Part 3: Three-candle formations · Chapter 8 of 14 · 8 min

Morning and evening stars

Morning and evening stars (three candles: a large one, a small one, one that comes back) appear 59 times on H1 gold, just above chance (50). Price goes in the expected direction in 46% of cases (control 37%) but the 55 trades give −0.05 ± 0.26 R; the three outside are 106 and give −0.10 ± 0.18 R.

The rules

  • Morning star: the first candle is bearish with a body of at least 1 ATR; the second has a body of at most 30% of that of the first and stays in its zone; the third is bullish and closes beyond the middle of the first. Expected direction: up. Without gaps, the middle candle opens at the close of the first: the "abandoned baby", the star with gaps, does not exist in these data.
  • Evening star: mirror image. Expected direction: down.
  • Three outside: an engulfing followed by a third candle in the same direction that closes beyond its close. The three inside is in chapter 6.

Stop beyond the extreme of the three candles plus 0.25 ATR (risk 0.5-4 ATR), target equal to the risk, 24 candles, cost 0.30 $.

How often they appear

FormationFound (chance, 5th-95th)In expected directionTradesMean RRandom (median)
Morning star31 (25, 19-31)45%27+0.12 ± 0.35−0.16
Evening star28 (25, 20-31)46%28−0.22 ± 0.36−0.09
The two stars59 (50, 42-59)46%55−0.05 ± 0.26−0.13
Stars with an almost-doji middle1953%16+0.17 ± 0.48−0.13
Three outside106 (139, 129-155)46%98−0.10 ± 0.18−0.02

The control is 37% up, 36% down, 27% no threshold. The stars sit at the upper edge of the range of chance (59 against 42-59). Price goes in the expected direction in 46% of cases (±13 points), above the control and above the 33% of the random series. The 55 trades give −0.05 ± 0.26 R (25 targets, 25 stops, 5 time exits): better than almost all the random series (−0.13, 26% do at least as well), but with an error of ±0.26 R. The morning star has the highest value (27 trades, +0.12 ± 0.35 R, and only 5% of the random series do as well); the evening star gives −0.22 ± 0.36 R. First half −0.16 R, second +0.13 R. With the middle candle almost a doji (19 stars, 16 trades) +0.17 ± 0.48 R. With the confirmation of the next candle: 20 confirmations and 15 trades at −0.11 ± 0.52 R. The samples are too small to say more: the star is a rare formation (about 6 a month) and an error of ±0.26-0.48 R does not tell it from chance.

The three outside are 106, fewer than chance (139). 51% of the 209 engulfings of chapter 5 are followed by a third candle in the same direction. Price goes in the expected direction in 46% (±10 points); the 98 trades give −0.10 ± 0.18 R (36 targets, 45 stops, 17 time exits), worse than 80% of the random series (−0.02). The confirmation is already inside the formation, and entering after three candles costs: a farther stop, less room. Adding the confirmation of the next candle: 19 trades at −0.34 ± 0.39 R.

Four examples

Figure 1 shows a winning morning star: on 12 April 2026 at 22:00 (4,711.89 → 4,661.39 $, low 4,647.18 $), at 23:00 (4,661.39 → 4,676.04 $) and on 13 April at 00:00 (4,676.04 → 4,696.46 $), after a fall of 5.0 ATR. Entry at 4,696.46 $, stop at 4,643.62 $ (52.84 $, 3.7 ATR), target at 4,749.30 $ reached on 13 April at 22:00: +0.99 R. Figure 2 shows a morning star that fails: on 1 February at 20:00-22:00 (4,852.05 → 4,826.69 $; 4,826.69 → 4,824.15 $; 4,824.14 → 4,849.30 $), after a fall of 2.5 ATR; entry at 4,849.30 $, stop at 4,808.77 $ (40.53 $), target at 4,889.83 $; stop hit at 23:00, one hour later: −1.01 R.

Figure 3 shows an evening star: on 21 January 2026 at 04:00 (4,843.78 → 4,870.51 $), 05:00 (4,870.22 → 4,868.69 $) and 06:00 (4,868.69 → 4,851.78 $), after a rise of 3.9 ATR; entry at 4,851.78 $ downward, stop at 4,895.43 $ (43.65 $, 1.9 ATR), target at 4,808.13 $ reached at 17:00: +0.99 R. Figure 4 shows a bullish three outside: on 30 March at 00:00-02:00 (4,448.82 → 4,433.95 $; 4,433.95 → 4,471.00 $; 4,470.98 → 4,484.42 $), after a fall of 4.8 ATR; entry at 4,484.42 $, stop at 4,414.23 $ (70.19 $, 3.5 ATR), target at 4,554.61 $ reached at 11:00: +1.00 R.

