A stop loss sells an asset once its price falls a set amount below the price paid. We tested it on 6,600 one-year purchases. A 10% stop sold in 46.8% of them. After a sale, the money stayed in cash. In 68.3% of those cases, holding to the year end would have finished above the sale price.
A 10% stop did one job well. It cut the worst one-year loss from 86.4% to 43.4%. It also cut the mean one-year return, which is the simple average, from 27.9% to 17.4%.
Do stop losses work? A stop loss is an order to sell an asset once its price falls a set amount below the price paid. In our tests, stops limited the worst losses, but they lowered the mean return. Holding finished above the sale price in most cases where a stop sold. The money left in cash earned nothing.
A 10% stop sold in 46.8% of one-year purchases. Holding to the year end finished above the sale price in 68.3% of those.
The worst one-year loss fell from 86.4% with no stop to 43.4% with a 10% stop.
The mean (simple average) one-year return fell from 27.9% with no stop to 17.4% with a 10% stop. After a sale, the money stayed in cash.
What is a stop loss?
A stop loss is an order that sells an asset when its price falls to a set level. A fixed stop sits a set percentage below the price paid. A 10% fixed stop on a purchase at 100 sells when the price reaches 90.
A trailing stop sits a set percentage below the highest price since the purchase, so it moves up with the price. This article tests fixed stops only.
A standard stop order turns into a market order once the stop level trades. A market order sells at the best price available at once, which can be well below the stop level. Our explainer on market orders and limit orders covers fills in a fast market.
How did we test stop losses?
We used 23 daily price files. Ten are single stocks (AMD, AMZN, GOOGL, INTC, META, MSFT, NFLX, NVDA, PLTR and TSM, which is Taiwan Semiconductor Manufacturing). Twelve are exchange-traded funds (ETFs) covering US stocks, sectors, gold and Treasury bonds, which are bonds issued by the US government. One is Bitcoin.
SPY (the S&P 500 ETF) and VOO (the Vanguard S&P 500 ETF) track the same index. We counted both, so that index is in the sample twice.
An entry is one purchase, made at the close on the first trading day of a month. The entry close is that day's closing price. The rule had four parts.
Entry: buy at the entry close of every calendar month.
Holding period: 365 calendar days.
Stop: sell at the first daily close at or below 5%, 10% or 20% under the entry close.
After a sale: hold cash that earns nothing, with no re-entry.
Entries run from 2 January 1990 to 1 October 2025. An entry counts only if a full 365 days of data follows it before 26 October 2026. Each file starts on its own date, so MSFT supplies 430 entries and PLTR supplies 62. The total is 6,600.
Consecutive entries overlap by 11 months, so the 6,600 are not independent tests. Returns count dividends for the 22 stocks and funds.
How often did a stop get hit?
A close 5% or more below the entry close happened within the year in 63.9% of entries. A 10% fall happened in 46.8% and a 20% fall in 26.4%.
Source: YX Insights price data
Holding with no stop finished the year below the entry price in only 25.8% of entries. A stop reacts to every dip, including the ones that recover.
Do stop losses limit the loss?
Chart 2 shows the worst loss across all 6,600 entries. It also shows the loss only 1 entry in 20 exceeded.
Source: YX Insights price data
With no stop, the worst one-year loss was 86.4%. It came from AMZN, bought on 3 April 2000.
A wider stop allowed a larger worst loss. The 5% stop capped it at 36.9%, the 10% stop at 43.4% and the 20% stop at 49.7%.
The loss that only 1 entry in 20 exceeded was 39.0% with no stop. It was 9.3%, 14.2% and 23.8% with the three stops.
A 10% stop did not hold losses to 10%. Our stop sells at the first daily close at or below the level. A close can sit far below it after a one-day fall.
On average, stops sold below their level. The mean sale came at a 6.7% loss for the 5% stop, 11.9% for the 10% stop and 22.4% for the 20% stop.
What did stop losses do to returns?
Chart 3 shows the mean and the median one-year return. The mean adds up all the returns and divides by the number of entries. The median is the middle return. After a sale, the test held cash that earned nothing, so a stop also missed any later rise.
Source: YX Insights price data
Holding with no stop earned a mean of 27.9% and a median of 14.7%. The 5% stop gave a mean of 12.8% and a median of -5.3%. The 10% stop gave 17.4% and 0.2%. The 20% stop gave 23.3% and 11.0%.
The stops lowered the mean by 15.0 points at 5%, 10.5 points at 10% and 4.6 points at 20%. The mean sits well above the median because a few entries gained several times their price.
The mean fell in 22 of the 23 files with a 10% stop.
Did the result differ for stocks, funds and Bitcoin?
