Part 4: Breakouts, system and test · Chapter 12 of 14 · 10 min

All the patterns together

The 14 pattern types together give +0.03 ± 0.19 R over 330 trades (random series +0.01) and no type is positive in a stable way in the two halves of the period. Also 207 patterns out of 268 (77%) overlap another one, as much as in random series (76%): the same chart can be read in several ways.

The single table

The rules are those of the previous chapters: patterns read on swings of at least 1.5 ATR, entry at the close of the breakout, stop beyond the opposite extreme plus 0.25 ATR, target at one height, exit after 48 candles, cost 0.30 $. Data: H1 gold, December 2025 - October 2026. On the right the comparison with 100 random series obtained by shuffling blocks of 6 candles (median; in brackets 5th and 95th percentile).

TypeFound (random, 5th-95th)TradesR ± errorR of the random series
Head and shoulders11 (5, 3-9)9−0.06 ± 0.600.00 (−0.85 / +0.78)
Double58 (45, 35-53)54+0.01 ± 0.220.00 (−0.17 / +0.24)
Triple3 (2, 0-5)3−0.88 ± 0.25−0.05 (−1.01 / +0.82)
Ascending triangle11 (11, 6-17)11−0.28 ± 0.68+0.02 (−0.66 / +0.79)
Descending triangle11 (11, 7-17)10+0.26 ± 0.76−0.09 (−0.64 / +0.98)
Symmetrical triangle33 (29, 22-37)32+0.54 ± 1.70−0.02 (−0.51 / +0.93)
Rectangle7 (4, 1-8)7+0.38 ± 0.53−0.02 (−1.00 / +0.74)
Channel25 (22, 15-29)21+0.08 ± 0.36+0.01 (−0.28 / +0.30)
Rising wedge31 (25, 19-31)31−0.14 ± 0.39−0.02 (−0.44 / +0.67)
Falling wedge17 (24, 16-32)16−0.39 ± 0.44+0.07 (−0.45 / +0.58)
Flag42 (49, 42-60)41+0.48 ± 0.45+0.02 (−0.31 / +0.37)
Pennant8 (16, 10-21)5−1.02 ± 0.01−0.06 (−0.65 / +0.77)
Broadening11 (11, 5-15)2+0.41 ± 0.15+0.13 (−0.36 / +0.34)
V spikes90 (123, 110-136)88−0.14 ± 0.17−0.01 (−0.14 / +0.10)

Errors under 20 trades are not reliable: with 2, 3 or 5 trades the formula gives narrow intervals that describe nothing. The triple, the pennant and the broadening must be read as anecdotes.

How many types leave the range of chance

By number of patterns found: four types out of 14 sit outside the 5th-95th interval of the random series. Two have more (head and shoulders 11 against 5; double 58 against 45) and two fewer (pennant 8 against 16; V spikes 90 against 123). The total, 358 against a median of 381 (from 351 to 405), sits inside. With 14 types and a threshold of 10% (5% per side) 1.4 are expected outside by pure chance.

By mean R: two types above the 95th percentile of chance (flag +0.48 R against +0.37, and broadening with 2 trades) and two at the limit or below the 5th (pennant with 5 trades and V spikes, −0.14 against −0.14). Again, the two with fewer than 10 trades do not count. The flag remains, as in chapter 7.

By stability: splitting the period in half, the flag gives +1.09 R in the first half (19 trades) and −0.04 in the second (22). No type with more than one trade per half has a positive R in both. Four types have a negative R in both: triple, falling wedge (8 and 8 trades), pennant and V spikes (39 and 49).

All together

The 14 types give 330 trades: 115 targets, 147 stops, 68 time exits, mean R +0.03 ± 0.19. The random series give +0.01 (from −0.10 to +0.11). 28 patterns out of 358 are skipped for too large a risk or for the end of the data. Without the V spikes the trades are 242 and the mean R +0.10. The sum of all types sits inside the range of chance.

Other timeframes

Same rules on H4 (swings of 2 ATR, exit after 48 H4 candles): 91 patterns, 81 trades, R +0.27 ± 0.24 (42 targets, 27 stops). The random series on H4 give +0.10 (from −0.13 to +0.38): gold sits inside. First half +0.30 (37 trades), second +0.25 (44). On D1 (swings of 1 ATR) 16 trades, +0.27 ± 0.56. The interval on H4 sits just above zero, but H4 is the second timeframe tried after H1 and its candles are the same: it is not an independent confirmation.

The same chart, several patterns

The detector of the book reads the same swings in several ways. Among the 268 patterns (without V spikes), 207 (77%) overlap another one for more than half of their duration, and in 81 of the 205 overlapping pairs (40%) the two breakouts go in opposite directions. In the random series the share is 76% (from 69% to 81%): it is a property of the reading, not of the market.

