Part 3: What happens to levels · Chapter 10 of 14 · 10 min

Confluence

85% of contacts have another level of a different type within half an ATR, and random levels have 86%: confluence is the norm. With more overlapping levels the hold does not rise (50%, 48%, 47%, 46% with 0, 1, 2, 3 or more other levels, like random ones). A level near a round number holds 44%, against 51% for the others. The best of 19 pairs (Asia plus trendline: 64%) is within the range of chance.

The idea of confluence

Courses talk a lot about confluence: yesterday's high that coincides with a round number and a four-hour swing, a week level that falls on the trendline. The idea is that if several levels sit at the same point, there are more orders and more traders watching there, and the reaction should be stronger. More reasons, more strength.

In the book confluence is measured like this. For each contact with a level we count how many other levels, of a different family, lie within half an ATR (9.3 $ as a median) of its price at that moment. The families are seven: previous day, previous week, previous month, Asia, H4 swing (1.5 ATR), 50 $ round numbers, H4 trendline. A level counts once per family. Then we look at the hold as a function of the number of families involved.

Confluence is the norm

The first result is a frequency: over 3,702 contacts with 1,847 levels, 85% have at least one other level of a different family within half an ATR. With shifted random levels the share is identical (86%). At an ordinary price (the close of an H1 candle) there are on average 1.79 families within half an ATR, none in 15% of cases, two or more in 55%.

A confluence of two levels, then, is not a rare event: it happens at more than one price in two. That is why, as in chapter 2, every contact has a confluence available, and whoever looks for it afterwards finds it.

The hold as a function of overlapping levels

Other levels (different families) within 0.5 ATRContactsHoldBreakRandom levels (same counts): hold
None53850% (± 4)48%49%
One1,00648% (± 3)51%51%
Two1,05247% (± 3)53%46%
Three or more1,10646% (± 3)52%46%

The hold does not rise with the number of levels: it goes from 50% to 46%, always inside the band of random levels with the same counts. If anything it falls a little, and falls in the same way with random levels. The idea that more nearby levels give more strength does not show.

The pairs

Confluence can be specific: not any other level, but a precise type. We tried all the pairs of families with at least 60 contacts (19 out of 21 possible): each contact counts in each of its pairs. The eight furthest from 50%:

PairContactsHold
Asia + H4 trendline13564% (± 8)
Round number + H4 trendline27739% (± 6)
Asia + round number24742% (± 6)
Previous day + round number62144% (± 4)
Previous week + round number35744% (± 5)
H4 swing + H4 swing36744% (± 5)
H4 swing + round number73445% (± 4)
Previous day + H4 swing76745% (± 4)

At the top is the pair Asia + H4 trendline with 64%: 14 points above 50%. It looks like the chapter's discovery. But we tried 19 pairs, and the right question is how large the maximum gap among 19 pairs is when levels are random. Over 30 trials with levels shifted at random, the maximum gap from 50% has a median of 9.5 points and a maximum of 17.6: in 7% of trials there is a pair at least as far from 50% as ours (14.4 points). A value like that turns up one time in fourteen even without real levels.

Five of the eight pairs in the list contain a round number and sit between 39% and 45%: it is the result of chapter 5 (round numbers hold less).

A simple comparison confirms it: time-based and swing levels with a 50 $ round number within half an ATR hold 44% (± 3, 1,274 contacts), those without a round number nearby 51% (± 2, 1,676). Confluence with a round number does not help: the hold falls by 7 points.

Figure 1. A confluence that holds (16-17 March 2026, UTC hours): yesterday's close (5,007.24 $) coincides with an H4 swing (5,004.60 $) and with the round number 5,000: the price arrives from above at midnight and the level holds.Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 16 Mar 2026 12:00 UTC to 17 Mar 2026 18:00 UTC.
Read the chart in words

H1 gold chart of 16-17 March 2026 with the zone of yesterday's close, the line of an H4 swing and that of the round number 5,000.

WhenWhatPrice
16 Mar 2026 12:00 UTCyesterday's close 5007.245,001.40 – 5,013.08
16 Mar 2026 12:00 UTCH4 swing 5004.605,004.60
16 Mar 2026 12:00 UTCround number 50005,000.00
17 Mar 2026 00:00 UTCtouch4,998.68
17 Mar 2026 04:00 UTChold5,036.92

Times in UTC.

