The fair value gap, and what it becomes when it breaks

A fair value gap is read across three consecutive candles. When the middle move is decisive enough, the first candle’s wick and the third’s do not overlap: a band of prices is left between them where, over that stretch of time, almost nothing traded.

That is the imbalance. Price moved so fast that some of the orders that would have wanted to trade there never got the chance. They are still waiting, and that is why the band pulls price back later.

Size matters: a gap of a few ticks on a low timeframe exists everywhere and means nothing. What makes an imbalance usable is a move that broke something on the way — otherwise you are drawing noise.

Price comes back into the band, and that is where the decision is made. A defensible entry waits for a reaction inside it — a visible rejection, a candle refusing to go further — rather than a limit order parked at the edge. The edge is your drawing, not the market’s.

Invalidation goes beyond the imbalance, not in the middle of it. If price cuts through the whole band without reacting, your read was wrong: either no orders were waiting there, or they have already been filled.

An imbalance already filled no longer counts. Same logic as a liquidity level: whatever was waiting was served on the first pass, and the band is now just a drawing on your chart.

Here is the part most explanations skip. When price closes through a bullish imbalance instead of reacting to it, the band does not disappear: it changes role. What was supposed to be support becomes a zone price tends to refuse when it comes back from below. That is what an inversion fair value gap is.

The mechanism is easy to picture: the buyers who were waiting in that band got caught out. Plenty of them have an exit order just above. When price comes back up there, it meets those exits — and that is no longer demand, it is supply.

In practice the inversion is worth waiting for because it gives you a dated level: you know exactly when it flipped, and the close that flipped it is your invalidation point.

The first is seeing them everywhere. On a low enough timeframe there is an imbalance every ten candles. The ones that count are those left by a move that broke structure: the rest is noise with a rectangle on it.

The second is confusing a close-through with a wick. A wick poking into the band and coming back out is not an inversion: it is exactly the reaction you were waiting for. An inversion requires a close on the other side.

The third is the most expensive: stacking bands. Three imbalances layered across three timeframes are not a signal three times stronger. Nothing measured on this approach says stacking improves anything — and a chart covered in rectangles will mostly help you find a reason to enter on every candle.