Figure 1 shows the case of 14-16 June 2026. The swings 4,336.15, 4,303.23 and 4,371.13 $ are shared by two patterns: a rising wedge (A1-A5, from 14 June at 15:00, downward breakout on 15 June at 16:00 below 4,368.80 $, target reached at 20:00: +0.60 R) and a rectangle (B1-B5, upward breakout on 16 June at 07:00 above 4,335.27 $, target on 17 June at 17:00: +1.13 R). The same swings also contain a symmetrical triangle (upward breakout at 08:00, stop hit on 17 June at 18:00: −1.01 R), a flag (upward breakout at 07:00, stop hit at 18:00: −1.01 R) and a channel (downward breakout on 16 June at 01:00, skipped for too large a risk). Five patterns, four trades: two win and two lose. Shared swings carry two labels, for example A3/B1.

Figure 2 shows the case of 29 July - 3 August: a falling wedge (A1-A5, from 4,120.44 to 4,062.49 $, upward breakout on 1 August at 02:00 above 4,054.04 $, +0.42 R with a time exit on 3 August at 02:00) and a symmetrical triangle (B1-B5, from 4,126.64 to 4,076.84 $, upward breakout on 2 August at 15:00 above 4,068.00 $, stop hit on 3 August at 01:00: −1.04 R). Same swings 4,126.64, 4,029.47 and 4,062.49 $, two patterns, two different outcomes.

Figure 1. The same stretch of chart (14-16 June 2026) read as a rising wedge (A, downward breakout) and as a rectangle (B, upward breakout): two patterns, two opposite directions.Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 14 Jun 2026 09:00 UTC to 17 Jun 2026 13:00 UTC.
Read the chart in words

H1 gold chart from 14 to 17 June 2026 with the lines of the rising wedge (A1-A5) and of the rectangle (B1-B5) and the two breakouts.

WhenWhatPrice
15 Jun 2026 16:00 UTCbreakout down4,336.15
16 Jun 2026 07:00 UTCbreakout up4,337.06
14 Jun 2026 15:00 UTCPoint A14,233.16
14 Jun 2026 18:00 UTCPoint A24,224.55
15 Jun 2026 02:00 UTCPoint A3/B14,336.15
15 Jun 2026 06:00 UTCPoint A4/B24,303.23
15 Jun 2026 14:00 UTCPoint A5/B34,371.13
15 Jun 2026 22:00 UTCPoint B44,307.46
16 Jun 2026 03:00 UTCPoint B54,335.39

Times in UTC.

Figure 2. A falling wedge (A) and a symmetrical triangle (B) on the same swings, 29 July - 3 August 2026: the first closes at +0.42 R with a time exit, the second takes the stop (−1.04 R).Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 29 Jul 2026 13:00 UTC to 3 Aug 2026 05:00 UTC.
Read the chart in words

H1 gold chart from 29 July to 3 August 2026 with the lines of the falling wedge (A1-A5) and of the symmetrical triangle (B1-B5) and the two breakouts.

WhenWhatPrice
1 Aug 2026 02:00 UTCbreakout up4,056.20
2 Aug 2026 15:00 UTCbreakout up4,068.68
29 Jul 2026 19:00 UTCPoint A14,120.44
30 Jul 2026 05:00 UTCPoint A24,039.00
30 Jul 2026 14:00 UTCPoint A3/B14,126.64
31 Jul 2026 14:00 UTCPoint A4/B24,029.47
31 Jul 2026 19:00 UTCPoint A5/B34,062.49
1 Aug 2026 18:00 UTCPoint B44,048.01
2 Aug 2026 04:00 UTCPoint B54,076.84

Times in UTC.

Procedure in four steps

  1. Choose beforehand which patterns you use and how you recognise them, with written rules and tolerances.
  2. Count all the patterns the rules find, including overlapping ones and those with opposite direction.
  3. Compare how many you find and how they do with a random series.
  4. Check the two halves of the period and how many types you tried before trusting the best one.

Exercises

Chapter exercises

1. On the chart the letters A-D are: the last swing of the wedge, the breakout of the wedge (downward), the breakout of the rectangle (upward) and the candle where the rectangle reaches its target. Which letter is the close of the breakout of the rectangle?

Chart exercise. Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 14 Jun 2026 09:00 UTC to 18 Jun 2026 01:00 UTC.
Read the chart in words
WhenWhatPrice
15 Jun 2026 14:00 UTCPoint A4,371.13
15 Jun 2026 16:00 UTCPoint B4,336.15
16 Jun 2026 07:00 UTCPoint C4,337.06
17 Jun 2026 17:00 UTCPoint D4,384.94

Times in UTC.

2. Across 14 pattern types a threshold of 10% (5% per side) is used to say that a number sits outside the range of chance. How many types do you expect outside by chance alone?

3. 77% of patterns overlap another one, and in random series 76%. What does it mean?

4. The flag gives +0.48 ± 0.45 R over 41 trades, but +1.09 in the first half (19 trades) and −0.04 in the second (22). How do you judge 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.