Fair Value Gap: What Filled Has to Mean Before You Test It

Ask whether fair value gaps get filled and you will be given a number. Ask what filling one means and the number stops being checkable. The two questions are usually answered as if they were one, and the first page of results for this term answers the second one nowhere at all.

That is the gap worth closing first. A fair value gap is easy to draw and easy to argue about, and almost every argument about it collapses into a disagreement over the fill rule nobody wrote down.

What Filled Has to Mean Before Any Fill Claim Can Be Checked

A fair value gap is a zone with two edges. Price returning to it can mean three different things, and the three are not interchangeable.

Under a first-touch rule, the gap is filled the moment any wick reaches the near edge. Under a midpoint rule, price has to reach the middle of the zone. Under a full-traverse rule, price has to cross the far edge, which means the whole zone has been retraded. On one chart, on one instrument, the same set of gaps will produce three different fill counts depending on which of these three you picked before counting.

The second missing term is the window. A gap that is untouched after ten candles and reached after four hundred is filled under one measurement and unfilled under another, and neither answer is wrong until somebody states the horizon they used. Fill claims are usually written without either term, which makes them unfalsifiable rather than accurate.

The divergence between the three is not a technicality that averages out. Consider a zone that price approaches once, prints a wick two ticks into the near edge, and then leaves for good. That zone is filled under a first-touch rule, unfilled under a midpoint rule, and unfilled under a full traverse. One event, three verdicts, and the only thing separating them is a choice made before the observation.

Any set of zones will contain a large share of exactly that kind of shallow return, which is why the criterion moves the headline number so much.

So the first thing to fix is not the entry, the stop or the timeframe. It is the definition. Write down the criterion and the window before you count anything, and any figure you produce afterwards can at least be checked by somebody else. This is the same discipline that makes a backtest worth reading: a result is only as meaningful as the rule that generated it.

Where the Three-Candle Rule Comes From

The construction is mechanical. Take three consecutive candles. In the bullish case, compare the high of the first with the low of the third: if the third candle’s low sits above the first candle’s high, the distance between those two prices is the zone. The bearish case is the same comparison read the other way, with the third candle’s high below the first candle’s low.

The middle candle decides nothing about where the boundaries sit. Its job is to be large enough to carry price from one side to the other, and its own high and low never bound the zone. That is worth stating plainly because most descriptions of the pattern lead with the middle candle, and a reader who draws the zone around it draws the wrong box.

Both edges are extremes of a period, which means the pattern is a statement about the chart you selected as much as about the instrument. Our page on the inside bar pattern works through what that dependence does to a two-candle comparison, and the same reasoning applies here without needing to be repeated.

The vocabulary question, of whether this zone is a genuinely new object or an older idea under a newer name, is argued out on our page on where these terms came from.

Price Traded Through All of It

The word gap invites a comparison that does not hold. In a true price gap, trading stops at one price and resumes at another, and no transaction happened at any level between them. That is what a weekend break or a post-announcement reopen produces, and it is covered on our page on gaps where no trading happened.

A fair value gap is not that. While the middle candle formed, price moved continuously from one end of its range to the other, and every level inside the zone was traded. What did not happen is an overlap between the wick of the first candle and the wick of the third. The zone marks an absence of overlap between two non-adjacent periods, not an absence of transactions.

The distinction matters for what you can infer. An untraded price range tells you something about liquidity that simply was not available. A traded range with no wick overlap tells you the move was quick relative to the period length, and quick relative to a period is a function of the period you chose.

How to Check a Fill Claim on Your Own Chart

Any fill figure you read can be tested, and testing it is more useful than believing it. The procedure below is deliberately small, because a procedure that takes an afternoon does not get run.

Fix the instrument, the chart period and a date range before you start, and do not change them once the count begins. Then state the criterion from the table above and the window in candles. Both go in writing, before the first gap is marked, so that the result is not quietly adjusted to a preferred answer as the count proceeds.

Mark every zone the rule produces in that range, including the ones price ignored. Skipping the ignored ones is the single easiest way to produce a high fill rate, and it happens without any intent to mislead, because unfilled zones leave nothing on the chart to draw attention to themselves.

Keep a note of anything you excluded and why. Zones cut off by the end of the date range have not had their full window and belong in a separate line rather than in the unfilled column, and zones you discarded as too small to trade are a judgement that changes the population being measured. An exclusion that is written down is part of the method. An exclusion that is not written down is the reason two people counting the same chart disagree.

