Part 2: Where levels go · Chapter 6 of 14 · 9 min
How many touches
A level touched several times holds neither more nor less: on time-based levels the hold is 47%, 50%, 48%, 50% and 55% (± 11) from the first touch to the sixth and beyond; after a hold the next touch holds 49%, after a break 51%. Neither the time elapsed nor the type of level changes the result: the price does not "remember" touches.
Two opposite beliefs
In technical-analysis courses two ideas about touches live side by side that cannot both be true.
- More touches, stronger. A level the price has already respected twice is "confirmed": there are more orders, more traders are watching it, next time it will hold.
- Every touch wears it down. The orders sitting on the level get executed at every touch: after a few contacts none are left, and next time the level gives way.
To decide you only have to count. For each level we numbered the contacts in order (first touch, second...) and measured the hold by number, with the chapter 1 rules.
The hold by touch number
| Touch | Time-based levels (day, week, month, Asia) | H4 swings (1.5 ATR) | H1 swings (1.5 ATR) | 50 $ round numbers |
|---|---|---|---|---|
| 1st | 47% (± 3, 879) | 48% (± 7, 184) | 48% (± 3, 965) | 38% (± 16, 34) |
| 2nd | 50% (± 5, 462) | 47% (± 8, 154) | 50% (± 3, 841) | 50% (± 17, 32) |
| 3rd | 48% (± 7, 175) | 53% (± 9, 125) | 47% (± 4, 718) | 47% (± 18, 30) |
| 4th-5th | 50% (± 9, 117) | 50% (± 7, 176) | 50% (± 3, 1,096) | 44% (± 13, 59) |
| 6th and beyond | 55% (± 11, 82) | 43% (± 6, 231) | 47% (± 3, 1,450) | 44% (± 4, 597) |
In each cell: hold (± error, number of contacts). Four families of levels, five touch numbers, twenty values between 38% and 55%, with errors of 3-18 points. No pattern: the hold does not rise with touches (one belief says "stronger", the other "weaker") and does not fall.
The two cells where the value strays furthest are those with the fewest contacts: the first touch of round numbers (38% over 34 contacts, error 16) and the sixth and beyond of time-based levels (55% over 82, error 11).
Does the previous touch matter?
Another question: if the last touch held, does the next hold more? For each level with at least two touches we compared.
| Family | After a hold, the next touch holds | After a break (and return), the next touch holds |
|---|---|---|
| Time-based levels | 49% (± 5, 355 cases) | 51% (± 4, 481) |
| H4 swings | 46% (± 6, 308) | 49% (± 5, 378) |
| H1 swings | 47% (± 2, 1,900) | 49% (± 2, 2,205) |
| 50 $ round numbers | 44% (± 5, 317) | 46% (± 5, 401) |
Nothing: the two numbers are equal within the error. The outcome of one touch says nothing about the next. The price keeps no tally.
And the time since the last touch? On time-based levels, when the next touch arrives after fewer than 12 candles the hold is 51% (± 4, 697 contacts), between 12 and 47 candles 49% (± 9, 119), after 48 or more 45% (± 22, only 20). On H1 swings, 47% (2,758), 48% (927) and 57% (± 5, 420) for contacts after two days or more. The last is the value furthest from 50%; among the many values computed in the chapter, one outside the band is normal. We record it as a question for a second period, not as a result.
Read the chart in words
H1 gold chart from 29 December 2025 to 2 January 2026 with the zone of a swing level and five contacts numbered and marked as hold or break.
| When | What | Price |
|---|---|---|
| 29 Dec 2025 07:00 UTC | level 4358.90 | 4,352.62 – 4,365.18 |
| 29 Dec 2025 15:00 UTC | 1: hold | 4,327.70 |
| 30 Dec 2025 14:00 UTC | 2: hold | 4,363.01 |
| 30 Dec 2025 19:00 UTC | 3: break | 4,346.27 |
| 31 Dec 2025 01:00 UTC | 4: hold | 4,373.49 |
| 2 Jan 2026 00:00 UTC | 5: break | 4,357.88 |
Times in UTC.
