Every FVG video ends with the same sentence.
"Price always comes back to fill the gap."
Say it slower. Always.
In a business where nobody can promise you Tuesday, an entire industry promises you always — about a three-candle pattern.
The pattern is real. The footprint is real. The always is marketing.
A fair value gap is a footprint of one-sided execution. It is not a debt the market owes you.This piece is the honest version of the most-searched question in SMC — what is a fair value gap? — answered mechanically: what the footprint actually records; why gaps do fill often, for a boring reason that has nothing to do with magnetism; when a gap is worth trading and when it's decoration; and a test you can run on your own chart to replace always with a number you measured yourself.
Two requests before we start:
- Save this for the next time a gap "had to fill" and didn't.
- Send it to one trader who just discovered ICT clips at 2 a.m.
Skip this if you already mark gaps
Sixty-second version: a fair value gap is the three-candle footprint left when execution ran one way faster than the other side could respond — candle two displaces so hard that candle one's high and candle three's low never overlap, and the unshared span between those wicks is the gap. It marks inefficient execution: trades that printed with thin opposition. Gaps often get revisited, but mostly for the boring reason that markets oscillate and retrace through recently traveled prices — not because imbalance emits gravity. The always fills promise survives because gaps that filled are easy to screenshot and gaps that never filled quietly age off the chart. A gap earns trade consideration the same way any zone does: displacement quality, freshness, location in structure, regime — the same grading discipline as order blocks. And the only fill-rate worth trusting is the one you counted yourself, on your instrument, your timeframe, with the misses included.
| What traders think | What's actually true |
|---|---|
| "Price always fills the gap" | Price often revisits recent prices — that's oscillation, not obligation |
| "The market seeks fair value" | The market seeks liquidity; "fair value" is a story stapled on afterwards |
| "A gap is a magnet" | A gap is a footprint; footprints don't pull |
| "Bigger gap = stronger signal" | Bigger gap = stronger displacement; meaning depends on context |
| "FVGs work / don't work" | Ungraded gaps behave like ungraded anything: chance, minus costs |
I — What a fair value gap actually is
The mechanics, stripped clean. Markets fill orders by matching buyers and sellers across a spread. When conviction arrives fast — news, a stop cascade, a large participant done waiting — one side consumes everything available across a span of prices in minutes or seconds. The candle that prints is long and one-sided, and the auction moved so fast that the candles on either side of it never traded back through part of its range: candle one's high and candle three's low don't touch. That untouched span is the fair value gap.
What it records is real and useful: execution ran one way with thin opposition. Somebody paid up, repeatedly, without the usual back-and-fill. That's evidence of urgency, and urgency is information.
What it does not record is an appointment. Nothing in market microstructure obliges price to return to prices where execution was thin. The gap is testimony about the past — "buyers ran through here unopposed" — not a contract about the future.
II — Why gaps fill — the boring truth
Here's the part the always crowd gets half-right, which is what makes the myth durable: gaps really do get revisited, a lot. But watch the mechanism.
Markets oscillate. Even strong trends retrace; even breakouts get retested; ranges get traversed top to bottom. Any price the market visited recently sits in the path of ordinary mean-reversion — and a gap, by definition, sits inside recently traveled prices. So retracement through a gap is the base case of normal price movement, the same way rain hits your street without aiming at your house. Add the definitional trick: the wider the retracement window you allow ("it filled — three weeks later"), the closer any fill-rate creeps toward certainty, because eventually revisiting a nearby price is what oscillating markets do.
Gaps don't pull price back. Price wanders back through gaps because wandering is what price does.(Yes, belief can be mildly self-fulfilling — enough traders watching the same gap can nudge a fill into happening. Notice what that argues for: measuring your market's actual rate, not trusting always.)
That distinction sounds academic until you're sizing a position on it. "The market must come rebalance this" justifies conviction, patience, doubling down. "Price often retraces through recent territory, on no particular schedule" justifies exactly what it says — a conditional setup that needs context to be worth money, and an invalidation you'll respect.
