WhitmanTrading

BOS vs CHoCH

Break of structure is a close beyond a swing point in the direction of the current read, while change of character is the first close beyond one against it. The bars can look identical; only the direction relative to the read separates them.

These two are a genuine pair rather than two names for one thing. Both describe price closing through a swing level; what separates them is whether that close agrees with the read the chart was already giving. Put them side by side over three decades of SPY and a third idea appears that neither term covers on its own: the read, which starts at one change of character and ends at the next.

What each one is

A break of structure continues the current read. In a run of higher highs and higher lows, price closes above the last swing high, and the read stays up. Break of structure covers the event and what followed it on SPY.

A change of character is the first close against the read. In that same run, price closes below the last swing low instead, and the read flips to down. Change of character covers that event and whether the next break confirmed it.

The bars themselves can look identical. What decides the label is the read that preceded the bar, so the classification lives in the context rather than in the candle. Both rely on the same written swing rule: here, a swing high is a day whose high beats the five sessions before it and the five after it, so it is only known five sessions later.

Where they differ

Which way the close points. With the read or against it. That is the entire definition, and it is why neither can be named from one bar alone.

Where each sits in a read. A change of character always opens a read. Breaks of structure only happen inside one, after it has opened, so a read can hold none, one, or a long chain of them.

How far past the level each closed. On SPY’s daily chart both usually cleared their level by well under one percent, and the change of character by a little more; the data section has the medians.

How old the level was. The level a change of character breaks has usually been known for a little longer than the one a break of structure takes out.

A read, from start to finish

The read is the unit that makes both terms make sense together. It begins on the day a change of character flips the direction and ends on the day the next one flips it back. Everything in between is a string of breaks of structure, or nothing at all.

Counting them answers a question neither term page asks: once a read has opened, how many times does it continue before it flips? On a daily SPY chart that number was usually small, and the reads that ran long were rare enough to count one by one.

Direction mattered more than the label. Reads that opened upward collected more breaks of structure than reads that opened downward, because SPY rose over the period and upward breaks were more common in general. The data section has the counts.

A read is only complete in hindsight. While it is running, nobody knows whether it will hold one break of structure or eight. The count is a description of what happened, not a forecast of what the next read will do.

Which one to use: BOS or CHoCH?

Use the change of character when you need to know where the current read began. It is the start marker, and the break of structure only means something relative to it.

Use the break of structure count when you want to describe how far a read has run. None, one or two is the ordinary case on a daily chart; three or more is the exception.

Use neither as a forecast. On SPY, each further break of structure inside an upward read arrived about as often as a coin flip would suggest, so the count so far did not say whether another was coming.

And when you cannot say which of the two you are looking at, you are missing the swing definition. Write the rule down before reading the next chart, because without it the labels get applied after the outcome.

The original data

The sample. Yahoo Finance daily SPY bars from 29 January 1993 to 25 September 2026, 8,472 sessions, prices only. The rule is the one on the break of structure and change of character pages: 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. That rule produced 419 breaks, the same list the two term pages use. A read here runs from one change of character to the next, which gives 219 complete reads from 19 May 1993 to 1 September 2026.

How many breaks of structure each read collected:

Read Reads None One Two Three or more Median length
Upward 110 42 32 17 19 31.5 sessions
Downward 109 68 29 10 2 24 sessions

The None column is the same count the change of character page gives for changes that the next break reversed: both pages work from one list of breaks.

Horizontal bars counting SPY's upward and downward structure reads from 1993 to 2026 by how many breaks of structure each collected: 42, 32, 17 and 19 upward, 68, 29, 10 and 2 downward.
SPY reads, change of character to change of character, by breaks of structure collected, 1993 to 2026, five-bar daily swings. Source: Yahoo Finance, SPY daily bars (m67-spy-structure-reads-choch-to-choch-1993-2026.csv).

Long chains were rare and nearly all upward. Of the 21 reads with three or more breaks of structure, 19 were upward. The longest held eight. Downward reads mostly flipped back before a single break of structure came.

