Imbalance in Trading: Chart Gaps, Footprint Imbalances and How Often They Fill

Michael Whitman · Trading since 2017, full-time since 2020 · YouTube creator, 137K subscribers · Published 2026-09-06 · Updated 2026-11-02 · Reviewed 2026-11-01

Covered on this page: TradingView.

An imbalance is a stretch of trading where one side swamped the other. On a candle chart it shows as a three-candle gap, read from the first candle’s wick to the third candle’s wick. On XLF’s daily chart since 1998, 83.20% of bullish gaps were re-entered within 20 sessions, against 84.89% of ordinary daily lows.

Three things called an imbalance

Search for the word and you get three separate ideas, often mixed up on the same page. They are measured from different data, and a reading from one tells you nothing about the others.

Most chart traders mean the first one, and most of this page is about it. The footprint version gets its own worked example below, because it is the one people confuse with the gap. The auction version is an exchange mechanism rather than a pattern: Nasdaq, for example, publishes its Net Order Imbalance information between 9:25 and 9:30 a.m. and between 3:50 and 4:00 p.m. Eastern time, according to its Opening and Closing Cross page (read 1 November 2026).

How it works

In a calm market, each candle overlaps the one before it. Buyers and sellers meet at nearly every price on the way up or down, so the bars share ground.

A gap appears when one bar travels far enough that its neighbors cannot touch. Take three candles in a row. If the first candle’s high sits below the third candle’s low, nothing in candles 1 and 3 traded in the space between them. That band is the bullish imbalance. For a fall it is the mirror: the third candle’s high below the first candle’s low.

The middle candle makes the gap. It is the long bar that covered the ground, and without it the outer two would overlap. Gaps between bars are where candlesticks mislead most: SPY’s first hourly bar of the session showed a color that disagreed with its change 31.64% of the time.

The measurement has no settings. Two prices from two candles, so two traders drawing the same chart should get the same box. The smart-money vocabulary calls the same three candles a fair value gap; the arithmetic does not change with the name.

It records an absence, not a presence. The band is where resting orders were too thin to slow the move. It does not say anyone is waiting there now.

A worked example: XLF in October 2026

The setup. XLF is the Financial Select Sector SPDR fund. On 14 October 2026 its high was $54.52. On 15 October it opened at $54.92 and closed at $55.27. On 16 October its low was $55.19 and it closed at $55.52.

The arithmetic. $55.19 is above $54.52, so the three candles leave a bullish band from $54.52 to $55.19. It is $0.67 wide, which is 1.21% of the $55.52 close.

What happened next. On 19 October XLF traded down to $55.03, inside the band, so it was entered after one session. Its lowest low over the next 10 sessions, through 30 October, was that same $55.03. The band’s $54.52 bottom was never reached.

The same week had a bearish one that went the other way. The 12 October low of $54.83 sat above the 14 October high of $54.52, a downward band of $54.52 to $54.83. On 15 October the high of $55.37 crossed the whole thing, so it was filled one session after it formed.

October was not unusual for this fund. Nine three-candle gaps by this rule were completed in its 22 sessions, seven bullish and two bearish. Of the eight with at least one later session, price came back into seven and crossed four completely by 30 October. One, the 19 to 21 October band from $55.50 to $55.64, was never touched: XLF’s lowest low over the next seven sessions was $55.75. Nasdaq’s own bars for the month agree with Yahoo’s to within half a cent and give the same nine gaps.

The footprint version, trade by trade

A footprint chart splits each bar into price rows and counts who started each trade. A trade that lifted the offer is an aggressive buy; one that hit the bid is an aggressive sell. A footprint imbalance compares a row’s aggressive buys with the aggressive sells one row lower, or a row’s sells with the buys one row higher, and marks the cell when one side is a set multiple of the other. This page uses 3 times; platforms let you pick the ratio.

The bar. Coinbase’s public feed holds 1,470 BTC-USD trades between 05:30 and 05:35 UTC on 2 November 2026. Rebuilt from those trades, the bar opened at $78,233.94, reached $78,289.00, dipped to $78,179.87 and closed at $78,289.00, the same four prices and the same 18.18 BTC as Coinbase’s own 5-minute candle.

Who was aggressive. Coinbase’s documentation says a trade’s side field is the maker’s side, so a “buy” there means the taker sold. Read that way, aggressive sellers took 13.05 BTC in 938 trades and aggressive buyers 5.13 BTC in 532 trades. The difference, called delta, was minus 7.92 BTC.

The diagonal test. In the $78,220 row sellers hit 4.00 BTC against 0.11 BTC bought one row higher, about 36 times. The $78,230 row (3.33 times), the $78,240 row (5.25 times) and the $78,270 row (3.82 times) also pass, so four sell cells, three of them stacked together. On the buy side, the $78,200 row (83.37 times) and the $78,190 row (44.76 times) pass, and so does the $78,180 row, where 0.26 BTC of buys faces a single sale of 0.00000007 BTC one row down.

And the bar closed $55.06 higher, at its high. Heavy selling and a stack of sell imbalances sat inside a bar that finished up. Many footprint tools offer a minimum-size filter for exactly the $78,180 case. Move the ratio and the count moves too: 11 cells pass at 1.5 times, 8 at 2 times and 5 at 4 times.

