Part 3: What happens to levels · Chapter 8 of 14 · 10 min
The retest after the break
After a true break the price comes back to touch the old level from the new side in 75% of cases within three days (median: 10 candles). The "support that becomes resistance" holds 50% (± 4), like random levels (50%). A resistance broken upwards: the retest holds 56% (± 6); a support broken downwards: 45% (± 6): differences inside the error.
Polarity
There is an idea in technical-analysis courses that links supports and resistances: polarity. When the price breaks a resistance, that resistance becomes a support: if the price comes back to touch it from above, it should bounce. And the same the other way round: a broken support becomes a resistance. The return is called a retest.
The explanation is that whoever sold below the resistance and found themselves at a loss buys to exit at breakeven, whoever bought the break defends the position: there are orders on the new side. It is a sensible explanation and it can be counted.
In the book: a true break (chapter 7: one close beyond the zone, then a close at 1 ATR beyond the zone) opens a window of 72 market-open H1 candles (three trading days). The retest is the first contact with the zone from the new side. The outcome is measured as in chapter 1: hold if the price moves away by 1 ATR in the direction of the break (the old obstacle holds), break if it goes back beyond by 1 ATR. The comparison is with random levels shifted by 2-6 ATR, treated with the same procedure.
How often there is a retest
On time-based levels (previous day, week, Asia) there are 820 true breaks with one close. In 613 (75%) the price comes back to touch the zone from the new side within three days. With random levels the share is the same (74%). The retest is fast: a median of 10 candles from the end of the break, 56% within 12 candles and 79% within 24.
The retest arrives often: it is the behaviour of a price that oscillates, not of a special level. The old level is close to the price, which finds it again without difficulty.
Does the retest hold?
| Levels | True breaks | True breaks, retest | Retest hold | Random levels (retest, hold) |
|---|---|---|---|---|
| Time-based levels, break with one close | 820 | 613 (75%) | 50% (± 4) | 74%, 50% |
| Time-based levels, break at 0.5 ATR beyond the zone | 1,227 | 918 (75%) | 50% (± 3) | 74%, 49% |
| H4 swings, break with one close | 405 | 302 (75%) | 49% (± 6) | 76%, 50% |
| H4 swings, break at 0.5 ATR beyond the zone | 623 | 455 (73%) | 47% (± 5) | 76%, 49% |
| 50 $ round numbers, break with one close | 337 | 260 (77%) | 48% (± 6) | 74%, 51% |
| 50 $ round numbers, break at 0.5 ATR beyond the zone | 543 | 403 (74%) | 45% (± 5) | 74%, 49% |
Six rows, six results close to 50% (45-50%) and all equal to random levels. Polarity does not show: the retest of a broken level holds like contact with any other price.
The round numbers at 0.5 ATR (45%) are the only row a little lower, and consistent with what we saw in chapter 5: round numbers hold a little less, on either side.
Resistance becoming support, and vice versa
Polarity has two directions. On time-based levels:
| Direction | Retests | Hold | Random levels |
|---|---|---|---|
| Resistance broken upwards: retest from above, expected "support" | 308 | 56% (± 6) | 51% |
| Support broken downwards: retest from below, expected "resistance" | 305 | 45% (± 6) | 48% |
The difference between the two directions is 11 points. At first sight it seems that "support" holds better than "resistance". With random levels the difference is 3 points (51% against 48%), and the error on the comparison is about 8 points: the larger part is explained by chance, not by the level. A hold of 56% over 308 cases is inside 50% ± 6.
When the retest arrives
A retest within 12 candles (346 cases) holds 53% (± 5), between 12 and 36 candles (198) 48% (± 7), between 36 and 72 (69) 43% (± 12). The more time passes, the less it holds: a pattern entirely inside the error and still to retest. The common thread: no band stands clearly away from 50%.