Once a read had one, the next was close to a coin flip. Of the 68 upward reads with at least one break of structure, 36 got a second (52.94%); of the 36 with two, 19 got a third (52.78%); of the 19 with three, 9 got a fourth. Tested against one half, none of those is distinguishable from chance (p = 0.72, 0.87 and 1.00). Downward reads ran the other way: 12 of 41 got a second (29.27%, p = 0.012), which most likely reflects a rising market rather than anything about the label.

Longer reads took longer, as they must. Across all 219, the median read with no break of structure lasted 18 sessions; with one, 36; with two, 54; with three or more, 105. Those four groups held 110, 61, 27 and 21 reads.

Four horizontal bars showing the median length of SPY structure reads from 1993 to 2026: 18 sessions with no break of structure, 36 with one, 54 with two and 105 with three or more.
Median sessions from one change of character to the next on SPY, by breaks of structure in between, 1993 to 2026. Source: Yahoo Finance, SPY daily bars (m67-spy-structure-reads-choch-to-choch-1993-2026.csv).

The label and the net move often disagreed. Measured close to close, from the change of character that opened a read to the one that closed it, SPY ended higher in 53 of 110 upward reads (48.18%) and in 86 of 109 downward reads (78.90%). That is mechanical rather than mysterious: a downward read ends when price closes above a swing high, and that swing high usually sits above the close that opened the read.

How far past the level, and how long after. The median break of structure closed 0.39% beyond its level and the median change of character 0.46%. A break of structure came a median 4 sessions after its swing became known; a change of character, 6.

The size of the break was no guide. Splitting the 219 changes of character at their median margin past the level (0.465%), 48 of the 109 bigger ones were followed by a break of structure, against 61 of the 110 smaller ones. A Fisher exact test of the two halves gives p = 0.11, so a wider close through the level did not make continuation more likely here, if anything the reverse. Every read, its count, length and close-to-close change is in the SPY structure reads file.

The video record. Of the 24,971 videos in this site’s search study, 47 name both terms in the title, at a median of 15,569 views.

The longest read in the sample

The eight-break read of 1995 shows what the rare case looks like. SPY closed at $46.73 on 13 January 1995, above the 27 December 1994 swing high of $46.41. The read had been down, so that close was a change of character, and it opened an upward read.

Eight breaks of structure followed, on 27 January, 10 March, 13 April, 15 June, 6 July, 27 July, 1 September and 19 October 1995, each a close above the latest known swing high. The first came ten sessions after the read opened.

The read ended on 10 January 1996, when SPY closed at $59.97, below the 19 December 1995 swing low of $60.58. That was 250 sessions later and 28.32% above the opening close: ($59.97 − $46.73) ÷ $46.73. The highest close inside the read had been 34.00% above it.

Nothing on 13 January 1995 marked this read as the long one. It began like the other 109 upward reads did, and 42 of those ended without a single break of structure.

When it fails

The failure is treating a long chain as the expected outcome of a change of character. The reads people remember are the ones like 1995, because they are the ones worth remembering. On SPY’s daily chart they were 19 of 110 upward reads. A method that assumes every flip opens a run is built on the exception, and it pays for each of the reads that flipped straight back.

The second failure is no written swing definition. Both labels then follow the outcome.

A third is reading a wide close past the level as confirmation. In this sample the wider changes of character were not followed by a break of structure more often.

A fourth is taking the read from one timeframe and trading on another. The reads on a five-minute chart and a daily chart disagree by design.

A fifth is a stop placed exactly at the break level. Everyone reading the same swing rule can see that price, and nothing in this data says price stays away from it once broken.

And a sixth is expecting the pair to give direction. They classify a break that already happened.

Break of structure covers the continuation break and what SPY did after one. Change of character covers the first opposing break and whether the next break confirmed it. And market structure shift is the other name the second one travels under.

The practical check

Count the breaks of structure in the current read before naming it a trend. None, one or two is the ordinary case on a daily chart; a long chain is the exception, and it is only visible once it has already happened.

— Michael Whitman

This page is educational, not financial advice. Test every idea on your own charts before risking money.