to say the moment the position is open.

The bulletin adds two operational points that matter later. These order types "may not be available through all brokerage firms," and different trading venues may use different standards for triggering them.

Three Exit Rules Behind One Stop

The setup is identical across all three rules. Entry at 100.00. Initial stop at 98.00. That 2.00-point distance is one unit of risk, written R throughout. Results are reported in R so that the exit rule is the only thing that varies.

• Rule A — fixed target. A sell limit at 104.00, which is +2R, with the stop left at 98.00. Whichever is reached first ends the trade.
• Rule B — fixed-width trailing stop. No target. The stop is placed 2.00 points below the highest price reached since entry, ratcheting up only, and never below the original 98.00.
• Rule C — volatility-scaled trailing stop. No target. The trailing distance is 6.7 times the mean absolute bar change over the trailing 20 bars, floored at 1.00 point and capped at 3.00. Same one-way ratchet.

The multiplier in rule C was chosen so its average distance would land near rule B's fixed 2.00. Across the run the realised averages were 1.91 points for C against 2.00 for B. The two trailing rules are matched on width and differ on responsiveness.

The assumptions, in full. Each path is 200 bars, each bar moving up or down with equal odds. Step size is 0.15 or 0.45 points, drawn from a two-state regime with a 5% chance per bar of switching. Prices are observed only at bar closes. A triggered stop is filled at the close that triggered it, at or below the stop price; a target is filled at its limit price. Seed 20260823, 20,000 paths. This is a synthetic process with no drift and no memory. It is not a price series.

On the single path above, rule A's limit filled at bar 24 for +2.00R, rule B's trail fired at bar 40 for +2.48R, and rule C stayed in until bar 71 for +3.15R. Early on, through the quiet stretch, rule C's stop sat above rule B's: its distance had compressed toward the 1.00-point floor. Through the volatile section it widened. On a different path the ordering reverses. One path establishes nothing about a rule, which is the reason for running twenty thousand.

The Specification Gets Longer, Not Shorter

This is the part that the phrase "no target" hides. A fixed target requires three numbers to be settled in advance. A fixed-width trail requires five. A volatility-scaled trail, as specified here, requires nine. Opening the upside is not the removal of a decision; it is the addition of six of them.

Three of those fields deserve naming, because they are the ones most often left implicit:

• Which price the trail follows. Last trade, the bid, or the session high are different references and produce different stop paths from identical price action. The SEC bulletin's note that venues may apply different triggering standards is the reason this cannot be assumed.
• The floor and the ceiling. Without a floor, a volatility-scaled distance collapses to near zero in a quiet stretch and exits on noise. Without a ceiling, one violent bar widens the distance so far that the stop is no longer a risk control. In this run the floor was binding on 29.3% of held bars.
• Whether the distance may narrow again. By the SEC definition the stop price never moves against the position. But a scaled distance can shrink when the volatility measure falls, and the specification has to say whether a shrinking distance is allowed to pull the stop up further, or whether the stop is frozen until a new high is made. Both are defensible. Neither is the default.

What the Arithmetic Produced

Across 20,000 paths	A fixed target	B fixed trail	C scaled trail
Mean result	-0.02R	+0.01R	+0.00R
Median result	-1.05R	-0.38R	-0.30R
25th percentile	-1.13R	-0.90R	-0.52R
75th percentile	+2.00R	+0.45R	+0.30R
90th percentile	+2.00R	+1.65R	+1.27R
99th percentile	+2.00R	+4.50R	+3.75R
Largest single result	+2.00R	+9.75R	+8.10R
Standard deviation	1.43R	1.21R	1.01R
Exited at the initial stop	60.5%	11.8%	10.4%
Exited at a trailing stop	—	86.2%	89.5%
Exited at the target	30.5%	—	—
Still open at bar 200	9.0%	2.0%	0.2%
Result above zero	36.9%	35.6%	33.4%
Result above +2R	0.0%	7.2%	5.2%
Result above +3R	0.0%	3.0%	2.0%
Share of all gains taken by the top decile	30.1%	62.2%	65.9%

Start with the row that does not separate them. The three means are -0.02R, +0.01R and 0.00R, against a standard error of roughly 0.01R on each. They are indistinguishable, and they were always going to be. The process was built with no drift, so no rule for choosing when to stop can move the average. Each exit rule took the same expectation and redistributed it. That is the whole finding, and everything below is a description of the redistribution.

Rule A produced two spikes and little else. 60.5% of paths landed at the initial stop, 30.5% at the target, 9.0% ran out of bars. The cumulative curve rises only 9 points between -1.05R and +2.00R, and that band holds nothing but the trades still open when the bars ran out. At +2.00R it is vertical: 30.5% landed on exactly that number. Above it there is nothing, and there could be nothing. The rule truncated its own right tail the moment it was written.

The trailing rules produced continuous distributions. Their medians are far less bad, -0.38R and -0.30R, because only about one trade in nine reached the original stop. Their tails run to +9.75R and +8.10R. But the concentration in those tails is severe: the largest 2,000 gains, one tenth of all trades, accounted for 62.2% of every point rule B made and 65.9% for rule C, against 30.1% for the fixed target.

