Swing Highs and Swing Lows: Why Price Keeps Stopping in the Same Places

Pull up any price chart and let your eyes go slightly slack. You will see hills and valleys. Places where price pushed up, ran out of enthusiasm and rolled back over. Places where it fell, found a floor and turned around. The tops of the hills are peaks. The bottoms of the valleys are dips. You already knew this before you read the sentence, which is the point: the ideas underneath technical analysis are mostly things a five-year-old could see. The vocabulary is what makes them look hard.

A swing high is just a peak

So here is the plain version first, and the name second. A peak — a candle that stuck up higher than the ones either side of it — is called a swing high. A dip, a candle whose low reaches further down than its neighbours, is a swing low. That's the whole vocabulary lesson. (If "candle" is doing unfamiliar work there, our guide to reading a candlestick chart covers the atoms. For today all you need is that each little bar is one slice of time, with a high and a low.)

One quick disambiguation, because search engines cheerfully mash these two together: a swing high has nothing to do with swing trading. Swing trading is a holding period — days to weeks, rather than minutes or decades. A swing high is a shape on a chart, and it sits there whether you trade it, invest around it, or merely admire it. Same word, two jobs.

What makes a peak worth naming is what it records. At a swing high, the people buying ran out of money, nerve or interest, and sellers took the wheel. At a swing low, the reverse. Each one is the fingerprint of a decision that real people made at a real price — which is why they are worth marking, and why the rest of this article is about what happens when several of them land in the same place.

Stack a few peaks at one price and you get a level

One peak is a fact. Two peaks at the same price is a coincidence. Three starts to look like an agreement.

On the chart below, price pushed up into roughly the same area three separate times and got knocked back on every attempt. The first push topped out at 109.21. The second at 109.23. The third at 108.61. Somewhere just short of 109, three different rallies independently decided that was enough.

An area where rallies keep failing like this is called resistance — a ceiling. Its mirror image is support: an area where falls keep stopping, because buyers keep stepping in. Those are the two words the whole field is built on, and you now have both. Everything else in this article is a variation on them.

And here is where most explanations quietly mislead you. They draw resistance as a line — one crisp horizontal at one exact price — because a line is easy to draw and even easier to sell. Real price refuses to cooperate. Our three pushes stopped sixty cents apart, and the wicks poked straight through wherever a single line would have gone. What we actually have is a band, roughly 108.4 to 109.4. A level is a neighbourhood, not an address. Treating it as an address and then being cross when price misses by forty cents is a fast route to being technically right and practically annoyed.

Which raises the fair question: why does price stall in the same neighbourhood at all? There is a real answer, and it is refreshingly boring. An economist studying the foreign-exchange market got hold of something outsiders almost never see — the actual order book of a large dealing bank, the live list of prices at which real customers had left standing instructions to buy and sell. The orders were not spread evenly across the price range. They came in piles.

Two piles did most of the work. Take-profit orders — instructions to cash out a position that is already winning — clustered heavily at round numbers. That is a whole crowd deciding, quite independently of each other, that a round number is a nice place to get out. When price arrives, the selling is already sitting there waiting for it, and the push runs into a wall of it. Stop-loss orders — instructions to bail out of a position that has gone wrong — clustered just beyond those round numbers. So when price finally does force its way through, it immediately trips a pile of orders pushing it further in the same direction.

That is the same fact seen from two sides. A level holds because orders are parked at it, and a break accelerates because more orders are parked just past it. Support is not a force field. It's a queue.

How many candles either side? You decide

Now go back and poke the definition, because it has a hole in it. "Higher than the candles either side" — how many candles either side? One? Three? Ten?

Nobody can tell you. There is no official answer, no standards body, no correct number. You pick one, and the number you pick decides what counts as a peak. Traders call that setting the lookback: how many candles either side a candle has to beat before you'll call it a peak.

Set the lookback to one and every little wiggle qualifies. This chart coughs up thirty-two swings, which is a polite way of saying none of them mean much. Set it to four and only the serious turns survive: seven. Same candles. Same chart. Nothing changed except the size of the move you decided to care about.

This is why two traders can look at the same stock, mark completely different peaks, draw completely different levels — and both be right. They are not disagreeing about the data. They are answering different questions. Someone trading over an afternoon cares about turns that someone holding for six months would file under noise, and neither of them is confused. A level is not sitting in the chart the way a closing price is sitting in the chart. It is what you get when you point a rule at the data. Change the rule, change the level.

Drag the slider below and watch it happen. Push it all the way to four and something genuinely inconvenient shows up: one of the three pushes that built our level stops counting as a peak at all. Two touches, not three. The level just got thinner — not because the market did anything, but because you changed a setting.