Read the chart in words
H1 gold chart from 29 June to 1 July 2026 with the zone of the round number 4,000 and four contacts numbered and marked as hold or break.
| When | What | Price |
|---|---|---|
| 29 Jun 2026 06:00 UTC | level 4000.00 | 3,995.65 – 4,004.35 |
| 29 Jun 2026 14:00 UTC | 1: hold | 4,001.68 |
| 30 Jun 2026 00:00 UTC | 2: break | 3,982.39 |
| 1 Jul 2026 00:00 UTC | 3: break | 3,996.20 |
| 1 Jul 2026 10:00 UTC | 4: break | 4,000.80 |
Times in UTC.
The life of a level
A different way of looking at the same thing: how many holds before the first break? Of the 879 time-based levels touched at least once, 301 (34%) never break within their validity period and 578 break at least once. Of those 578, 467 (81%) break already at the first touch, 96 after one hold, 12 after two and 3 after three or more. On H4 swings: 184 levels touched, 156 with at least one break, of which 96 at the first touch, 41 after one hold, 11 after two and 8 after three or more.
The 81% seems to say the first touch is the one that decides. It is a selection effect, not the level's: levels touched only once and held sit among the 301 "never broken" and do not enter among the broken; whoever breaks necessarily had a break, and more often it is the first. The hold at the first touch is 47%, like the second (50%). Every touch is a new coin flip, with the same weight.
Read the chart in words
H1 gold chart from 30 December 2025 to 3 January 2026 with the zone of a level and four contacts marked with the letters A, B, C and D.
| When | What | Price |
|---|---|---|
| 30 Dec 2025 22:00 UTC | level 4333.00 | 4,327.65 – 4,338.35 |
| 31 Dec 2025 04:00 UTC | Point A | 4,358.08 |
| 1 Jan 2026 17:00 UTC | Point B | 4,332.69 |
| 1 Jan 2026 23:00 UTC | Point C | 4,356.20 |
| 2 Jan 2026 15:00 UTC | Point D | 4,357.53 |
Times in UTC.
What to do with it
- Do not treat the number of touches as a strength measure. "Tested three times" does not say the level is stronger (hold 48% at the third touch) or weaker.
- Do not trust the sequence. A hold does not say the next will be a hold: after a hold the next touch holds 49%, after a break 51%.
- Use touches as memory, not as prediction. Counting touches tells you where the price has already been and lets you delete old levels.
- Do not pick the most "proven" level of a group. Every level is a new coin flip.
- Keep a touch log for your levels. If after 200 contacts your number 1 is very different from number 2, you have found something; with 30 you have not.
Procedure in four steps
- For each level log the contacts in order, with the outcome (hold or break).
- Compute the hold by touch number with its error (± 1.96 times the square root of p(1 − p)/n).
- Compare the hold after a hold with that after a break.
- If there is no difference within the error, ignore the number of touches.
Exercises
Chapter exercises
1. On the chart the grey box is an H4 swing level (4,333.00 $) and the letters A-D are its four contacts, in the order they occurred. Which letter is the third contact?
Read the chart in words
| When | What | Price |
|---|---|---|
| 30 Dec 2025 22:00 UTC | level 4333.00 | 4,327.65 – 4,338.35 |
| 31 Dec 2025 04:00 UTC | Point A | 4,358.08 |
| 1 Jan 2026 17:00 UTC | Point B | 4,332.69 |
| 1 Jan 2026 23:00 UTC | Point C | 4,356.20 |
| 2 Jan 2026 15:00 UTC | Point D | 4,357.53 |
Times in UTC.
2. After a hold the next touch holds 49% and after a break 51% (time-based levels, 355 and 481 cases). What does it mean?
3. The hold by touch number is 47%, 50%, 48%, 50% and 55% from the first to the sixth. What does it say about the two beliefs ("more touches, stronger" and "every touch wears it down")?
4. Of the 578 time-based levels that break, 467 (81%) do so at the first touch, but the first touch holds 47%. How is it explained?
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.