The cost was largest where holding earned most. The table gives the mean return from holding and the cost of each stop in points.
Group | Entries | Mean return from holding | Change with a 5% stop | Change with a 10% stop | Change with a 20% stop |
|---|---|---|---|---|---|
Single stocks | 3,052 | 40.9% | -23.8 | -16.2 | -7.2 |
Funds | 3,406 | 12.1% | -5.6 | -3.7 | -1.6 |
Bitcoin | 142 | 124.5% | -53.6 | -50.2 | -20.2 |
All 23 files | 6,600 | 27.9% | -15.0 | -10.5 | -4.6 |
One-year returns, with dividends counted, from entries between 2 January 1990 and 1 October 2025. Changes are in percentage points of mean return against holding. Source: YX Insights price data
Single stocks averaged a 40.9% one-year return from holding. Funds averaged 12.1%. A 10% stop cost single stocks 16.2 points and funds 3.7 points. Bitcoin has only 142 entries, so its row is the least reliable.
The 23 files are not a random draw of listed securities, so these figures describe these files only.
What happened after a stop sold?
Chart 4 takes only the entries where a stop was hit. It compares holding to the year end with selling at the stop.
Source: YX Insights price data
After a 10% stop sold, holding finished above the sale price in 68.3% of cases. The figure was 70.3% for the 5% stop and 61.9% for the 20% stop. The stop finished ahead in the rest.
Did the 10% stop beat holding in any year?
Across the 36 entry years from 1990 to 2025, the 10% stop beat holding on the mean in five: 2000, 2001, 2007, 2008 and 2021.
Does the holding period change the result?
We reran the test at six months (6,738 entries) and three years (6,048 entries). The mean fell at every stop level in both.
What does the test leave out?
The test uses daily closes, while a real stop order can trigger during the day. The test has no re-entry, no interest on cash, no fees and no taxes.
Bitcoin has 365 closes a year against about 252 for stocks, so it has more chances to touch a stop level.
For the measure of how far prices fall from a high, see our explainer on what a drawdown is. Our explainer on loss aversion covers why a loss feels larger than a gain.
In these 6,600 entries, a stop loss limited the worst loss but lowered the mean return. Holding finished above the sale price in most cases where a stop sold. The money left in cash earned nothing. The test cannot say how a stop performs with re-entry or inside a different plan.
Get the next explainer
The Academy is free to read, here on the web. The Long Trend, our free weekly letter, comes by email.
Learn more with YX Insights
This explainer is part of the YX Insights Academy. Each one takes a single idea and checks it against real data.
The same approach runs through everything else we publish:
Systematic Portfolio: ready-made portfolios for Macro & Megacaps and for Commodities. We publish the holdings and every change.
Multi-model Signals: a daily long-or-flat call on every name we cover. Each call shows how strongly our four models agree.
Research: company deep dives, macro commentary and essays on how we test.
Good places to start on the website:
Academy: more explainers like this one.
Start here: get our free weekly letter by email.
Macro & Megacaps Systematic Portfolio and Commodities: the two portfolios in detail.
Track record: live results since 3 August 2026, plus the ten-year backtest.
Upgrade: get the Systematic Portfolio, Multi-model Signals and research in full.
Common questions about stop losses
Does a stop loss protect you in a crash?
Only partly. A stop loss limits a loss down to the price at which it sells. In our tests the mean sale for a 10% stop came at an 11.9% loss, below the stop level. The worst came at 43.4%. A price that jumps below the level sells below it.
What is a good stop loss percentage?
None of the three stops we tested (5%, 10% and 20%) beat holding on the mean one-year return. The 20% stop gave up the least, 4.6 points, while the 5% stop gave up 15.0 points. Tighter stops also sold more often: 63.9% of entries at 5% against 26.4% at 20%.
What is a trailing stop loss?
A trailing stop loss is a stop that moves up with the price. It sits a set percentage below the highest price since the purchase, so it rises when the price rises and stays put when the price falls. We tested fixed stops only, so our figures do not describe trailing stops.
Is a stop order the same as a stop-limit order?
No. A stop order turns into a market order when the stop level trades, so it sells at the next available price. A stop-limit order turns into a limit order, which sells only at the limit price or better, so by design it may not sell in a fast fall. Our market order explainer covers both.
DISCLAIMER: This article is intended for educational purposes only. Any information or analysis in this note does not constitute an offer to sell or a solicitation of an offer to buy any securities. Nothing in this note is intended to be investment advice, nor should it be relied upon to make investment decisions. Any opinions, analyses, or probabilities expressed in this note are those of the author as of the note's date of publication and are subject to change without notice.
