Candidate
Fair value gap fill rates
Fair value gaps get filled — price almost always returns to close the imbalance.
A candidate test of how often three-bar fair value gaps are actually revisited, and over what horizon, once you fix a definition in advance.
Pre-registration
Not pre-registered.
How claims are specified, judged and corrected is set out in the methodology.
Results
No results yet — this claim is candidate.
Why this is only a candidate
The claim is popular and sounds falsifiable, but “fair value gap” is not defined consistently enough to test off the shelf. Before this could be pre-registered we would have to fix, in advance:
- The pattern definition. The common three-bar form (bar 1’s high below bar 3’s low, or bar 1’s low above bar 3’s high) has several variants that disagree about wicks, minimum gap size, and whether the middle bar’s direction matters.
- What “filled” means. First touch of the gap edge, a close inside the gap, or full traversal to the far edge give very different fill rates.
- The horizon. Almost everything fills eventually. A fill rate without a deadline is unfalsifiable, so the test needs a fixed horizon — same session, N bars, or N days — chosen before we look.
- The baseline. A high fill rate means nothing without a comparison. We would need a null, such as fill rates for randomly placed price windows of the same size and distance from spot.
Listing this here is not a commitment to test it. If we do pre-register it, the choices above will be published before any data is pulled.