I lost an hour once arguing with a friend about whether a chart was in an uptrend. Same pair, same timeframe, opposite reads, and both of us could point at something on screen that backed us up. Turned out he was counting a different set of swings than I was. Neither of us had a rule for what counted as a swing in the first place, so we were debating vibes with extra steps. That is the entire problem market structure is meant to solve, and most people skip the part where they actually define it.
The pitch is simple. If you can label swing highs and swing lows with a mechanical rule, the trend of any chart becomes a statement of fact instead of an opinion. Two people applying the same rule to the same candles get the same answer, and everything downstream, break of structure, change of character, continuation versus reversal, falls out of that one discipline.
Swing points you can defend
The definition I use removes all judgment. A swing high is a candle whose high sits above the highs of the two candles before it and the two candles after it. A swing low is the mirror image. This is the old Williams fractal idea, and there is nothing magic about the number two. Some traders use three candles per side, which yields fewer and more significant swings. The number matters less than the commitment, so pick one and never change it mid chart.
Notice what the definition costs you. A swing point does not exist until two more candles have closed, so you are always confirming structure slightly after the fact. Beginners hate the lag because it feels late, and it is late on purpose, because the delay is the price of objectivity. Paying it is a lot cheaper than labeling a swing that later stops being one.
Once swings are mechanical, trend is just a sequence. An uptrend is higher highs and higher lows, each confirmed swing high above the previous one, each confirmed swing low above the previous one. A downtrend is the reverse. Anything else, higher highs paired with lower lows, overlapping swings, a broken sequence that has not rebuilt, is a range, and a range is a legitimate answer. Most charts spend more time ranging than trending cleanly, and forcing a trend label onto a range is where a lot of bad trades are born.
Break of structure versus change of character
These two terms get used interchangeably and they should not be, because they point in opposite directions.
A break of structure is price breaking a swing point in the direction of the existing trend. In an uptrend, that means taking out the most recent confirmed swing high, which is continuation evidence. The market is still doing what it was doing, and the pullback after a break of structure is the classic spot where trend traders look for entries, somewhere between the broken level and the last higher low.
A change of character is the first break against the trend. In an uptrend, that means price closing below the most recent confirmed higher low. This is the earliest structural evidence that the trend might be done, with heavy emphasis on might. On its own it is a warning light, and plenty of warning lights turn off by themselves. The reversal is only confirmed when the market follows through, prints a lower high, and then breaks another swing low. Until that second break, the honest label is a range that used to be an uptrend.
The detail that saves the most pain is deciding in advance what counts as a break. Wicks poke through levels constantly, and in crypto a stop run through an obvious swing low is practically a scheduled event. My rule is that a break requires a candle body closing beyond the level on the timeframe I am reading. A wick that pokes through and closes back inside usually means someone collected stops and the level held. You can pick a stricter convention if you like, one close beyond, two closes beyond, whatever suits you. Written down beforehand, any of them work fine. Decided in the moment, none of them do.
A routine two traders can agree on
The exercise only works if you do it the same way every time. Here is the version I follow, and it takes about two minutes per chart.
- Pick one timeframe and commit to it for the whole read. Structure exists on every timeframe at once and they disagree constantly. The daily can be trending up while the one hour is mid reversal, and both reads are true. Mixing them in a single sentence is how you end up confident and wrong.
- Fix your fractal length, two candles per side or three, and mark the last five or six confirmed swings. Ignore anything older unless price is about to test it.
- Write the sequence as letters. HH, HL, HH, HL is an uptrend. LL, LH, LL is a downtrend. Anything that does not read cleanly one way is a range.
- Note the two levels that matter right now, the most recent swing high and the most recent swing low. One of them breaking means continuation, the other means change of character, and you should know which is which before the next candle closes.
- State the trend in one sentence with evidence. Uptrend on the four hour, three ascending swing lows, last break of structure above the prior high, no change of character. If you cannot write that sentence, the honest label is a range and the honest trade is often none.
The failure modes are predictable. Timeframe hopping until you find the trend you wanted to see. Relabeling a swing after the fact because a trade went against you. Treating the small wiggles inside a pullback as the same order of structure as the swings that define the trend. That last one deserves its own rule, internal structure tells you about the pullback, major structure tells you about the trend, and keeping the two in separate mental boxes removes most of the confusion.
What this actually buys you
None of it predicts anything. A perfectly labeled uptrend can die on the next candle, and a change of character can turn out to be a fakeout that resumes the trend right after taking your stop. What structure gives you is a falsifiable read. You know what the chart is doing, you know exactly which level breaking would change your mind, and someone using the same rules can hand you back the same sentence.
It also makes ideas testable, which is where this gets useful beyond winning arguments. A rule like buy the pullback after a break of structure, exit on change of character, is mechanical enough to code, and running that through the strategy tester on Blockcircle has killed more of my clever structure ideas than it has confirmed, which is exactly what a backtest is for.
Start with one chart you actually care about. Fix the fractal length, mark the swings, write the one sentence. If the sentence comes out mushy, that is useful information too, and standing aside from a mushy chart is a habit worth building early.