The retest is not a light touch: within the 12 candles from the contact, the price goes beyond the level (in the case of a resistance broken upwards, below it) by more than a quarter of an ATR in 82% of retests, by 1.4 ATR as a median. Whoever puts the stop just beyond the zone is stopped almost always (chapter 12).
Read the chart in words
H1 gold chart of 21-23 April 2026 with a level's zone, the break and the retest marked.
| When | What | Price |
|---|---|---|
| 21 Apr 2026 15:00 UTC | level 4731.76 | 4,724.85 – 4,738.67 |
| 21 Apr 2026 23:00 UTC | break | 4,739.91 |
| 22 Apr 2026 14:00 UTC | retest: hold | 4,732.42 |
Times in UTC.
Read the chart in words
H1 gold chart from 28 May to 1 June 2026 with a level's zone, the break and the retest marked.
| When | What | Price |
|---|---|---|
| 28 May 2026 08:00 UTC | level 4492.09 | 4,486.31 – 4,497.87 |
| 28 May 2026 16:00 UTC | break | 4,512.40 |
| 1 Jun 2026 07:00 UTC | retest: break | 4,495.51 |
Times in UTC.
Read the chart in words
H1 gold chart of 14-15 July 2026 with a level's zone and four candles marked with the letters A, B, C and D.
| When | What | Price |
|---|---|---|
| 14 Jul 2026 03:00 UTC | level 4024.54 | 4,021.70 – 4,027.38 |
| 14 Jul 2026 07:00 UTC | Point A | 4,032.03 |
| 14 Jul 2026 11:00 UTC | Point B | 4,033.96 |
| 15 Jul 2026 05:00 UTC | Point C | 4,039.37 |
| 15 Jul 2026 15:00 UTC | Point D | 4,064.90 |
Times in UTC.
What to do with it
- Expect the retest, not the bounce. After a true break, the price comes back to the level in 75% of cases within three days, with a median of 10 candles. It is a chance for a limit order, not a signal of direction.
- Do not count on polarity. The old support holds like an ordinary price (50%).
- Put the stop beyond the typical depth of the retest. As a median the price passes the level by 1.4 ATR in the 12 candles after the contact; a stop a quarter of an ATR from the level is hit in 82% of cases.
- Do not tell the stories of retests that held. With a 50% hold, half the retests hold and make a nice chart; the other half do not.
- Compare with random levels for retests too. The 75% of retests and the 50% hold are found with levels shifted at random.
Procedure in five steps
- Define the true break (one close beyond the zone and then 1 ATR beyond).
- Open a window of 72 candles and mark the first contact from the new side.
- Measure the retest hold (1 ATR in the direction of the break against 1 ATR back).
- Redo it on random levels and compare.
- Measure the depth of the retest to choose the stop distance.
Exercises
Chapter exercises
1. On the chart the grey box is the zone of a level (4,024.54 $) broken upwards on 14 July, and the letters A-D are on four H1 candles. Which letter is the retest candle, the first contact with the zone from the new side (from above)?
Read the chart in words
| When | What | Price |
|---|---|---|
| 14 Jul 2026 03:00 UTC | level 4024.54 | 4,021.70 – 4,027.38 |
| 14 Jul 2026 07:00 UTC | Point A | 4,032.03 |
| 14 Jul 2026 11:00 UTC | Point B | 4,033.96 |
| 15 Jul 2026 05:00 UTC | Point C | 4,039.37 |
| 15 Jul 2026 15:00 UTC | Point D | 4,064.90 |
Times in UTC.
2. After a true break the price comes back to touch the zone in 75% of cases within three days; random levels give 74%. What does it mean?
3. The retest of a broken level holds 50% (± 4) on time-based levels, and random levels give 50%. How do you read it?
4. In the 12 candles from the retest, the price passes beyond the level by more than a quarter of an ATR in 82% of cases. What does it imply for the stop?
Have a question or want to share your exercise?
Post in the community, or join the free signals room on Telegram.
Educational content, not financial advice. Trading involves risk.