Figure 2. A confluence that gives way (4-5 March 2026, UTC hours): yesterday's close (5,141.12 $), an H4 swing (5,133.05 $) and the round number 5,150 sit at the same point, but the price passes through them.Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 4 Mar 2026 17:00 UTC to 5 Mar 2026 23:00 UTC.
Read the chart in words

H1 gold chart of 4-5 March 2026 with the zone of yesterday's close, the line of an H4 swing and that of the round number 5,150.

WhenWhatPrice
4 Mar 2026 17:00 UTCyesterday's close 5141.125,134.95 – 5,147.29
4 Mar 2026 17:00 UTCH4 swing 5133.055,133.05
4 Mar 2026 17:00 UTCround number 51505,150.00
5 Mar 2026 05:00 UTCtouch5,145.86
5 Mar 2026 13:00 UTCbreak5,116.10

Times in UTC.

Why it seems to work

When you look at a chart, a confluence that held is noticed ("three levels at the same point, a nice bounce"). One that gave way is forgotten or explained away ("they broke it of course, the stops were there"). And confluence is always available: at any point there are on average 1.79 families, and two or more in 55% of cases.

Figure 3. Exercise: four H1 candles (4-5 June 2026, UTC hours) around a point where yesterday's open (4,446.40 $), an H4 swing (4,451.52 $) and the round number 4,450 coincide, marked with the letters A-D.Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 4 Jun 2026 13:00 UTC to 5 Jun 2026 19:00 UTC.
Read the chart in words

H1 gold chart of 4-5 June 2026 with the zone of yesterday's open, the line of an H4 swing and that of the round number 4,450.

WhenWhatPrice
4 Jun 2026 13:00 UTCyesterday's open 4446.404,442.17 – 4,450.63
4 Jun 2026 13:00 UTCH4 swing 4451.524,451.52
4 Jun 2026 13:00 UTCround number 44504,450.00
4 Jun 2026 20:00 UTCPoint A4,484.39
5 Jun 2026 01:00 UTCPoint B4,471.83
5 Jun 2026 10:00 UTCPoint C4,471.00
5 Jun 2026 16:00 UTCPoint D4,349.10

Times in UTC.

What to do with it

  1. Do not count overlapping levels as proof of strength. With 0, 1, 2 or 3+ other levels the hold is 50%, 48%, 47%, 46%, like random ones.
  2. Distrust the confluence you found afterwards. In 85% of contacts there is already another level nearby.
  3. Do not prefer levels near a round number. They hold 44%, against 51% for the others.
  4. Count the pairs you try. With 19 pairs, a 64% is within the range of chance (7% of random trials have one as large).
  5. Use confluence to know how crowded a point is. A piece of context: where there are more levels, the price finds more obstacles to cross, it does not bounce more.

Procedure in four steps

  1. Define the maximum distance for confluence (half an ATR) and the families of levels.
  2. For each contact count the families within that distance and measure the hold by number.
  3. If you try pairs, count how many: the best of N must be compared with the best of N on random levels.
  4. Repeat on levels shifted at random and compare.

Exercises

Chapter exercises

1. On the chart the grey box is yesterday's open (4,446.40 $), and the lines are an H4 swing (4,451.52 $) and the round number 4,450. The letters A-D are on four candles of 4-5 June 2026 (UTC hours). Which letter is the contact with the zone where the three levels coincide, coming from above?

Chart exercise. Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 4 Jun 2026 13:00 UTC to 5 Jun 2026 19:00 UTC.
Read the chart in words
WhenWhatPrice
4 Jun 2026 13:00 UTCyesterday's open 4446.404,442.17 – 4,450.63
4 Jun 2026 13:00 UTCH4 swing 4451.524,451.52
4 Jun 2026 13:00 UTCround number 44504,450.00
4 Jun 2026 20:00 UTCPoint A4,484.39
5 Jun 2026 01:00 UTCPoint B4,471.83
5 Jun 2026 10:00 UTCPoint C4,471.00
5 Jun 2026 16:00 UTCPoint D4,349.10

Times in UTC.

2. 85% of contacts have another level of a different family within half an ATR, and random levels have 86%. What does it mean?

3. The hold with 0, 1, 2 and 3 or more other nearby levels is 50%, 48%, 47% and 46%; random levels give 49%, 51%, 46% and 46%. How do you read it?

4. Among 19 pairs of families, the best holds 64% (± 8) over 135 contacts. In 30 trials with random levels, 7% have a pair at least this far from 50%. How do you treat it?

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.