What Is a Stop Loss and Where Should It Go?
Michael Whitman
·
Trading since 2017, full-time since 2020
·
Published 2026-09-05
·
Updated
·
Reviewed
A stop loss is an order that closes your trade automatically at a price you set in advance. Its job is to make the size of a loss a decision you made calmly, though a market that opens beyond the stop fills it at the next available price, not at your line.
Everyone knows they should use one. The part that is rarely taught is how the price gets chosen, and how often a given choice actually gets hit. The second part can be measured, and this page does it on SPY and eight large stocks.
How it works
A stop loss is an order that closes your trade at a price you set in advance.
Mechanically it is a resting order sitting at that price. When price reaches it, it fires and your position closes; you do not have to be watching, and you do not have to decide anything.
That last part is the whole point. The decision gets made before you have money on the line, which is the only moment you are able to make it well.
Stops are a trading tool, not an investing one. If you are buying a broad fund to hold for years, you generally do not need a stop loss: a normal dip would trigger it and sell you out of a holding the plan expected to fall and recover, which is the same point the dollar cost averaging page makes about regular investing.
Two kinds of stop order
Worth knowing before you place one, because brokers offer both and they behave differently in exactly the situation you need them.
A stop-market order becomes a market order when your price is reached. It gets you out, at whatever price is available. That may be worse than your line.
A stop-limit order becomes a limit order instead. It will not fill worse than the price you set, and in a fast move that means it may not fill at all, leaving you in a position you thought you had closed.
The protection against a bad fill and the risk of no fill are the same feature. For a stop whose job is getting you out, the market version is usually the one that does it.
The four ways people pick the price
1. A level on the chart
Put it beyond a swing low, a level, or the edge of a range. This is the only method where the price means something: it marks the point at which your reason for the trade has stopped being true.
2. A fixed percentage
Simple, consistent, and it takes no account of the chart or of how much the instrument moves. The data below shows what that costs: the same 5% is a wide stop on a quiet fund and a tight one on a volatile stock.
3. A multiple of recent range
Measure how big a normal day is lately and place the stop some multiple of that away. This adapts to the instrument, so a quiet market gets a tight stop and a wild one gets a wide one, without you deciding anything. It still does not know where the levels are.
4. Time
Close the trade if it has not done what you expected within a set period. Rarely taught, occasionally sensible, because money tied up in a trade that is going nowhere is not available for the next one.
Too tight and too wide
Too tight is the common one. The stop sits inside the range price moves in on an ordinary day, so it gets hit by nothing in particular. On SPY since 1996, a 2% stop was reached within 20 sessions after 55.05% of closes.
Too wide is not the opposite mistake. A wider stop with the same money at risk simply means a smaller position; that is the arithmetic on the risk management page, and it is a trade-off rather than an error.
The obvious price is the crowded one. A stop just under the swing low feels safe because it is obvious, and obvious means many other stops sit there too; the liquidity page is the long version of this.
A worked example
A real gap through a stop, on SPY. Buy 100 shares at the 30 August 2001 close of $113.32 and place a 5% sell stop at $107.65. The planned loss is 100 × ($113.32 − $107.65) = $567.00, before costs.
SPY did not trade at or below the stop in the six sessions that followed; the lowest low was $107.70, five cents above it. US stock markets then closed after the attacks of 11 September and reopened on 17 September. SPY opened that day at $101.00, already $6.65 below the stop.
A stop-market order fills at about the open, so the loss becomes 100 × ($113.32 − $101.00) = $1,232.00: 10.87% of the entry instead of 5%. The extra $6.65 a share is 5.87% of the entry price, the largest gap-through of a 5% stop in SPY’s daily record since 1996. A stop-limit at $107.65 would not have filled at all that day, because the high was $106.40.
The lesson is not to skip the stop. It is that position size has to allow for a fill worse than the line, because the line is a request.
