Fair Value Gaps Fill No More Often Than Random Levels, Test Shows

A backtest of 15,490 fair value gaps on gold and Bitcoin found they fill no more often than arbitrary price levels at the same distance — stripping the pattern of its claimed predictive edge. The test, published by the analyst Indicatr on TradingView, compared gaps against a random-level control and found the gaps performed slightly worse in five of six data sets.

Fair value gaps are a staple of Smart Money Concepts trading. The claim: price leaves an imbalance behind, and sooner or later it returns to fill it. Indicatr wrote: "I tested how often that really happens on gold and Bitcoin, and compared it with what price does anyway."

What is a fair value gap?

The pattern is a three-candle formation. In a bullish gap, the low of candle three sits above the high of candle one — price moved through the zone so fast that it traded there in one direction only. A bearish gap is the mirror image: the high of candle three below the low of candle one.

The definitions matter. "Filled" means price trades back through the entire gap. "Touched" means it merely reaches the near edge. Traders often conflate the two; the test kept them separate.

How was the test built?

The study covered gold (OANDA spot) and Bitcoin (Bitstamp) across three timeframes:

• Daily data: gold from 2010, Bitcoin from 2014
• Four-hour data: gold from 2013, Bitcoin from 2017
• One-hour data: gold from 2023, Bitcoin from 2024

Every gap of at least 0.1 ATR(14) was included. The control measured how often price reaches any level at the same ATR-distance from the close within the same number of bars. If gaps genuinely pull price back, they should beat that baseline.

They did not. Within 20 bars, gaps were touched 82.3% of the time versus 82.5% for arbitrary levels; fully filled 68.7% versus 69.3%; fully filled within five bars 45.4% versus 45.8%. Every dataset landed within 2.5 percentage points of the control. Gaps get filled often — but only as often as their distance from price already predicts. The gap itself adds nothing measurable.

Do big gaps behave differently?

Yes, and in the wrong direction for gap traders. Small gaps under 0.5 ATR filled within 20 bars 76.5% of the time, against 76.2% for the control — a wash. Big gaps of 1 ATR or more filled only 41.1% of the time, versus 46.0% for random levels at the same distance. A big gap usually marks a strong move, and strong moves tend to continue.

The pattern's other promise — that a gap should hold as support on a retest — failed too. Indicatr tested entries at the near edge on first touch, with a stop just beyond the far edge and a target of twice the risk. Across 12,824 trades, 29.7% hit the target. Random entries with the same stop distance reached target 30.6% of the time. Trading the retest was, in plain words, no better than entering at random.

What do the live examples show?

The positive case: on October 30, gold's four-hour chart printed a 1.8 ATR bearish gap during the sell-off to 3,870. Gold turned and filled it completely on November 2, the next trading day.

The negative case: on September 25, a 1.4 ATR bearish gap formed between 4,232 and 4,280 on the same timeframe. Roughly seven weeks later it remained open, with gold trading about 175 points below its lower edge. Traders waiting for that fill waited through the entire decline.

What are the study's limits?

Indicatr is explicit about the caveats. The test used one definition — the three-candle gap with a 0.1 ATR minimum — and did not test extra filters such as displacement or market structure that some traders apply. Data came from a single CFD feed for spot gold and a single exchange for Bitcoin, with no transaction costs. Test windows overlap and gaps cluster, so the samples are not fully independent.

The practical guidance that survives the data: treat an open gap as a reference level, not a target price owes you; the closer the gap, the more likely the fill, exactly as with any other level; and be most skeptical of big gaps, which are the likeliest to stay open. Indicatr says the series continues with liquidity sweeps — reversal or trap — which promises another empirical check on a popular trading claim.