Figure 1. A winning morning star: 12-13 April 2026, after a fall of 5.0 ATR; entry at 4,696.46, stop at 4,643.62, target at 4,749.30 reached on 13 April at 22:00 (+0.99 R).Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 12 Apr 2026 14:00 UTC to 14 Apr 2026 04:00 UTC.
Read the chart in words

H1 gold chart of 12-14 April 2026 with the three numbered candles and the lines of entry, stop and target.

WhenWhatPrice
12 Apr 2026 22:00 UTCPoint 14,647.18
12 Apr 2026 23:00 UTCPoint 24,648.81
13 Apr 2026 00:00 UTCPoint 34,668.63
13 Apr 2026 00:00 UTCEntry4,696.46
13 Apr 2026 00:00 UTCStop4,643.62
13 Apr 2026 00:00 UTCTarget4,749.30
13 Apr 2026 22:00 UTCtarget reached4,759.40

Times in UTC.

Figure 2. A morning star that fails: 1 February 2026, 20:00-22:00, after a fall of 2.5 ATR; entry at 4,849.30, stop at 4,808.77 hit at 23:00 (−1.01 R).Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 1 Feb 2026 12:00 UTC to 2 Feb 2026 05:00 UTC.
Read the chart in words

H1 gold chart of 1-2 February 2026 with the three numbered candles and the lines of entry, stop and target.

WhenWhatPrice
1 Feb 2026 20:00 UTCPoint 14,820.00
1 Feb 2026 21:00 UTCPoint 24,813.34
1 Feb 2026 22:00 UTCPoint 34,824.01
1 Feb 2026 22:00 UTCEntry4,849.30
1 Feb 2026 22:00 UTCStop4,808.77
1 Feb 2026 22:00 UTCTarget4,889.83
1 Feb 2026 23:00 UTCstop hit4,703.34

Times in UTC.

Figure 3. An evening star after a rise of 3.9 ATR (21 January 2026, 04:00-06:00): entry at 4,851.78, stop at 4,895.43, target at 4,808.13 reached at 17:00 (+0.99 R).Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 20 Jan 2026 20:00 UTC to 21 Jan 2026 23:00 UTC.
Read the chart in words

H1 gold chart of 20-21 January 2026 with the three numbered candles and the lines of entry, stop and target.

WhenWhatPrice
21 Jan 2026 04:00 UTCPoint 14,872.37
21 Jan 2026 05:00 UTCPoint 24,879.61
21 Jan 2026 06:00 UTCPoint 34,889.61
21 Jan 2026 06:00 UTCEntry4,851.78
21 Jan 2026 06:00 UTCStop4,895.43
21 Jan 2026 06:00 UTCTarget4,808.13
21 Jan 2026 17:00 UTCtarget reached4,802.05

Times in UTC.

Figure 4. A bullish three outside after a fall of 4.8 ATR (30 March 2026, 00:00-02:00): entry at 4,484.42, stop at 4,414.23, target at 4,554.61 reached at 11:00 (+1.00 R).Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 29 Mar 2026 16:00 UTC to 30 Mar 2026 17:00 UTC.
Read the chart in words

H1 gold chart of 29-30 March 2026 with the three numbered candles and the lines of entry, stop and target.

WhenWhatPrice
30 Mar 2026 00:00 UTCPoint 14,419.21
30 Mar 2026 01:00 UTCPoint 24,433.66
30 Mar 2026 02:00 UTCPoint 34,461.37
30 Mar 2026 02:00 UTCEntry4,484.42
30 Mar 2026 02:00 UTCStop4,414.23
30 Mar 2026 02:00 UTCTarget4,554.61
30 Mar 2026 11:00 UTCtarget reached4,582.63

Times in UTC.

Procedure in four steps

  1. Write the rules of the three candles (bodies, zone, close) before looking for them.
  2. Count how often they appear: a rare formation has few cases.
  3. Report the two mirror images together and tell the sample from the anecdote.
  4. Consider how much it costs to enter after the third candle.

Exercises

Chapter exercises

1. On the chart the letters A-D are on four candles of 12-13 April 2026. Which letter is the small middle candle of the morning star?

Chart exercise. Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 12 Apr 2026 14:00 UTC to 13 Apr 2026 09:00 UTC.
Read the chart in words
WhenWhatPrice
12 Apr 2026 19:00 UTCPoint A4,730.50
12 Apr 2026 23:00 UTCPoint B4,676.98
13 Apr 2026 00:00 UTCPoint C4,699.27
13 Apr 2026 03:00 UTCPoint D4,722.18

Times in UTC.

2. The first candle falls from 100 to 90 (ATR 8); the second opens at 90 and closes at 91; the third rises from 91 to 96 (hypothetical numbers). Is it a morning star under the rule of the book?

3. The morning star gives +0.12 ± 0.35 R over 27 trades and the evening star −0.22 ± 0.36 R over 28. What do you conclude?

4. The three outside with the confirmation of the next candle gives −0.34 ± 0.39 R over 19 trades. Why does entering with more confirmations not help?

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.