A stop-loss rule prescribes when an exposure should be reduced or closed after a specified loss or adverse market move. It can discipline a strategy's risk-taking, but the threshold that prompts an exit is not necessarily the price at which the exit happens. A rule to sell below 95 does not guarantee a fill at 95 or cap the loss at that level.
A stop can also protect part of an existing gain; activation need not mean the trade as a whole has lost money. The rule is broader than a broker order. A trader can implement it through a resting conditional order, a monitored signal followed by an order, or a portfolio policy that reduces several holdings. A portfolio rule may measure loss from a high-water mark, whereas an individual trade rule may use its entry price. Paleologo's institutional treatment distinguishes partial from full liquidation and permanent shutdown from a specified restart policy. The loss baseline and response belong in the rule's definition.
Trigger first, execution second
For U.S. securities, FINRA Rule 5350 defines a sell stop order as becoming a market order when a transaction occurs at or below the stop price. A sell stop-limit instead becomes a limit order. The rule permits other activation events, including quotations, under separately distinguishable order types with disclosure of the trigger. These are U.S. rules, not a universal convention for every instrument or venue.
The distinction has an economic cost:
| Instruction after activation | What it controls | What remains uncertain |
|---|---|---|
| Market order | Seeks execution against available liquidity | Fill price; completion can also be interrupted |
| Sell limit order | Any sale must be at the limit price or higher | Whether and how much will sell |
The SEC's stop-order bulletin explains both price uncertainty and the risk that a stop-limit remains unfilled. A brief adverse move can also trigger an exit before a recovery.
Check the actual broker terms. For example, Fidelity's equity-order FAQ describes transaction-based activation and possible market-centre filters intended to exclude erroneous trades; it also says not all securities or trading sessions are eligible for stops. A chart touching a horizontal line alone does not establish that a particular order activated or filled.
A 95 trigger can produce a 90 fill
Consider a fictional long position of 10 whole shares bought at USD 100 per share, costing USD 1,000. The sell stop is USD 95. Assume it is accepted and active in the relevant session, and that the next eligible transaction after 96 is 90, with no intervening transaction at 95. Stipulate that the resulting market order then sells all 10 shares at 90. This is a scenario, not an execution forecast.
| Cash ledger, USD | Conditional exit at 95 | Stipulated exit at 90 |
|---|---|---|
| Entry cash flow | −1,000 | −1,000 |
| Exit cash flow | +950 | +900 |
| Gross P&L | −50 | −100 |
For quantity , entry price and actual fill , gross P&L is
The conditional loss at the stop price is . The stipulated gross loss is : twice the conditional amount. The extra USD 50 is the effect of the worse fill relative to 95. Fees, dividends and financing are zero here; separately charged transaction costs would further reduce net P&L.
Original synthetic sequence and cash ledger, USD. A gap crosses the trigger without an assumed trade at 95. The full 10-share fill at 90 is stipulated, not predicted; fees, dividends and financing are zero.
Changing the instruction to a stop-limit does not manufacture a 95 buyer. With a 95 stop and 94 sell limit, and only available bids at 90 after activation, the order cannot sell at 90. It may remain unfilled while the position continues to lose value. This is a separate branch, not the completed-sale ledger above.
Position sizing therefore needs an explicit exit assumption. A USD 50 planning budget supports 10 shares conditional on a 95 fill and zero costs; it does not convert that fill assumption into a maximum loss. Scenario-testing a 90 fill must retain the original quantity rather than resize it after seeing the gap.
Make the rule reproducible
A backtest must specify the monitored variable, loss baseline, trigger source, equality convention, observation frequency, eligible session, order instruction and action after partial or failed execution. It must also state whether stopping prevents re-entry, allows a cooldown or requires a new signal. Otherwise two implementations of the same “stop” can trade different paths.
Daily high and low prices do not reveal the order of intraday events or the depth available after activation. Automatically booking every stopped trade at its threshold can erase gap risk. Selecting the best threshold on the evaluation sample also creates overfitting; the stop is part of the strategy being tested, not an automatic improvement to it.
For the mechanics behind that execution branch, Market, Limit and Conditional Orders follows activation, partial fills, lifetime and cancellation. The full lesson requires membership; the linked page offers a public preview.
Sources
- Giuseppe A. Paleologo, Advanced Portfolio Management: A Quant's Guide for Fundamental Investors (Wiley, 2021), section 9.1, pp. 150–151 — institutional loss baselines, exposure reduction and shutdown versus restart.
- FINRA Rule 5350: Stop Orders — transaction triggers, conversion to market or limit orders and disclosure of alternative activation types; checked 26 October 2026.
- SEC Investor.gov: Stop, Stop-Limit, and Trailing Stop Orders, updated 18 August 2026 — trigger versus execution and nonfill risk; checked 26 October 2026.
- Fidelity: Trading FAQs — Order Types — a named broker's stock-stop handling and eligibility qualifications; checked 26 October 2026.