Swing High and Swing Low: Definition, Settings and the Confirmation Lag
Michael Whitman
·
Trading since 2017, full-time since 2020
·
YouTube creator, 137K subscribers
·
Published 2026-09-05
·
Updated
·
Reviewed
A swing high is a candle whose high is higher than the candles on both sides of it. A swing low is the mirror — a candle whose low is lower than its neighbors. They are the turning points that market structure, trend lines and most stop placements are built from.
Covered on this page: TradingView.
A swing high is a candle whose high is above the highs of the candles on both sides of it; a swing low is a candle whose low is below its neighbors on both sides. Neither can be named until the candles to its right have printed, and on QQQ the wait cost a median 1.28% to 3.69% of price, depending on the setting.
Almost every structural idea on this site is built from these two points. Two sentences of definition, then the setting that decides how many you see, what that setting costs, and how to read one.
How it forms
A swing high is a candle whose high is higher than the candles either side of it.
A swing low is the mirror: a candle whose low is lower than its neighbors on both sides.
That is the whole definition, read straight off each candlestick’s high and low. Some tools require two, three or five candles either side rather than one, which produces fewer and larger swings. The number is a setting, not a law.
What a swing high records is where buying stopped. Price rose while buyers kept paying more, and at that candle they stopped, either because they were finished or because sellers arrived in enough size to end it. That is why the level draws attention later: some who sold there watch it, and some who bought just before it want out at breakeven. The liquidity page has the rest of that story.
What the setting actually does
Every tool that marks these has a number: how many candles either side must be lower. It is usually labeled length, strength or period, and some tools take separate counts for the left and the right.
A small number finds many swings. Most are pauses rather than turns, and the chart fills up.
A larger number finds a handful. Those are the turns you would have pointed at anyway.
The cost sits on the other side: a bigger number confirms later. A five-bar swing high cannot be known until five candles after the peak, so a cleaner chart is also a slower one. No setting gives you both.
The part that catches everyone
A swing high is not a swing high when it prints. It becomes one later.
Every swing point is confirmed late, by definition, because the definition requires candles that have not happened yet. Anything built on swing points, from structure to trend lines to most stops, inherits that lag. It is worth knowing so you do not go looking for a version that arrives on time.
A worked example
QQQ, 6 Oct 2026. The session traded up to $762.86. The five sessions before it reached at most $756.92 (5 Oct); the five after it reached $758.30, $757.18, $752.87, $749.79 and $744.20.
As a one-bar swing high it was confirmed at the close of 7 Oct, when the next candle’s high of $758.30 stayed below it. QQQ closed that day at $757.73, 0.67% under the peak.
As a five-bar swing high it was confirmed at the close of 13 Oct, five candles later. QQQ closed at $743.57, 2.53% under the peak. Same high, same chart: the stricter setting named it four sessions and almost two more percent later.
Then it held. QQQ did not trade above $762.86 in the nine sessions after 13 Oct, to the end of the data; it closed at $712.20 on 26 Oct. That is one case, chosen because it is recent, not evidence about swings in general.
A second example: the swing low that cost 8.49%
The 6 Oct high above was a quiet case. The low QQQ made in late July shows how expensive the wait can get when price leaves fast.
The candle. On 29 Jul 2026 QQQ traded down to $661.14 and closed at $661.73. The lows of the five sessions before it were $703.62, $687.79, $682.48, $675.95 and $667.88, each one lower than the last.
One bar each side. The next session’s low was $673.30, above $661.14, so the one-bar swing low was confirmed at the close of 30 Jul. QQQ closed that day at $683.55, already 3.39% above the low.
Five bars each side. The next five lows were $673.30, $680.05, $685.82, $707.53 and $716.92, so the five-bar swing low was confirmed at the close of 5 Aug. QQQ closed at $717.30, which is 8.49% above the low. The median for five-bar swing lows in the table below is 3.94%, so this one cost more than twice the usual amount.
Then it held. QQQ did not trade below $661.14 again through 26 Oct, the end of the data, and its lowest low after 29 Jul was $673.30 on the very next day. Like the first example, this is one case chosen to show the mechanics, not a claim about swing lows in general.
Read the chart left to right and the setting trade-off is visible. In these 64 sessions the one-bar rule marked 15 swing highs and 14 swing lows; the five-bar rule kept 6 and 5. The five-bar highs also tell a story on their own: $748.65 on 3 Jun, $745.45 on 22 Jun, $737.62 on 30 Jun and $728.54 on 5 Aug, each high lower than the one before, until $734.58 on 17 Aug broke the sequence, and $724.13 on 28 Aug was lower again. The five-bar lows were $686.37 on 9 Jun, $702.81 on 26 Jun, $700.91 on 8 Jul, $661.14 on 29 Jul and $702.70 on 24 Aug. That is the raw material the market structure page joins up.
