A trailing stop follows the price up and sells when it falls back by a set distance. Here is how brokers and exchanges run one, how to size it, the mistakes, and two real examples on 2026 prices, one where the wider trail won and one where it lost.
November 15, 2026
17
min to read
The short answer
A trailing stop loss is a stop order that follows the price. On a long position it sits a set distance below the highest price since you bought: a percent, a dollar amount or a multiple of ATR. It rises with the price and never falls. When the price drops back by that distance, it sells, usually at market.
A stop loss, or stop order, is an instruction to sell once the price falls to a level you choose, the stop price. Investor.gov puts it simply: when the stop price is reached, the stop becomes a market order. A trailing stop is a stop whose level moves by itself. Schwab describes trailing stops as “not ‘orders’ per se but a means to automatically move or ‘trail’ stops” that are already in place.
On a long position, the stop sits a set distance below the highest price reached since you placed it. That distance is the trail. When the price makes a new high, the stop rises with it. When the price falls, the stop stays where it is. If the price falls far enough to touch it, the stop triggers and sells. A trailing stop on a short position is the mirror image: it sits above the lowest price and only moves down.
Say you buy at $100 with a 10% trail. The stop starts at $90. The price climbs to $120, so the stop climbs to $108. The price then slips to $112: the stop stays at $108. If it keeps falling and touches $108, the stop fires and you sell near there, with a gain of about 8% instead of the 20% you had at the top. That gap between the peak and the exit is the price of staying in the trend.
| Term | What it means |
|---|---|
| Trail, or offset | The distance between the best price and the stop: a percent, a dollar amount or a multiple of a volatility measure such as ATR. |
| Peak, or high-water mark | The highest price since the trailing stop was placed (the lowest, for a short). The stop is measured from it. |
| Stop price | Where the stop sits right now. It only moves in your favor. |
| Reference price | The price the venue watches to move and trigger the stop: the last trade, the bid, the midpoint or a mark price. |
| Trailing stop limit | A version that becomes a limit order instead of a market order when it triggers. |
There are three common ways to set the trail, plus one classic recipe.
That is why traders reach for ATR: the same percent means different things on different markets. Over the three months to November 13, 2026, BTC’s typical daily range was about 2.8% of its price, so a 10% trail was about 3.6 normal days of movement wide. Gold’s was about 1.6%, so the same 10% was about 6.2 days wide: far looser. HYPE moved about 5.5% on a typical day; a 5% trail on it would sit inside one ordinary day.
Here are both trails run on real prices, the same day, on two markets that rallied in August 2026: BTC, the Bitcoin perpetual on Hyperliquid, and xyz:GOLD, trade.xyz’s gold perpetual, which references the dollar spot price of one troy ounce of gold. A perpetual, or perp, is a futures contract with no expiry date. We picked a rally on purpose: a trailing stop only has something to follow when the price rises. The next section runs the same rules from the first day of every month.
BTC opened August 1 at $62,857. Its 14-day ATR was $1,745, 2.8% of the price, so the 3-ATR trail started about 8.3% below, close to the 10% trail. Both climbed with the rally. The 3-ATR trail sold first, on September 15 at $75,903, +20.8% in 45 days, when a dip from $82,268 touched it. BTC then went on to $87,471 on September 21. The 10% trail sat out that dip and sold on October 29 at $78,724, +25.2% in 89 days. BTC closed November 13 at $68,275.
Gold opened August 1 at $4,051.10. Its ATR was $70.50, 1.7% of the price, so 3 ATRs put the stop only about 5.2% below: half the distance of the 10% trail. Gold peaked at $4,696.90 on August 25. The 3-ATR trail sold on August 28 at $4,464.75, +10.2% in 27 days. The 10% trail waited until September 28 and sold at $4,227.21, +4.3%, after giving back most of the gain. Gold closed November 13 at $3,883.00.
| Bought August 1, 2026 | 10% trail | 3-ATR trail |
|---|---|---|
| BTC | Sold Oct 29 at $78,724: +25.2% | Sold Sep 15 at $75,903: +20.8% |
| Gold (xyz:GOLD) | Sold Sep 28 at $4,227.21: +4.3% | Sold Aug 28 at $4,464.75: +10.2% |
Same rules, same day, opposite winners. On August 1, a 10% trail was about 3.6 times BTC’s average daily range, wide enough to sit through a normal dip. On gold it was about 5.7 times, so it gave back most of the rally before it fired. The chandelier exit with its classic settings, the 22-day high minus 3 times the 22-day ATR, run under the same rules, would have sold on the same days, at $75,938 and $4,466.85. Two trades prove nothing about which trail is better; they show why the width has to fit the market.
No single number does, and the data shows why. Below, the same rules run from the open on the first day of every month in the data: 13 entries on BTC (Nov 2025 to Nov 2026) and 10 on gold (Feb to Nov 2026). Trades still open on November 13 are valued at that day’s close.
| Trail | BTC, 13 entries | Gold, 10 entries |
|---|---|---|
| 5% | +1.1% on average; 9 of 13 at a loss; held 8 days on average | -1.2% on average; 8 of 10 at a loss; held 11 days on average |
| 10% | +1.7% on average; 8 of 13 at a loss; held 33 days on average | -3.6% on average; 7 of 10 at a loss, 2 of them still open; held 41 days on average |
| 20% | -8.3% on average; 11 of 13 at a loss, 1 of them still open; held 52 days on average | -10.2% on average; 10 of 10 at a loss, 7 of them still open; held 101 days on average |
| 3 x ATR(14) | +0.7% on average; 6 of 13 at a loss; held 22 days on average | -0.4% on average; 7 of 10 at a loss; held 17 days on average |
| No stop, held to November 13 | -10.1% on average; 9 of 13 at a loss | -12.1% on average; 10 of 10 at a loss |
These are 13 and 10 trades in one year: far too few to crown a winner, as our guide to backtesting a trading strategy explains. What the data does support is a method, not a number: measure how far the market moves on a normal day, set the trail at a multiple of that, then test the width on past prices before you rely on it.
