Do Fair Value Gaps Get Filled? 273,191 FVGs Measured

The short answer

Fair value gaps fill, and quickly. Of 221,004 gaps on 15-minute charts, 87.3% were filled completely within 24 hours, and the median wait for price to come back to one was 30 minutes. On the 1-hour chart, 73.7% of 52,187 gaps filled within a day.

The number that matters is the next one. For every real gap we drew a fake one, the same size and the same distance from price but on the other side, where there is no gap at all. The fake gaps filled 87.2% of the time on the 15-minute chart and 74.0% on the 1-hour. A fair value gap does not pull price back. It fills because it sits close to price, and price revisits nearby levels all the time.

The textbook trade agrees. Buying or selling the first retest of the gap with the stop at its far edge and a 2R target, 189,423 times, won 26.2% and averaged −0.21R a trade before costs. An FVG is a precise map of where price just moved fast. It is not, by itself, an edge.

How often fair value gaps get filled

A gap counts as touched when price trades back to its near edge, as reaching consequent encroachment when it trades to the midpoint, and as filled when it trades through the far edge, measured from the bar after the gap is complete.

Two things stand out. Gaps are everywhere: every market printed about 13–17 of them a day on the 15-minute chart. And the first visit rarely stops at the edge or the midpoint. Once price got back to a 15-minute gap, it went through the whole of it within five days 97.4% of the time (94.1% on the 1-hour). As support or resistance, the far edge does not hold for long.

The test that matters: a gap that isn't there

"Most gaps get filled" sounds like evidence until you ask what else gets filled. A level a quarter of a bar away from price gets traded through constantly, gap or no gap. So for each real gap we built a placebo: the gap mirrored through the close of its third candle, the same distances from price on the opposite side. After a bullish gap the fake one sits above price, in the direction it was already moving. Then we walked both forward the same way.

The rows are almost identical. Price came back to the real gap no more often than it went on to the fake one, which means the gap itself carries no pull. The same is true of every split below: by size and by session, real and fake gaps never differ by more than 1.4 points within a day.

Size and session

Large gaps fill less often than small ones, and gaps from the New York morning fill least of any session: 82% within a day on the 15-minute chart against 90% for the Asian evening. It is tempting to read that as stronger gaps. The fake gaps say otherwise, moving in step every time: a large gap's far edge is simply further away, and the New York morning is when price leaves a level and keeps going. The hour-by-hour volatility shows the same clock.

The FVG retest trade

We fixed the trade before looking at any result, in the form it is usually taught. When the gap completes, place a limit order at its near edge and leave it working for 24 bars. The stop goes at the far edge, because a gap that fills completely has failed. The target is 2R, with 1R as a variant. A trade still open after five days is closed at the market. A bar that touched both the stop and the target is scored as a loss, since we cannot see inside it, and the best-case column shows the result had every one of those been a win instead.

A 2R target only needs to win one trade in three to break even, and this won 26%. The reason sits in the fill numbers above: when price comes back to a gap it usually keeps going straight through it, so the stop at the far edge is the most likely exit. Small gaps fared worst; their stop is a sliver of an ordinary 15-minute bar, and that is before spread, which on a 1.8 pips EUR/USD gap is a large share of the risk. Large gaps come out at about zero, which is what a price moving at random would score.

Every market

The markets are strikingly alike, which is itself the finding: a pattern that worked on some markets and not others would show up here as spread. The retest came out between −0.25R and −0.19R on every one, none of them clearly above zero, and no market's best case rose above −0.07R. The 15-minute figures are shown; the tool above switches to the 1-hour.

What this means if you trade fair value gaps

• Use the gap as a map, not a magnet. It tells you exactly where price moved fast. It does not tell you price will react there.
• Don't lean on the far edge. Once price is back inside a gap, it finished the job 97% of the time within five days. A stop just past the gap is the stop most likely to be hit.
• Skip the small ones. Small gaps were the worst trade by a distance (−0.44R), and that is before the spread eats a stop measured in a pip or two.
• Test what you add, against this baseline. The bare retest is worth zero at best. If your edge is a gap after a liquidity sweep, after a shift in structure, or inside a session window, measure that version and compare it to these numbers.

That is what the FVG backtester is for: gaps are drawn on the chart as they form, you replay it one candle at a time with the future hidden, apply your own rules and log every trade. The fair value gap guide covers the pattern itself, and the order block, previous day high and low and CRT statistics run the same kind of test on other ICT setups.

How we measured it

CRTLAB's own 15-minute price history, 1 Jul 2022 – 18 Sept 2026, for every market except the DAX (from 11 Mar 2025), and 1-hour bars built from it on clock hours. A fair value gap is three consecutive bars, with no session break between them, whose first and third bars do not overlap; a gap across a weekend or a closed market is a price gap, not a fair value gap, and is left out. Gaps smaller than a tenth of the average true range of the 14 bars before them are noise and are not counted. Everything is measured from the bar after the third, the moment the gap exists, using clock time: 1 hour is four 15-minute bars or one 1-hour bar.

The fake gap mirrors each real one through the third bar's close, keeping both distances. The trade is walked forward bar by bar, with no spread or commission: this measures the market's behaviour, not a broker. The definitions, the placebo, the trade and every split were fixed before the results were seen.

Frequently asked questions