FVG vs imbalance vs liquidity void: what separates them and what our data shows

A fair value gap (FVG) is one kind of imbalance, not another name for it, and a liquidity void is a longer one-way move that can contain several of them. Used loosely, the three words make the same chart read three different ways. Below is what each one actually is, and what happened when we measured them on five years of our own hourly forex data.

Each one in one sentence

Imbalance is the family, not a synonym

The cleanest definitions we found come from LuxAlgo's indicator documentation, which describes imbalances as areas where "a great disparity between supply and demand" leaves price trading away from fair value. Among the types it lists under that heading are the fair value gap, the volume imbalance and the opening gap.

So when someone says "imbalance", ask which one. The FVG is the one the Smart Money Concepts (SMC) framework popularised, by way of ICT (Inner Circle Trader). If you want the full anatomy of that pattern, we cover it in our guide to the fair value gap.

The opening gap is the simplest of the three: two candles whose wicks do not touch at all. On our hourly data it is rare, 871 in five years across ten pairs, and most of them come where trading had paused, like the weekend open we deal with in our post on the weekend gap.

The volume imbalance barely exists in forex

For two candle bodies not to overlap, the second candle has to open away from the first one's close. Forex trades around the clock, so the next candle almost always opens where the last one closed.

On our data, 96% of hourly candles open less than one pip from the previous close. Of the 51,134 volume imbalances our scan found, 47,230 (92%) measure less than one pip, with a median of a tenth of a pip. That is the smallest step the price feed records, not a footprint of anyone's orders.

The ones of a pip or more are rare: about 12 for every 1,000 hourly candles, with a median width of 1.9 pips. On an hourly forex chart, treat a volume imbalance of a fraction of a pip as noise.

A liquidity void is a run of gaps

In its comparison of fair value gaps and liquidity voids, FXOpen describes a liquidity void as a significant, abrupt move between two levels "without the usual gradual trading activity in between", spanning "multiple candles and FVGs". The tutorials at innercircletrader.net, a third-party ICT tutorial site, put it in one line: a void "can contain two or more fair value gaps inside it".

Here is a real one from our data. On 17 February 2026, EUR/USD left three FVGs across five hourly candles, from the first candle's high of 1.18239 to the last one's low of 1.18431:

The three gaps add up to 9.5 pips; the void is 19.2. Each FVG band was crossed by a single candle, the middle one of its three; the rest of the range by two, both on their way up.

That is the difference in practice: the FVG marks the bands that only one candle crossed, the void marks the whole stretch the market crossed in one direction. Price came back into this void six candles after it formed and reached its far edge after 18 candles.

Most voids are short: 77% are just two FVGs back to back, 17% are three and only 5% run to four or more.

Do they really pull price back?

The usual claim is that price returns to these zones "to rebalance". Return rates on their own prove nothing, though: price wanders, and sooner or later it reaches almost any band near it. So we compared every zone with ten controls, bands of the same width taken at random hours nearby with similar volatility.

We built those controls two ways, because the answer turned out to depend on it. One puts the band at the same distance from the closing price as the real zone; the other places it right under the control candle's low (above its high for a bearish zone), the way a gap sits under its last candle. We also ran a placebo: the sub-pip volume imbalances, which we already know are noise.

For the FVG, the verdict flips with the comparison: 1.7 points above one control and 1.8 below the other, while even the placebo beats its controls by about a point. In other words, we found no sign that an FVG pulls price back any more than an ordinary band of the same size. Over 120 candles it is the same story: 87.0%, against 85.5% and 87.3%.

The liquidity void is different. It beats both controls, by 3.4 and 4.5 points at 24 candles (95% confidence intervals of +2.6 to +4.2 and +3.7 to +5.3), more than the placebo manages. Against the control under the candle it comes out ahead in all ten pairs, from +2.6 points (GBP/USD, the weakest) to +6.3 (USD/CAD).

Over 120 candles the edge shrinks to about 3 points: 81.8%, against 79.2% and 78.6%. FXOpen notes that voids "can remain open longer" than FVGs, and that shows up too: in absolute terms voids fill less often than FVGs. They are also more than twice as wide.

How we measured it

• Data: hourly (H1) candles from our own database for ten forex pairs (AUD/USD, EUR/CHF, EUR/GBP, EUR/JPY, EUR/USD, GBP/USD, NZD/USD, USD/CAD, USD/CHF, USD/JPY), January 2021 to April 2026: 329,283 candles. They come from a MetaTrader 5 broker's history, in the broker's server time, and the analysis was run on 6 October 2026; the code and random seeds are available on request.
• FVG: the low of candle 3 above the high of candle 1 (bullish), or the reverse (bearish), with three consecutive hours and no weekend in between. 45% of all FVGs are under 3 pips wide; we kept those apart, because a gap of a pip or two is comparable to the spread plus noise.
• Liquidity void: two or more FVGs in the same direction on consecutive candles. That is our own working rule, built from the descriptions above; nobody publishes an official threshold. Here FVGs of any width count, including those under 3 pips, and the ones of 3 pips or more inside a void also appear in the FVG row.
• Filled: price reaches the far edge of the zone within 24 or 120 candles. Candles, not hours, because a stretch can cross a weekend.
• Controls: for each zone, ten random hours within 500 candles on the same pair, each with an ATR (average true range) within 10% of the zone's own. The confidence intervals come from resampling whole pair-months, so one volatile month cannot carry the result.
• Tested vs found: zones in the last 120 candles of the data, and the few without ten matching controls, are left out of the fill test, so the tested counts sit slightly below the totals found (47,039 against 47,230 for the placebo).

What this does not measure: whether trading the retracement makes money. A fill is not a trade. Spread, entry, stop and the moves that never come back all decide that, and it would need a proper backtest, which this is not.

What changes in practice

• Say which imbalance you mean. "Imbalance" is the family; the FVG and the void are different sizes of the same idea.
• Ignore sub-pip volume imbalances on forex. They are a side effect of a market that never closes during the week.
• Do not trade a gap because it will "get filled". Most will, and so will most bands of the same size. The reason to act still has to come from market structure, an order block or your own plan, as we argued in our comparison of order blocks, supply and demand zones and support and resistance.
• A void is wider, so expect it to stay open longer. Within 24 candles, four in ten voids are still not completely filled.

The takeaway

An imbalance is any stretch the market crossed one-sidedly. An FVG is the three-candle version, a volume imbalance is the two-candle one that hardly exists in forex, and a liquidity void is a longer one-way run that can contain several FVGs. On five years of hourly data, an FVG gets revisited about as often as any band of the same size, and a void only a few points more often: they tell you where price has been in a hurry, not where it has to go.