All three are exit orders that wait for a price. They differ in whether the level moves and what they protect.
| Fixed stop loss | Trailing stop | Take profit | |
|---|---|---|---|
| Where it sits | A set price below your entry (above, for a short) | A set distance below the best price since you placed it | A set price above your entry (below, for a short) |
| Does it move? | No, unless you move it | Yes, in your favor only | No, unless you move it |
| What it protects | Your capital, by capping the loss | Your capital first, then part of the gain | A gain you have decided is enough |
| When it fires | When the price falls to it | When the price falls back by the trail | When the price rises to it |
| What you give up | Any recovery after it fires | The gap between the peak and the exit, every time | Everything above the target |
| Typical order | Stop market or stop limit | Trailing stop or trailing stop limit | Limit order |
They also combine. Schwab’s guide to advanced order types walks through a one-cancels-other order, or OCO: a stock bought at $40, a take profit at $52 and a stop at $36, where whichever fills first cancels the other. A trailing stop can play the stop’s part in the same pair, or take over from a fixed stop once a trade is working.
The idea is the same everywhere; the details are not. Three of them decide what you get.
A plain trailing stop becomes a market order. Schwab’s thinkorswim guide sums up the trade-off: a trailing stop “could fill quickly but at an uncertain price during times of volatility,” while a trailing stop limit “may offer more control over the fill price but may not execute at all” if other orders take priority, liquidity is thin or the price falls below the limit. Interactive Brokers describes the same pair: as the price rises, its sell trailing stop rises by the trail, and when the price falls the stop stays put until a market order is sent. It also notes that it may simulate stop orders on some exchanges.
Each venue watches a reference price. On thinkorswim you link the trail to a price yourself: the mark, the midpoint between bid and ask, which Schwab calls a common choice for stocks with narrow spreads because it can protect against “temporary, anomalous price swings,” or the bid, which can sit closer to your real exit on wide spreads but reacts more to short swings. Kraken, a crypto exchange, lets you trail the Index, Last or Mark price, by a percent or a fixed amount, and places a market order once the price reverts from its best level by that offset. Robinhood’s trailing stop follows the highest price of the stock for a sell and triggers a market order at the best price available.
A day order expires at the close; a good-’til-canceled order stays until it fills or you cancel it. Schwab’s tutorial sets the trailing stop to GTC for that reason. On crypto exchanges and perpetuals, trading runs around the clock, so a trail can move and fire at any hour, weekends included.
| Venue | Trail set as | Follows | When triggered |
|---|---|---|---|
| Charles Schwab (thinkorswim) | A percent or a dollar amount | The price you link it to, such as the mark or the bid | Market order (trailing stop) or limit order (trailing stop limit) |
| Interactive Brokers | A trailing amount below the market price | The market price | Market order (trailing stop) or limit order (trailing stop limit) |
| Robinhood | A trail you specify | The highest price of the stock | Market order |
| Kraken | A percent or a quote-currency amount | Index, Last or Mark price | Market order |
Hyperliquid has its own trailing stop order type for perpetuals, with its own rules; our separate guide to stop losses on Hyperliquid covers it.
Ethy is an AI trading agent you ask before you trade, and it trades Hyperliquid’s crypto perps and trade.xyz’s stock, index and commodity markets. You can ask it for a trailing stop in plain words, on a position you already hold or inside an automation, a set of rules that trades on its own once you activate it.
There is no universal one. Size it to the market’s normal movement: in our data a 10% trail was about 3.6 typical daily ranges on BTC but about 6.2 on gold, so it suited one and was loose on the other. ATR-based trails express the gap as a multiple of the ATR instead; StockCharts’ ATR trailing stop uses 3 by default. Test any width on past prices first.
Neither is better in general. A fixed stop caps the loss on a trade and stays put; a trailing stop does the same at first and then moves up to protect part of a gain. The cost of a trailing stop is that it always sells some distance below the peak, and in choppy markets it can sell on noise.
No. A standard trailing stop becomes a market order when it triggers, and FINRA warns the fill can be significantly worse than the stop in a fast market. If the price gaps below the stop, the sale happens near the new price. A trailing stop limit avoids a bad fill but may not fill at all.
A volatility-based trailing stop developed by Charles Le Beau. For a long position it sits three times the Average True Range below the highest high of the last 22 periods, using a 22-period ATR. It rises as the period high rises and widens when volatility grows, which is meant to keep you in a trend through normal pullbacks.
Both trail the price the same way. When triggered, a trailing stop becomes a market order, which fills quickly at an uncertain price; a trailing stop limit becomes a limit order, which controls the price but may not execute if the market falls through the limit. Schwab and Interactive Brokers both offer the pair.
It depends on the venue and the order. At a stock broker, a day order expires at the close, while a good-’til-canceled order stays live; stocks can also open far from the last close. Crypto exchanges and perpetual markets trade around the clock, so a trail there can move and fire at any hour.
Trading involves risk, including the loss of your money. Ethy AI is a research and trading tool and does not give investment advice or recommendations. Backtests are simulations on historical data and do not predict future results.
Deploy your AI trading agent in one click.