Frequency and average point in opposite directions. Head to head on identical paths, rule B produced the better result on 55.5% of them, rule A on 29.4%, with 15.1% identical; rule C beat rule A on 57.1%. And the means are still the same. How often a rule is the better one, and how much it produces on average, are two different measurements. A comparison reporting only the first will rank rules the second cannot separate.

One figure sets the scale of what these rules were reaching for. Of the 20,000 paths, 37.6% traded above 104.00 at some point, 8.5% above 108.00, and 1.2% above 112.00. The median path's highest print was 103.00. A rule that reaches for the tail has to survive the majority of paths where it is not there.

The Give-Back Is the Rule Working

A trailing exit has an arithmetic identity built into it. When the trail fires, the exit is the highest price reached, minus the trailing distance, minus whatever gap sits between stop price and fill. A trailing rule therefore cannot exit at the high. Not sometimes. Ever.

The run puts numbers on it. Among trades that exited on the trail, the median give-back from the running peak was 1.13R for rule B and 0.68R for rule C. Rule B's figure is close to its 1.00R trailing distance, and the excess of about 0.13R is roughly the average shortfall between its stop price and its fill, which the run put at 0.11R. Rule C gave back less because its exits cluster where its distance had narrowed: the median trailing distance at the exit bar was 1.51 points, against 1.91 across all held bars.

This matters for review discipline more than for design. A trade that peaked at +4R and closed at +2.9R has not revealed an error. That gap is the price of the specification, knowable before entry and writable into the plan as a number. The error would be discovering it afterwards and shortening the trail in response, which converts a written rule back into an unwritten one.

A Note on the Fill

By the SEC's definition a triggered stop becomes a market order, and Investor.gov's order primer is explicit that a market order "guarantees that the order will be executed, but does not guarantee the execution price." Here, fills were taken at the bar close that triggered the stop, averaging 0.23 points below the stop price for rules A and B and 0.17 for rule C, or 0.11R and 0.08R. That leak is the only mechanism that moved any mean off zero, and it applies to all three rules, since all three exit on a stop most of the time. What happens to a stop across a session boundary is a separate subject.

What Would Invalidate This

• The process has no drift. That assumption is what forces the three means to be equal. Under a drifting series the equality no longer holds, and a rule that holds a position longer collects more or less of that drift depending on its sign. Every distributional statement above is conditional on a driftless generator.
• No memory, no jumps, bar closes only. The largest possible single-bar move is 0.45 points, and no path within a bar is modelled. Real series trend, mean-revert and gap, and any of those changes both the fill distribution and the trail's hit rate, unequally across the three rules.
• The volatility proxy is crude. Mean absolute bar change over 20 bars is one measure among many, and a different measure or lookback is a different rule C.
• The widths were matched deliberately, 1.91 against 2.00. Nothing here describes a 0.5R trail or a 4R trail.
• Platform behaviour was assumed, not observed. If a broker's trailing stop follows a different reference price, recalculates on a different schedule, or is unavailable, then the rule tested is not the rule that will be sent. The SEC bulletin flags both availability and triggering standards as firm- and venue-specific.
• No costs, no financing, no taxes, no position sizing changes, one position at a time.
• This does not rank the rules. Ranking would require the drift and jump structure of a specific instrument over a specific horizon, and none of that is in the generator.

Concrete Framework

1. Write the stop as a price before the entry order is sent. That number caps the loss and is finished doing its job the moment it exists.
2. Write the exit rule for a working position in the same sitting. A blank field means the operative rule is "decide later," applied under pressure.
3. If the choice is a fixed target, settle three fields: stop price, target price, and the time limit at which a trade that reaches neither is closed.
4. If the choice is a fixed-width trail, settle five: initial stop, trailing distance, the reference price the trail follows, how often the stop is recalculated, and the time limit.
5. If the distance is scaled to volatility, settle nine: the five above plus the volatility measure and its lookback, the multiplier, the floor, and the ceiling.
6. Confirm with the broker which of these the platform supports and which price triggers the order. Availability and triggering standards are firm- and venue-specific: a question with an answer, not an assumption.
7. Write the expected give-back down in advance. A trailing exit is the peak minus the distance minus the fill gap, so state that number before entry, not in the review.
8. When reviewing a closed trade, separate the rule from the path. One trade where the trail beat the target, or the reverse, carries no information about either.
9. Compare shapes, not averages alone: median, the top decile's share of total gains, and the fraction of results landing beyond where a fixed target would have capped them.
10. Re-run any such comparison with the fill convention that matches the account. Filling a stop at its stop price rather than at the price that triggered it flatters every rule tested, unequally.

The distinction is narrow and load-bearing. A stop limits the loss. It does not cap the gain, and removing a target does not uncap it either. What shapes the upper half of the distribution is a second rule, with its own fields, written down before the order or written by the position while it is open.