How big should a fair value gap be?
The three-candle rule has no minimum size, so it marks tiny gaps. Most traders add a minimum. This page shows what each threshold removes, counted on the archive, so the choice is made on numbers and not on habit.
Educational content, not advice. 1557 gaps formed 2026-10-12 to 2026-11-10, all markets together. A pip is 0.0001, 0.01 for yen pairs and 0.10 for gold, so one threshold means a different distance on different markets. Cells with fewer than 5 gaps show no figure. These are counts of what price did to gaps, not of any trade. Read every figure as indicative: it is a count about gaps, not a result for any trade, and a different period would count differently.
The short answer
There is no agreed size. The rule used here, wicks of candles 1 and 3 with no minimum, is exact, and a size filter is a choice laid on top of it. Educators choose in pips, as a share of the average range, or by the size of the middle candle. Each choice is a trade-off: a higher minimum leaves fewer gaps, and says nothing by itself about whether those gaps are worth acting on.
The table shows the first half of that trade-off for real: how many gaps stay at each pip threshold.
What each threshold leaves
| Minimum size | Gaps left | Touched within 4 candles | Filled within 4 candles | Sample |
|---|---|---|---|---|
| Any size | 1557 of 1557 (100%) | 731 of 1252 (58%) | 446 of 1252 (36%) | n 1557 · 2026-10-12 to 2026-11-10; watched 4 candles: n 1252 |
| 1 pip or more | 1096 of 1557 (70%) | 544 of 909 (60%) | 282 of 909 (31%) | n 1096 · 2026-10-12 to 2026-11-10; watched 4 candles: n 909 |
| 2 pips or more | 715 of 1557 (46%) | 363 of 602 (60%) | 159 of 602 (26%) | n 715 · 2026-10-12 to 2026-11-10; watched 4 candles: n 602 |
| 5 pips or more | 242 of 1557 (16%) | 123 of 199 (62%) | 30 of 199 (15%) | n 242 · 2026-10-12 to 2026-11-10; watched 4 candles: n 199 |
| 10 pips or more | 61 of 1557 (4%) | 28 of 52 (54%) | 2 of 52 (4%) | n 61 · 2026-10-12 to 2026-11-10; watched 4 candles: n 52 |
"Touched" means a later candle traded into the gap; "filled" means a later candle traded through its far edge (exact definitions). Only gaps watched for at least 4 candles after forming are in the touched and filled columns, so a young gap is not read as a miss.
Three ways to set a minimum
- A fixed number of pips. Simple, but a pip is a different distance on each market and each timeframe, so one number does not carry across. Compare sizes by timeframe on the statistics page before choosing one.
- A share of the average range. The threshold follows the market's own movement: the same share means a bigger gap in a wider-ranging market. It needs a range figure and a window for it, and both are choices.
- The middle candle. Some traders ask that candle 2 be large against its neighbours, a displacement idea. It filters on the move that made the gap and not on the gap itself.
Whichever you pick, write it down before you look at the chart, and keep it the same across the period you are studying. Changing the threshold after seeing which gaps "worked" only fits the rule to the past.
Cost matters more on small gaps
A gap a pip or two wide can be smaller than the spread you pay to enter. The table's low rows are the gaps where that is most likely, which is one reason traders drop them. It is a reason about cost, not a finding about how gaps behave.
Written by Michael H.