III — The always problem
The fill-the-gap promise survives on two selection tricks, and you should be able to name both.
Screenshot survivorship. Filled gaps are photogenic — a shaded box, an arrow returning, a satisfying tap. Unfilled gaps are invisible: they age, scroll off the left side of the chart, and no one posts them. The evidence you see for the magnet theory is pre-filtered to confirm it — the same survivorship trick we dissected for zone indicators, wearing a different costume.
The missing deadline. "It always fills" comes with no when. A promise with no deadline is unfalsifiable — and untradeable. Your stop-loss has a location; your margin has a clock; "eventually" has neither. Any claim that can't be wrong can't fund a position.
IV — When a gap is worth trading
Strip the magnetism and the FVG still earns a place — as evidence inside a graded setup, not as a standalone prophecy.
As a displacement record. The gap certifies the move that made it: real urgency, thin opposition. A zone or structure break accompanied by an FVG carries better evidence of participation than one without.
As a retrace map. If you already have directional context — trend, structure, a graded zone from the order-block framework — the gap marks where a pullback finds thin prior trading, where entries with structural invalidation live. You're not trading the gap; you're using the gap to time something that already deserved a trade.
As confluence, never as cause. Gap + fresh zone + structure + live regime is a setup. Gap alone is a shaded rectangle and a hope. The five-property grading discipline from the order-block piece applies verbatim — displacement, freshness, location, regime, liquidity — because an FVG is just another zone whose grade decides whether it matters.
Use the gap to time a trade that already earned its thesis — never as the thesis.V — Measure your own fill rate
Twenty minutes replaces always with a number. No indicator required — a notebook works.
Minutes 0–5 — Define fill, in writing. Full overlap or midpoint touch? Within how many bars? Your numbers are meaningless until "filled" has one fixed definition — the missing deadline is where the myth hides.
Minutes 5–10 — Mark blind. Scroll back three months on YOUR instrument and timeframe. Mark every FVG by the wick rule alone — no cherry-picking, including the ugly ones. This is the step the screenshots skip.
Minutes 10–15 — Count. For each gap: filled within your window, or not? Write the fraction down. Then split it: fills with trend context vs against; fresh gaps vs stale. Watch the useful pattern appear — it's the splits, not the headline rate, that trade.
Minutes 15–20 — Price the claim. Whatever your measured rate is, ask: does a setup this frequent, with this fill window, beat its costs with a stop I can actually hold? That question — never does price fill gaps — is the tradable one.
Where the SMC toolkit sits
Our own ledger — caveats first. No gap, graded or not, is a promise — and a tool's job is to grade evidence, not to sell magnetism. Our SMC AI-Scored Toolkit treats FVGs exactly the way this article argues they deserve: every order block and FVG scored 0–100 on five transparent factors, re-weighted to the live regime, weak ones dropped before they reach your eyes — a gap only surfaces when its grade and context qualify, and the dashboard's verdict (ARMED / MANAGE / STAND-ASIDE) says so in plain English, with stand-aside as a first-class answer. Zones commit on the closed bar and don't repaint, and the toolkit shows its Resolved % — how its own zones actually held, recomputed on your chart with the misses left in — so the survivorship trick from §III isn't available to us either. It scores; it never predicts. Run the fill-rate test from §V against anything it shows you — that's the point of §V.
Disclosure
This is education about a footprint, not a promise about returns. Fair value gaps record one-sided execution; they don't obligate the market to anything, and neither does any tool that draws them — ours included. The one question that sorts gap educators instantly: "What's the fill rate, over what deadline, measured on what sample — and where are the unfilled ones?" If the answer is a highlight clip, you have your answer.
Your first 20 minutes
Run §V on your own chart tonight — three months, one instrument, one written definition of "filled." Whatever number you get, you now know more about fair value gaps on your market than the person who taught you always. Then look at how many of the filled ones happened inside trend context with a graded zone nearby — that intersection, not the gap itself, is where the tradable thing lives.
The gap was never a magnet.
It was a witness.
Witnesses tell you what happened.
Grade what's happening now — and let the witness testify, not promise.
