Change of Character (CHoCH), Explained

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

Covered on this page: TradingView.

A change of character is the first close through a swing point that goes against the trend: a close below the last swing low while the read is up, or above the last swing high while the read is down. This page defines it with a fixed swing setting and then measures what followed every one on SPY’s daily chart since 1993.

How it works

A trend is a sequence. An uptrend is a run of higher highs and higher lows, and a downtrend the reverse. Each close beyond a swing point in the trend’s own direction keeps the sequence going.

A change of character is the first time the sequence breaks the other way. In an uptrend it is a close below the most recent swing low. In a downtrend it is a close above the most recent swing high. After it, the read flips: the next break in the new direction would be a break of structure for the new trend.

Nothing here works until the swing is defined. The definition used on this page: a swing low is a day whose low is below the lows of the five sessions on each side of it. It cannot be marked until those five later sessions have closed, so each level becomes usable about a week after it forms. A newer swing replaces an older one on the day it becomes known, and each level can be broken only once. This is the same rule as the market structure and break of structure pages, so the three pages count the same events.

Use the close, not the wick. Under this rule a day that trades below the level and closes back above it is not a change of character.

Marking it before it happens

The level is knowable in advance. While the read is up, the live swing low is the change-of-character level. It is fixed before anything happens, and price either closes through it or does not.

That is the honest strength of the idea. An indicator gives a reading after the bar; this gives a price in advance. The rest of this page asks whether crossing that price told you anything about what came next.

It looks exactly like a pullback when it happens. Both are price moving against the trend. A pullback resumes and makes a new high; a reversal does not. Nothing on the break day itself separates the two, which is why the data below tracks the next break rather than the break day.

A worked example

SPY, February 2020. The read was up after a break of structure on 6 February 2020. The live swing low was the 31 January low of $320.73, which became known on 7 February. SPY made a record close of $338.34 on 19 February. On 25 February it closed at $312.65, below $320.73: a change of character down.

What came next was a crash. The next break was a downward break of structure on 9 March 2020, a close of $274.23 below the 28 February swing low of $285.54. The lowest close within 60 sessions was $222.95 on 23 March, 28.69% below the change-of-character close. Note where the warning arrived: SPY was already 7.59% below its record when the level broke.

Now the same event in February 2018. The live swing low was $266.64, set on 29 December 2017. On 5 February 2018 SPY closed at $263.93, below it, and already 7.90% below its 26 January record close of $286.58. The down read was confirmed by a break of structure on 22 March. Yet the lowest close in the next 60 sessions was $257.47 on 2 April, only 2.45% lower, and 20 sessions after the break SPY was 3.39% higher.

Same rule, same signal, two very different outcomes. In both cases SPY was already more than 7% below its record when the level broke. A worked example like the 2020 one is easy to find in hindsight; the data section shows how often each kind happened.

The original data

The sample. 8,472 daily SPY bars from 29 January 1993 to 25 September 2026, prices only. With the five-bar rule there were 419 breaks in total: 199 breaks of structure and 220 changes of character, 110 downward and 110 upward. The market structure page covers how often they occur; this section asks what followed them.

Question one: did the next break confirm the change or reverse it?

Read the direction, not the label. In both rows the next break was upward 61.82% of the time. Across all 419 breaks, 252 (60.14%) were upward. So the next break after a change of character went up about as often as any break did, and the change itself added nothing measurable: tested exactly against 60.14%, p = 0.77. A rising market makes upward breaks more common, and that, not the change of character, set the odds.

The swing setting did not rescue it. With two-bar swings, the next break after a downward change was upward 59.45% of the time, against 60.27% for all 886 breaks (p = 0.84). With ten-bar swings it was 68.00% of 50, against 59.49% of 195 (p = 0.25). Neither is distinguishable from chance.

Question two: was a downward change of character a useful warning of a deeper fall? For each one, the lowest close in the next 60 sessions was compared with the close on the break day, and the same was done for every day in the sample.

The warning and the base rate were close. 17 of 109 downward changes were followed by a further 10% fall, p = 0.49 against the all-day rate; at the 5% line it was 34 of 109, p = 0.76. Twenty sessions out, SPY was higher after 78 of 109 (71.56%), against 63.91% for any day, p = 0.11: if anything a little more often, and still within chance.

Question three: did the big falls print one? SPY had five falls of 20% or more from a record close in the sample. Each printed a downward change of character before its low:

That is the case for it, and the cost of it. Every large fall printed a change of character while most of the fall was still ahead. But 21 of the 110 downward changes printed inside those five falls, and the other 89 printed outside them. A rule that acts on every one acts on all 110. Every event, its next break and its 20- and 60-session changes are in the SPY change of character file.

What it is not

It is not a reversal. It is one break. A reversal is a new sequence in the opposite direction, and on SPY’s daily chart half of all changes of character were undone by the very next break.

It is not an entry. Acting at the break means acting when the outcome is least settled. In both 2020 and 2018 SPY was already more than 7% off its record by the time the level broke.

It is not the same as a break of structure. That one breaks with the trend; this one breaks against it. The names are close enough that people mix them up, and mixing them up inverts the read.

When it fails

In a range it fires constantly. Price alternates between two boundaries, so the read flips back and forth and almost every leg produces a change of character in one direction or the other. That describes the range; it does not forecast anything.

Timeframe shopping is the second failure. Drop to a fast enough chart and a change of character exists at almost any moment you want one. The market structure page shows how the swing width alone moves the count: 12.9 a year with two-bar daily swings against 3.0 with ten-bar swings, on the same prices.

The third is reading it as a forecast. On SPY the next break after a change of character matched the market’s general upward lean and nothing more. The level tells you the old sequence stopped holding, not which sequence comes next.

The fourth is cost. Each change of character acted on is a round trip in spread and any commission. Since half of them were reversed by the next break, a rule that trades every one pays for many that come back.

The version worth keeping is a risk instruction. Mark the swing low while the trend is running. If price closes below it, stop adding in the old direction and check your stop. On this evidence that does not predict a crash; it limits how much you add at the moment the trend’s own definition has stopped holding.

Related

Market structure is the sequence this breaks and the page that counts how often it happens. Break of structure is the opposite event, and the pair only make sense read together. Trend reversal is what a change of character is sometimes the first sign of, and on SPY about half the time is not.

A change of character can leave an order block behind it, the last candle that went the other way before the break.

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.