Which timeframe for fair value gaps?
There is no best timeframe, and this page does not pick one. It sets out what changes as the candle gets longer, and counts it: how many gaps each timeframe forms, how large they are and how soon price comes back.
Educational content, not advice. 5059 gaps formed 2026-10-12 to 2026-11-10, from the delayed delayed feed. A timeframe with fewer than 5 gaps shows no figure. These are counts of what price did to gaps, not of any trade, and they are indicative only.
What a timeframe changes
The same three-candle rule runs on any candle length. Three things move with it. The gap spans more price, so it is wider in pips. It takes more time to be reached, because "within 4 candles" is 20 minutes on 5m and four days on 1D. And there are fewer of them: a day holds many more 5-minute candles than daily ones, so the shorter timeframes form far more gaps.
Many traders read the gap on a longer timeframe and look for confirmation on a shorter one. That is a method, not a finding, and the counts below do not test it.
Gaps by timeframe
One row a timeframe, all markets together. The "within N candles" columns count candles of that timeframe: 4 candles are 20 minutes on 5m and 16 hours on 4h.
| Timeframe | Gaps | Not filled so far | Size in pips: median (10th to 90th) | Touched within 4 candles | Filled within 4 candles | Touched within 8 candles | Filled within 8 candles | Median time to first touch |
|---|---|---|---|---|---|---|---|---|
| 5m | 3000n 3000 · 2026-10-28 to 2026-11-10 | 346 of 3000 (12%)n 3000 · 2026-10-28 to 2026-11-10 | 1.1 (0.2 to 4.2)n 3000 · 2026-10-28 to 2026-11-10 | 1140 of 2233 (51%)n 2233 · 2026-10-28 to 2026-11-10 | 573 of 2233 (26%)n 2233 · 2026-10-28 to 2026-11-10 | 1096 of 1906 (58%)n 1906 · 2026-10-28 to 2026-11-10 | 634 of 1906 (33%)n 1906 · 2026-10-28 to 2026-11-10 | 10 minn 2783 · 2026-10-28 to 2026-11-10 |
| 15m | 1557n 1557 · 2026-10-12 to 2026-11-10 | 190 of 1557 (12%)n 1557 · 2026-10-12 to 2026-11-10 | 1.7 (0.3 to 6.3)n 1557 · 2026-10-12 to 2026-11-10 | 731 of 1252 (58%)n 1252 · 2026-10-12 to 2026-11-10 | 446 of 1252 (36%)n 1252 · 2026-10-12 to 2026-11-10 | 758 of 1132 (67%)n 1132 · 2026-10-12 to 2026-11-09 | 517 of 1132 (46%)n 1132 · 2026-10-12 to 2026-11-09 | 30 minn 1449 · 2026-10-12 to 2026-11-10 |
| 1h | 389n 389 · 2026-10-12 to 2026-11-10 | 103 of 389 (26%)n 389 · 2026-10-12 to 2026-11-10 | 3.7 (0.5 to 12.4)n 389 · 2026-10-12 to 2026-11-10 | 174 of 329 (53%)n 329 · 2026-10-12 to 2026-11-09 | 98 of 329 (30%)n 329 · 2026-10-12 to 2026-11-09 | 167 of 275 (61%)n 275 · 2026-10-12 to 2026-11-09 | 116 of 275 (42%)n 275 · 2026-10-12 to 2026-11-09 | 2 hn 321 · 2026-10-12 to 2026-11-09 |
| 4h | 102n 102 · 2026-10-12 to 2026-11-10 | 45 of 102 (44%)n 102 · 2026-10-12 to 2026-11-10 | 9.1 (1 to 25.5)n 102 · 2026-10-12 to 2026-11-10 | 35 of 66 (53%)n 66 · 2026-10-12 to 2026-11-09 | 26 of 66 (39%)n 66 · 2026-10-12 to 2026-11-09 | 40 of 52 (77%)n 52 · 2026-10-12 to 2026-11-06 | 34 of 52 (65%)n 52 · 2026-10-12 to 2026-11-06 | 8 hn 66 · 2026-10-12 to 2026-11-09 |
| 1D | 9n 9 · 2026-10-20 to 2026-11-09 | 2 of 9 (22%)n 9 · 2026-10-20 to 2026-11-09 | 18.7 (6.6 to 58.8)n 9 · 2026-10-20 to 2026-11-09 | 4 of 6 (67%)n 6 · 2026-10-20 to 2026-10-31 | 4 of 6 (67%)n 6 · 2026-10-20 to 2026-10-31 | 5 of 5 (100%)n 5 · 2026-10-20 to 2026-10-29 | 5 of 5 (100%)n 5 · 2026-10-20 to 2026-10-29 | 24 hn 9 · 2026-10-20 to 2026-11-09 |
| 1W | n 2, under 5: not shown | n 2, under 5: not shown | n 2, under 5: not shown | n 1, under 5: not shown | n 1, under 5: not shown | n 0, under 5: not shown | n 0, under 5: not shown | n 1, under 5: not shown |
How big the gaps are
Gap size in pips by timeframe, gaps formed 2026-10-12 to 2026-11-10: the median, the middle 80% (10th to 90th percentile) and how many gaps fall in each size band. Each row shows its own n and dates; under 5 gaps it shows no figure. Counts and sizes only. A pip is 0.0001, 0.01 for yen pairs and 0.10 for gold, so sizes of different markets are not like for like.
| Timeframe | Median (pips) | 10th to 90th | under 1 | 1 to under 2 | 2 to under 5 | 5 to under 10 | 10 to under 20 | 20 and over | Sample |
|---|---|---|---|---|---|---|---|---|---|
| 5m | 1.1 | 0.2 to 4.2 | 1385 | 737 | 652 | 167 | 36 | 23 | n 3000 · 2026-10-28 to 2026-11-10 |
| 15m | 1.7 | 0.3 to 6.3 | 461 | 381 | 473 | 181 | 46 | 15 | n 1557 · 2026-10-12 to 2026-11-10 |
| 1h | 3.7 | 0.5 to 12.4 | 64 | 63 | 121 | 85 | 42 | 14 | n 389 · 2026-10-12 to 2026-11-10 |
| 4h | 9.1 | 1 to 25.5 | 10 | 6 | 14 | 23 | 36 | 13 | n 102 · 2026-10-12 to 2026-11-10 |
| 1D | 18.7 | 6.6 to 58.8 | 0 | 0 | 1 | 2 | 2 | 4 | n 9 · 2026-10-20 to 2026-11-09 |
| 1W | – | – | – | – | – | – | – | – | n 2, under 5: not shown |
Is it changing?
Change on last month: not enough history yet. The stored gaps cover 2026-10-12 to 2026-11-10 (29 days); a comparison of one month with the one before needs 60 days, so no figure is shown.
Choosing one without fooling yourself
- Pick the timeframe from how long you can watch the chart and how long you will hold, not from which one shows the most gaps.
- Set the minimum size per timeframe (see minimum size): a pip threshold that suits 5m removes nothing on 1D.
- Count the stop as part of the choice. A wider gap needs a wider stop, and the loss if it is hit is larger for the same position.
- Keep the choice fixed over the period you study before you compare results.
Written by Michael H.