Then count. A rate is the number of zones meeting your criterion inside your window, divided by every zone the rule produced. If you repeat the exercise with a second criterion, keep the two numbers separate rather than averaging them, and label each with its criterion and window when you write it down. Two numbers that measure different things do not combine into a better one.

What an Inverse Fair Value Gap Assumes

An inverse fair value gap is the same zone, read after price has passed all the way through it. The claim attached to it is that a zone which failed to hold in one direction becomes relevant in the other, so a bullish zone that price traversed downward is watched afterwards as resistance.

The assumption underneath is worth naming, because it is not a mechanical property of the construction. It is a behavioural claim: that participants who acted at a level once will act at it again with their positions reversed. Nothing in the three-candle rule implies that, and the rule cannot be used to support it. The inversion is a separate idea sharing the same drawing.

That does not make it useless, but it does change what evidence would settle it. A fill rate says nothing about whether inverted zones hold, because holding and filling are different events. Testing an inversion needs its own criterion, its own window and its own count, defined before the test rather than after it.

There is a second thing to settle before testing one. An inverted zone and an ordinary support level that broke are, on the chart, the same picture. If your rule for the inversion would also mark every broken level, then it is not measuring the inversion at all, and a result from it says nothing about fair value gaps specifically. Deciding in advance what would distinguish the two is the part of the test that makes it worth running.

What Nobody Publishes About Fill Rates

Fill percentages circulate widely and none of them arrives with a method. Across the pages that rank for this term, the figures that appear carry no dataset, no instrument, no date range, no criterion and no window, and several carry no source of any kind. A figure without those five things cannot be reproduced, which means it cannot be checked and should not be repeated.

Most of those pages also carry no publication or update date of any kind, so a reader cannot tell whether a figure was produced this year or five years ago. Of the six pages read while this one was written, one showed a dated update and one carried a byline two years old with no revision date. The remaining four showed neither.

There is also no institutional source to fall back on. Exchanges publish trade and quote data, and regulators publish rules and disclosures, but neither publishes chart-pattern statistics, and the pattern is not defined in platform documentation either.

The term does not appear in the MetaTrader 5 timeseries documentation or in the chart-settings help pages checked while this page was written.

What the platform documents is that each period stores an open, high, low and close, and that the time value a terminal works from is formed on the trade server rather than by the clock on your own computer. Two brokers running different server offsets cut the same price history at different moments, which is one more reason a rate measured on one feed does not transfer to another. Our page on the clock your candles follow sets out where that offset comes from.

The practical consequence is that the only fill rate you can trust is one you produced yourself, on the instrument you trade, with the rule you wrote down. That is a smaller claim than the ones in circulation, and it is the only one that survives being questioned.

Which Figures This Page Does Not State, and Why

This page states no fill rate, no win rate, no recommended timeframe and no typical number of candles to a fill. Every one of those appears on pages ranking for this term, and every one that we found was stated without an attributed source, a dataset or a method.

One further caution belongs here. Loss percentages published by brokers are firm-specific regulatory disclosures covering that firm’s own retail clients over a defined period. They are not a market-wide measurement of how traders perform, and a page that reuses one as if it were a general statistic has changed what the number means. Where you see a round figure for how many traders lose, check whether it began life as one firm’s disclosure.

Our standing rule is that a figure appears here only when an official source states it, or it does not appear at all. For this topic that leaves the mechanism, the definitions and the method, which are the parts that were actually missing.

Who This Page Is Not For

If you want an entry rule, a stop distance and a target for trading these zones, this page does not contain one and is not an attempt to write one. It also will not tell you whether the pattern works, because that question cannot be answered without the criterion and window discussed above, and answering it for you would defeat the point.

Before you test any of this, decide which criterion and which window you are committing to, confirm that your platform’s zone tool uses the same boundaries described here rather than boundaries of its own, and check the results against a second instrument before you conclude anything. Three checks, run once, are worth more than a figure repeated a hundred times.

Sources checked 15 August 2026: MetaQuotes MQL5 Reference, Timeseries and Indicators Access documentation · MetaQuotes MQL5 Reference, TimeCurrent date and time function documentation · MetaQuotes MQL5 Reference, ENUM_TIMEFRAMES chart period constants · MetaTrader 5 user guide, chart settings help pages