Break of Structure (BOS), Explained

Michael Whitman · Trading since 2017, full-time since 2020 · Published 2026-09-06 · Updated 2026-10-03 · Reviewed 2026-10-03

Video: Michael Whitman on YouTube, a beginner's guide to break of structure. Watch it on YouTube.

Covered on this page: TradingView and LuxAlgo.

How it works

In an uptrend, price makes higher highs and higher lows. A break of structure is the moment a new high closes above the previous swing high — the sequence continuing, confirmed.

In a downtrend it is the mirror: a close below the previous swing low.

That is the entire definition, once the swing is defined. No indicator and no calculation beyond one setting: how many bars on each side a high must beat to count as a swing. The rule used on this page is five, so a swing high is a day whose high is above the highs of the five sessions before it and the five after it. Since five more sessions have to print first, nobody can mark that swing until about a week later.

The level exists before the event does, which is worth more than it sounds. Mark the last swing high while price is still below it and the decision point is fixed in advance rather than rationalized afterwards.

The distinction that actually matters

Break of structure breaks WITH the trend. Change of character breaks AGAINST it. They are opposite events with nearly identical names, and confusing them inverts the read completely.

In an uptrend: a close above the last high is a break of structure. A close below the last low is a change of character. Same chart, same day, opposite meanings.

Use closes. A wick through the level is not a break by any usable definition — wicks poke through levels constantly on every timeframe, and treating them as structural events produces a signal every few candles.

A worked example

SPY’s daily chart from May to October 2025 is a clean chain. The read turned up on 12 May 2025, when SPY closed at $582.99 above the 25 March swing high of $576.41 — a change of character, because the read had been down, last confirmed by a downward break of structure on 3 April.

Then five breaks of structure in a row, each a close above the latest known swing high:

The chain ended on 10 October 2025. SPY closed at $653.02, below the 25 September swing low of $654.41, which had been known since 2 October. That close was a change of character down, and the string of breaks of structure stopped there.

Read the margins column before the last one. Two of the five breaks cleared the level by less than a dollar on a $600 share price. A swing rule of four or six bars would have drawn different levels and dated some of these breaks differently, which is why the rule has to be fixed before the chart is read.

And the 20-session column looks good only in hindsight. All five were higher 20 sessions later, but so were most stretches of 2025. The question is whether that happens more after a break than on any other day, which is what the data section below tests.

In practice: a trend is a chain of these

Strung together, breaks of structure ARE the trend. That is not an analogy — an uptrend is literally a sequence of closes above successive prior highs. Counting them is counting the trend.

Which gives the concept its most useful reading: how many in a row, and how much distance each covered. A trend making its sixth break of structure with each leg shorter than the last is a different situation from one making its second with each leg longer.

Volume is one independent check, and a weak one. Participation at the break is more encouraging than none, which is a statement about plausibility rather than outcome.

The timeframe changes the count. A five-minute break of structure inside an hourly range is a much smaller event than a daily one, and checking the timeframe above removes many of the signals a fast chart produces.

And nothing structural sits underneath the level. The order book has no memory of where a swing high was. If the level matters at all, it is because many people are watching the same obvious price.

The original data

The sample. 8,472 daily SPY bars, 29 January 1993 through 25 September 2026. The rule is the one in the worked example and on the market structure page: five-bar swings, each known five sessions after it forms; a newer swing replaces the older one the day it becomes known; a close beyond the live swing breaks it, and each level breaks once. A break in the direction of the current read is a break of structure, and the first break of the sample only sets the read.

That rule found 199 breaks of structure: 142 upward and 57 downward. The question here is what SPY did next. “Higher” means the close N sessions later was above the close on the break day, price only. The base rate is every session with N sessions still to come.

The day count is for the 20-session column; the five-session column has 8,467 days and 57 downward breaks.

Neither difference is distinguishable from chance. Tested exactly against the 63.91% base rate, the upward breaks give p = 0.49 and the downward breaks p = 0.68. Five sessions out, the upward breaks were actually higher a little less often than any day (p = 0.40).

A downward break did not mean lower. After 34 of the 56 downward breaks, SPY was higher 20 sessions later, and the median change was +1.84%. A long-rising market pulls every starting day upward, and the break did not overcome that.

Changing the swing width did not change the answer. With two-bar swings, 65.51% of 316 upward breaks and 65.91% of 132 downward breaks were higher 20 sessions later. With ten-bar swings it was 67.69% of 65 and 72.41% of 29. Each is within chance of its base rate.

One result out of 24 cleared the usual 5% bar, and it was not a break of structure: with five-bar swings, 70.00% of 110 upward changes of character were higher five sessions later (p = 0.012). Running 24 tests at the 5% level, about one such result is what luck alone produces.

Breaks also overlap. In 40 of the 141 gaps between upward breaks, the next one came within 20 sessions, so those windows share days and the effective sample is smaller than 142. Every event and its 5- and 20-session changes are in the SPY break of structure file.

The honest summary is that this is a description, not a signal. It tells you a trend is still doing what a trend does. That is worth knowing and, on SPY’s daily chart, it is not an edge, because what followed a break looked like what followed an ordinary day.

What it is not

It is not a signal to buy. It confirms a trend that was already visible. Buying the break means buying after the move that produced it, and the table above shows no extra follow-through to pay for that.

It is not new. Dow theory, drawn from Charles Dow’s Wall Street Journal editorials before his death in 1902, was describing markets in terms of trends long before the smart money vocabulary named this break.

It is not proof of anything. A break of structure that immediately reverses is common, and it happens most often in exactly the conditions where the tool is least applicable.

And it is not the same as a breakout. A breakout leaves a range. A break of structure continues a trend. Different context, different premise.

When it fails

A range breaks it completely. Price alternates between boundaries, so each swing closes beyond the last one in the opposite direction, and the tool produces breaks both ways on every leg. It is describing oscillation, not confirming a trend.

The second failure is entering at the break. By definition the move has already happened. On SPY’s daily chart the 20 sessions after an upward break were no better than any 20 sessions, so the break itself offered no extra reward for buying late.

A third is frequency. Each attempt is a full round trip, paying the spread in and out plus any commission. A one-minute chart producing eight of these in a session costs eight round trips, and the tool has to be right very often to survive that.

A fourth is redrawing the swing after the fact. Which high counts as “the” swing high is a judgment, and moving it once price has moved converts a pre-marked level into a post-hoc explanation.

And a fifth is treating a late-stage break as an early-stage one. The sixth consecutive break in a mature trend and the first break out of a base are the same event on the chart and different propositions.

The use that survives all of that is as a state, not a trigger. While breaks keep occurring in one direction, the trend is intact by definition. When one stops occurring and a change of character appears instead, the state has changed.

Read that way it costs nothing, because it is not a trade. It is a way of describing what the chart is currently doing, in language precise enough that two people looking at the same chart would describe it the same way — which is more than most technical vocabulary manages.

Related

Change of character is the opposite event and the one this is most often confused with. Market structure is the framework both sit inside, and it counts how often each one fires. Smart money concepts is where both terms come from.

The last opposing candle before that break is an order block, and without the break there is no block to mark.

This page is educational, not financial advice. Test every idea on your own charts before risking money. Some links on this page earn a commission if you buy through them. It costs you nothing and it does not decide what appears here or in what order — how these pages are made is set out in our methodology.