The original data
Sample: QQQ daily bars from Yahoo Finance, 10 Mar 1999 to 26 Oct 2026, 6,951 sessions (27.63 years), split-adjusted highs, lows and closes. A swing high needs a high strictly above every high within N sessions on each side; highs that tie a neighbor are left out (54 at one bar, 16 at five). The distance is measured from the swing to the close of the session that confirms it, N sessions later.
| Bars each side | Swing highs a year | Confirmed below the high, median | Topped within 10 sessions of confirmation | Swing lows a year | Confirmed above the low, median |
|---|---|---|---|---|---|
| 1 | 54.61 | 1.28% | 72.88% | 55.23 | 1.73% |
| 2 | 31.41 | 1.94% | 64.59% | 33.15 | 2.39% |
| 3 | 21.86 | 2.26% | 59.37% | 24.00 | 2.91% |
| 5 | 13.75 | 2.97% | 51.98% | 15.02 | 3.94% |
| 10 | 6.41 | 3.69% | 33.52% | 7.49 | 5.68% |
The finding: the setting trades how many swings you see against how much price is gone before you can name one. Going from one bar to five cut the swing highs found from 54.61 to 13.75 a year and raised the median gap at confirmation from 1.28% to 2.97%. The quicker setting paid for its speed in pauses: 1,099 of 1,508 one-bar swing highs (72.88%) were topped within ten sessions of being confirmed, against 197 of 379 five-bar ones (51.98%) and 59 of 176 ten-bar ones (33.52%).
Swing lows were confirmed further away than swing highs at every setting, 3.94% against 2.97% at five bars. In other words, QQQ moved further from its lows in the days after them than from its highs, which describes one fund over one period, not a rule.
The check. Nasdaq’s own QQQ history, 26 Oct 2016 to 26 Oct 2026, finds the same 141 five-bar swing highs and the same 153 five-bar swing lows as Yahoo’s bars over those ten years, date for date. At one bar Nasdaq finds 554 highs to Yahoo’s 552, all 552 shared: its two-decimal prices split two ties that Yahoo’s unrounded prices keep equal. Every swing at every setting, with its confirming close, is in the QQQ swing points file.
Which ones are worth marking
Technically, a chart contains dozens. Practically, most are noise.
The test: did price turn there and go somewhere? If the high was exceeded two candles later, it was a pause, not a turn, and the table above shows how often that happens at each setting. The other test is the one on the timeframes page: if it vanishes when you combine candles, it was never a swing.
What they are for
Market structure is swing points joined up. “Higher highs and higher lows” is a sentence about swing highs and swing lows and nothing else, which is why this page comes first. A pullback runs between two of them.
They are also where invalidation usually sits. The last swing low under a long position is the price at which the sequence you identified stops being true. It is also the obvious price that many traders pick, which is why stops are commonly placed some distance beyond it rather than on it.
When it fails
The swing gives way
A swing low holding is what an uptrend is. A swing low breaking is that sequence ending, and it is the event that stops placed against it are there for.
You marked thirty of them
A chart with a mark on every turn is a chart you cannot read, and it will always contain a swing point supporting whatever you already wanted to do. At one bar each side QQQ produced about one swing high a week.
You are waiting for confirmation that cannot come early
Covered above and worth repeating, because it is where beginners look for a fix that does not exist. The lateness is the definition, and on QQQ it cost a median 2.97% at five bars.
You found it afterward
Every swing point is obvious with the right-hand side of the chart visible. At the candle itself it is a guess, and it stays a guess for as many candles as the setting asks for.
Questions people ask
How many candles make a swing high?
As few as one on each side, which is the strict definition. Charting tools let you raise it; on QQQ’s daily chart one bar each side found 54.61 swing highs a year and five bars found 13.75.
Why does a swing high take so long to confirm?
Because the definition needs the candles after it. A five-bar swing high is only known five sessions later, and on QQQ the close on that day sat a median 2.97% below the high.
Do swing highs act as resistance?
Often they are where traders draw it, but many give way quickly. On QQQ, 51.98% of five-bar swing highs were traded above within ten sessions of being confirmed, and 72.88% of one-bar swing highs.
Should I use the same setting on every timeframe?
The table measures daily bars only. On any timeframe the same trade-off applies: more bars each side means fewer, later swings, so the right number depends on how many turns you want to see.
Is a swing high the same as a fractal?
Nearly. A Williams fractal is a swing high or low with two bars required on each side, so it is the two-bar row of the table: on QQQ’s daily chart that setting found 31.41 swing highs a year.
Do swing high and low indicators repaint?
They draw the label on the peak candle, but only after the right-hand candles have printed, so on a finished chart the mark looks as if it was there in real time. It was not. A five-bar label on QQQ appeared five sessions late, with the close a median 2.97% from the high.
Related
Market structure is the next page: swing points joined up are what a trend is made of.
Pullback is what happens between two swing points, and its higher-low test is this page applied in sequence.
And risk management is where a swing low turns into a stop and a position size.
I do not use the swing-point setting on indicators, and that is a deliberate choice rather than laziness. Turned on, it marks every high and low it can find and clutters the screen, and I trust the levels I draw from order blocks more than a label the tool put there. The turning points that matter are usually the ones you would have pointed at anyway.
— Michael Whitman
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.