Why a ceiling turns into a floor

Now the idea that makes any of this useful, and it fits in nine words: the ceiling of one floor is the floor above.

Stand in a room. The ceiling stops you going up. Go upstairs and that same slab is now the thing holding you up. Nothing about the slab changed. You changed sides.

Price does exactly this. Our band stopped three rallies dead. Then price broke above it and kept climbing. Later it fell back to that same band — and this time the band held it up. Same prices. Same zone. New job. Traders say the level flipped, or talk about polarity, but the plain version is the one that sticks: broken resistance tends to become support, and broken support tends to become resistance.

The mechanism is people, and it is slightly undignified. Three different groups converge on the same price. The traders who sold into that ceiling and then watched price run away without them would very much like another go at it. The traders who bought the break want to buy more, and this band is the last price that felt cheap. And the traders who bought the third failed push, sat underwater for weeks, and now just want out at the price they paid — well, that price is this one. Three different feelings, one price, and all three produce orders in the same place.

When the levels slope: trendlines

Everything so far has been flat — a ceiling at one price, a floor at another. But markets often don't stall at a fixed price. They stall at a price that keeps moving, a little higher each time. Buyers keep stepping in, but each time they do it a bit sooner and a bit dearer.

Draw a straight line touching a run of rising dips and you have drawn a trendline. It is the same idea as support, tilted. A trendline is support that's going somewhere. Do it through a run of falling peaks instead and you have the ceiling version, sloping down.

That also gives you the plainest definition of a trend there is, and it is made entirely of peaks and dips. An uptrend is a staircase: each dip stops higher than the last one, and each peak reaches higher than the last one. A downtrend is the same staircase upside down — lower peaks, lower dips. When neither is happening, when the peaks and dips are roughly level with each other, price is in a range, going sideways, and most of the fun is elsewhere.

The discipline with trendlines is drawing the line the chart offers rather than the line you'd prefer. A chart will politely agree with almost any line you are determined enough to draw, especially if you allow yourself to ignore the two touches that spoil it. If a line only works when you squint, it isn't a level. It's a hope with a ruler.

Breaks, retests, and the ones that lie to you

A break is simply a swing that failed to form. Price arrives at the ceiling as it has three times before, only this time the sellers aren't there in enough size, and it closes cleanly out the other side. No peak forms. The level has been broken, and traders call it a breakout on the way up and a breakdown on the way down.

Often — not always — price then comes back to the level and touches it from the new side. That's a retest, and it's where a lot of chart-readers prefer to act, because the level has now done its new job once, in public, where everyone could see it. A retest that holds is the strongest evidence a break was real that you can get without simply waiting longer.

Then there is the irritating half. Sometimes price pushes through, everyone concludes it's away, and it snaps straight back inside the range within a couple of candles. That is a fakeout, or a false break, and it happens constantly. The tells people lean on — a big decisive candle, heavy volume behind it, a proper close outside the band rather than just a wick — are tendencies, not guarantees. The only thing that reliably separates a breakout from a fakeout is what happens next, which is precisely the thing you don't have yet.

The honest bit: you can only spot a peak too late

Here is the part the cheat sheets leave out, and it is not a small part.

Read the definition once more. A peak is a candle higher than the ones either side of it. Either side. So you cannot know a candle is a peak until the candles after it have printed. If your lookback is four, that's four more candles of waiting. On a daily chart, that is four more days.

A swing high is therefore never a live signal. It is a label the chart only lets you attach once the moment has passed. And this — not discipline, not psychology, not your broker — is the real reason levels look so clean in a backtest and so slippery in the moment. Scroll back through any chart and the peaks are unmistakable. Of course they are. Everything after them has already happened.

The right-hand edge of the chart is a different animal. That newest candle might be a peak. It might be a pause before another push. The two look identical, because at that moment they are identical. Nothing distinguishes them yet.

Nothing fixes this. It is not a flaw in your method and there is no indicator that patches it — it is arithmetic. Confirmation costs candles, candles cost time, and by the time you are sure, part of the move has already gone. Everyone using structure is making the same trade-off in some ratio: act early and be wrong more often, or wait for confirmation and pay for it in a worse price. Anyone showing you a chart of beautiful, obvious levels marked after the fact is showing you the easy half of the job.

None of which makes levels useless. It makes them what they always were — a way to organise a chart into places worth paying attention to, and to decide in advance where you would admit you were wrong. That is a genuinely useful thing to own. It is just not a prediction, and it was never sold to you as one by anybody honest.