The original data

The sample. XLF daily bars from Yahoo Finance, 7,002 sessions from 22 December 1998 to 23 October 2026, weekly bars built from them (1,453 weeks), and hourly bars for 718 complete sessions from 27 November 2023 to 22 October 2026, with half days and one bad final session left out. On the hourly chart all three candles had to sit in one session. A gap counts as entered when a later bar trades back to its near edge within 20 bars, and filled when one reaches the far edge.

The base rate. The same 20-bar test on every bar’s own low: did a later bar trade at or below it? That happened after 84.89% of daily bars, 84.37% of hourly bars and 82.95% of weeks. Highs were retraded more often, 90.20% on the daily chart, in a fund that rose over the period.

Daily: 2,209 imbalances, 31.65 per 100 bars. Of 1,244 bullish ones, 1,035 (83.20%) were entered within 20 sessions and 860 (69.13%) filled; of 965 bearish ones, 868 (89.95%) were entered and 782 (81.04%) filled. The median band was 0.56% of the price for bullish gaps and 0.69% for bearish ones, and the median wait for an entry was one session.

Hourly: 633 imbalances, 12.65 per 100 bars, or about 0.9 a session against roughly one every three sessions on the daily chart. Of 330 bullish ones, 83.03% were entered and 73.94% filled; of 303 bearish ones, 92.41% and 87.46%.

Weekly: 356 imbalances, 24.88 per 100 weeks. Of 217 bullish ones, 81.57% were entered and 72.81% filled; of 139 bearish ones, 93.53% and 87.77%.

What it adds up to. On every timeframe the bullish share sat within 1.7 points of the base rate for any bar’s low, and below it. The bearish share sat close to the base for highs too, a little under it on the daily chart and a little over it on the hourly and weekly ones. Against a stricter base, only bars that closed higher and made a higher low, the bullish gaps came out slightly ahead instead: 81.91% of those daily lows were traded again, against the gaps’ 83.20%, and the difference stayed within 2.3 points on all three timeframes.

A band that price “often returns to” was revisited about as often as any recent bar was. Every imbalance is in the XLF imbalance file, the totals in the summary by timeframe, October’s bars in the October 2026 file and the Bitcoin bar in the footprint file and the trade list.

Checked against Nasdaq. Nasdaq’s XLF bars cover 2,514 of the same sessions, 24 October 2016 to 23 October 2026. Through the same code they gave 841 daily imbalances against 835 from Yahoo, 833 on the same date and side, with 84.42% entered against 84.43% and 71.82% filled against 71.86%.

How traders use it, and what the numbers allow

The usual plan is to wait for price to come back into the band and trade from it. The numbers above set the limits on that plan. A return is likely, but no more likely than a return to any recent low, so the return alone is not evidence that the band matters.

Entered and filled are different bets. A trade from the near edge assumes the band holds; a target at the far edge assumes it fills. On XLF’s daily chart 30.87% of bullish bands were still not filled 20 sessions later, and the 16 October 2026 band is one of them so far.

Timeframe changes the count more than the outcome. Per bar, the daily chart printed more gaps than the hourly one, because each daily window spans two overnight gaps. Per session, the hourly chart printed nearly three times as many, so a trader on the faster chart meets far more of them, and each attempt pays the spread and commission of a round trip.

When it fails

Waiting for a fill that never comes. Nothing obliges price to return. The money tied up waiting is money not used elsewhere, and “eventually” cannot be traded.

Treating a band as a line. The band has a top and a bottom. Price often trades into it and straight out the far side, so an entry at the first touch assumes the very thing being tested.

Marking every one. A fast chart with a box on every screen has a level everywhere, which tells you nothing. XLF’s daily chart alone gave nine in one month.

Ignoring the cause. A gap made by a scheduled announcement is the market repricing on news. One made in a thin overnight book is just thin liquidity. They look identical and do not have to behave alike.

Reading the footprint as a forecast. The Bitcoin bar above had four sell imbalances, three stacked, and closed at its high. Aggressive orders show who crossed the spread, not who will win the next five minutes, and a ratio test flags tiny cells unless you filter them.

Stacking correlated marks. An imbalance, an order block and a liquidity sweep drawn on the same region are three names for one area. Counting them as three reasons adds no evidence.

Questions people ask

What is an imbalance in trading?

On a candle chart, a band of prices a fast move skipped, read from the first candle’s wick to the third’s. The 14 to 16 October 2026 XLF example left a band from $54.52 to $55.19. On a footprint chart the word means one side out-trading the other at a price, and at an exchange auction it means unmatched shares.

How do you trade with an imbalance?

The common approach waits for price to come back into the band. Measure the return rate on your own chart first and set it against how often price returns to any recent low; on XLF the two were within 1.7 points.

Are imbalance and FVG the same?

On a candle chart, yes: the same three candles and the same two prices. A footprint imbalance is a different thing, read from trades rather than candles.

Do imbalances always get filled?

No. On XLF’s daily chart 69.13% of bullish bands and 81.04% of bearish ones were filled within 20 sessions, so a large minority were not.

Related

Fair value gap is the same measurement under the more common name. Order block is the level most often drawn alongside it, and an opening gap is the version that forms while the market is closed. And liquidity is what was missing from the band in the first place.

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.