The original data
What was measured. Every daily close after the first in each series was treated as an entry (so SPY entries start on 3 January 1996 and the stocks on 4 January 2000). A sell stop was placed a fixed percentage below that close, and the stop counts as hit if any later day’s low reached it within the next 5, 20 or 60 sessions. If that day opened below the stop, the stop was gapped through and the fill is taken at the open. Prices are Yahoo Finance daily bars downloaded 3 October 2026, adjusted for stock splits but not for dividends, as a resting order sees them. SPY starts on 2 January 1996 because some early-1990s Yahoo opens are copies of the prior close; the stocks start on 3 January 2000. Entries overlap, so these are shares of start days, not independent trials.
| SPY stop | Hit within 5 sessions | Within 20 | Within 60 | Gapped through (20 sessions) |
|---|---|---|---|---|
| 2% below | 31.13% | 55.05% | 70.46% | 14.83% of hits |
| 5% below | 6.27% | 22.86% | 41.25% | 14.74% of hits |
| 10% below | 0.92% | 6.75% | 18.47% | 20.15% of hits |
A fixed percentage is a different stop on every stock. The same 5% stop over 20 sessions, with each name’s median daily high-to-low range (as a share of the prior close) for scale:
| Ticker | Median daily range | 5% stop hit within 20 sessions | Hits gapped through |
|---|---|---|---|
| SPY | 1.09% | 22.86% | 14.74% |
| JNJ | 1.29% | 23.14% | 11.02% |
| KO | 1.30% | 24.94% | 13.93% |
| XOM | 1.66% | 37.17% | 15.20% |
| MSFT | 1.77% | 38.41% | 16.19% |
| JPM | 1.87% | 43.57% | 15.85% |
| AAPL | 2.19% | 46.16% | 17.38% |
| AMZN | 2.53% | 51.83% | 16.66% |
| NVDA | 3.37% | 59.36% | 14.02% |
The ranking follows the range. SPY (median daily range 1.09%) and JNJ (1.29%) sat near 23%; NVDA (3.37%) and AMZN (2.53%) above 50%. Across the eight stocks pooled, a 5% stop was hit within 20 sessions after 21,769 of 53,656 closes, 40.57%. These are stocks that are large today, so the list leaves out companies that shrank or vanished along the way. NVDA is the extreme case: its split-adjusted close rose about 2,400-fold from 3 January 2000 to 2 October 2026, so its high hit rate came on the way up, not from a falling stock.
On one fund, volatility moved the number threefold. Splitting SPY’s entry days into thirds by the standard deviation of the previous 20 daily returns, a 5% stop was hit within 20 sessions after 12.00% of the 2,566 entries in the calm third (under 0.71% a day; the exact cut is 0.7071%), 21.20% in the middle and 35.54% in the volatile third (1.10% a day or more; the exact cut is 1.1023%); the volatility table is published too.
Gaps were a steady minority of hits. For a 5% stop over 20 sessions, 260 of SPY’s 1,764 hits were gapped through, 14.74%, at a median of 0.40% of the entry beyond the stop. The share ran from 11.02% of hits on JNJ to 17.38% on AAPL, and single gaps were brutal: AAPL’s worst cost 42.31% of the entry beyond the stop on 29 September 2000. The full stop grid is published with the per-stock table.
In the search study behind this site, 307 of 24,971 videos have stop loss in the title, at a median of 7,126 views; 76% never pass 50,000.
How this connects
An ATR stop is the formal version of the range-multiple method, and the data above is the case for it: a distance measured in the instrument’s own movement means roughly the same thing on SPY and NVDA, where a fixed 5% does not.
A trailing stop moves only in the trade’s favor, which turns the protective order into an exit for profits as well.
Risk per trade is the other half of the calculation. The stop sets the distance; the fixed share of the account sets the amount.
A bracket order places it together with a take profit, so neither can be forgotten after entry.
When it fails
It does not fill where you put it
The stop price triggers the order; it does not promise the fill. If the market opens below your price, over a weekend, a closure or on news, it fills at the next available price, which can be well past your line.
The percentage ignored the instrument
A stop copied from one market to another carries a different meaning. The same 5% that held through most months on SPY was reached within a month after most closes on NVDA.
You moved it
Moving a stop away from price is not patience. It converts a planned, survivable loss into an unplanned one of unknown size, and it feels reasonable every single time.
You are choosing the price to fit the position
If the stop keeps ending up wherever makes the position size comfortable, the chart is not choosing it. That is the failure the risk management page is built around.
Related
Risk management is the page this one feeds: the stop gives you a distance, and the distance gives you a position size.
Swing highs and lows are where the best stop prices come from, because they are prices the market made rather than numbers you chose.
And liquidity explains why the obvious stop price is the one most likely to be reached.
Choose the stop price from the chart first and size the position from that distance, never the other way round. The hardest part is not choosing the price. It is leaving it alone once the position is open and price is getting close.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.