Part 2: One- and two-candle formations · Chapter 6 of 14 · 9 min
Harami and inside bars
The harami (a small candle inside the body of the previous one) appears 185 times on H1 gold, fewer than the 288 of a random series, and the trades give −0.10 ± 0.14 R; with confirmation +0.10 ± 0.26 R. The inside bar after a trend appears 439 times: the trend continues in 45% of cases (control 37%), but the trades give −0.06 ± 0.10 R.
The rules
The harami has two candles: the first has a body of at least 1 ATR; the second has a body between 0.05 ATR and half the body of the first, of the opposite colour and entirely inside the body of the first. In the bullish harami the first is bearish; in the bearish, the reverse. The inside bar has a lower high and a higher low than the previous candle (the "mother", with a range of at least 0.8 ATR) and the book counts it after a trend of at least 1.5 ATR in the previous 10 candles; the expected direction is that the trend continues. The three inside is a harami followed by a third candle that closes beyond the open of the first. The stop sits beyond the extreme of the candles plus 0.25 ATR (risk 0.5-4 ATR), the target equals the risk, 24 candles, cost 0.30 $.
How often they appear
| Formation | Found (chance, 5th-95th) | In expected direction | Trades | Mean R | Random (median) |
|---|---|---|---|---|---|
| Bullish harami | 92 (136, 126-145) | 39% | 91 | −0.15 ± 0.20 | −0.16 |
| Bearish harami | 93 (152, 142-163) | 40% | 91 | −0.05 ± 0.20 | −0.09 |
| All haramis | 185 (288, 273-302) | 39% | 182 | −0.10 ± 0.14 | −0.12 |
| Three inside | 7 (14, 10-19) | 43% | 7 | −0.15 ± 0.79 | −0.08 |
| Inside bar after a trend | 439 (470, 447-496) | 45% | 398 | −0.06 ± 0.10 | −0.05 |
The control is 37% up, 36% down, 27% no threshold. Haramis are fewer than chance (185 against 288; with blocks of 24 candles 257): as with engulfings, a small body inside a large body is less frequent in the real market than in the shuffled series. The expected direction comes in 39% of cases (±7 points), the same as the control. The 182 trades give −0.10 ± 0.14 R (82 targets, 91 stops, 9 time exits): in line with the random series (−0.12). First half −0.24 R, second +0.02 R.
With the confirmation (the next candle closes beyond the extreme of the harami in the expected direction) haramis give 55 confirmations out of 185 and 53 trades at +0.10 ± 0.26 R; the random series with the same rule give −0.13 R (from −0.33 to +0.04): gold sits above the 95th percentile. Bullish 24 trades at +0.21 ± 0.38 R, bearish 29 at +0.01 ± 0.37 R. It is one of the highest results of the book for a two-candle formation, with 53 trades and an error of ±0.26 R: a candidate to retest on new data, not a rule.
The inside bar and the break of the mother
Inside bars after a trend are 439 (470 in chance). The trend continues in 45% of cases (±5 points) and reverses in 34%, against 37% of the control: the direction is slightly in favour of the theory. But the 398 valid trades give −0.06 ± 0.10 R (184 targets, 190 stops, 24 time exits), as much as the random series (−0.05): the small direction edge is not enough to pay for cost and stop.
The break of the mother candle can be looked at on its own. Taking all 936 inside bars (without requiring the trend), the first break within 10 candles is upward in 45% of cases and downward in 46% (in 9% both sides or neither break). After a rise (224): up 42%, down 50%. After a fall (185): up 44%, down 50%. Without a trend (527): up 46%, down 43%. The most frequent break in both contexts is downward: it comes from the period (gold spent long stretches falling), not from the prior trend. In 48% of cases (448 out of 936) the next candle is itself inside the mother.
Three examples
Figure 1 shows a winning bullish harami: on 29 December 2025 at 15:00 (open 4,383.73 $, close 4,341.09 $, low 4,327.70 $) and at 16:00 (4,342.07 → 4,352.90 $), after a fall of 5.3 ATR. Entry at 4,352.90 $, stop at 4,320.62 $ (32.28 $, 1.1 ATR), target at 4,385.18 $ reached on 30 December at 05:00, thirteen candles later: +0.99 R. Figure 2 shows a harami that fails: on 17 February 2026 at 05:00 (4,954.85 → 4,893.15 $, low 4,866.07 $) and at 06:00 (4,893.15 → 4,913.47 $), after a fall of 2.4 ATR; entry at 4,913.47 $, stop at 4,860.34 $ (53.13 $, 2.3 ATR), target at 4,966.60 $; stop hit at 14:00: −1.01 R.
Figure 3 shows an inside bar after a rise: on 5 January 2026 the mother candle of 01:00 (4,385.05 → 4,399.98 $, high 4,404.74 $, low 4,385.05 $) contains the one of 02:00 (4,399.87 → 4,403.29 $, high 4,404.00 $, low 4,390.00 $), after a rise of 5.1 ATR. Entry at 4,403.29 $, stop at 4,382.33 $ (20.96 $, 1.9 ATR), target at 4,424.25 $ reached at 06:00: +0.99 R.
Read the chart in words
H1 gold chart of 29-30 December 2025 with the two numbered candles and the lines of entry, stop and target.
| When | What | Price |
|---|---|---|
| 29 Dec 2025 15:00 UTC | Point 1 | 4,327.70 |
| 29 Dec 2025 16:00 UTC | Point 2 | 4,341.10 |
| 29 Dec 2025 16:00 UTC | Entry | 4,352.90 |
| 29 Dec 2025 16:00 UTC | Stop | 4,320.62 |
| 29 Dec 2025 16:00 UTC | Target | 4,385.18 |
| 30 Dec 2025 05:00 UTC | target reached | 4,392.78 |
Times in UTC.
Read the chart in words
H1 gold chart of 16-17 February 2026 with the two numbered candles and the lines of entry, stop and target.
| When | What | Price |
|---|---|---|
| 17 Feb 2026 05:00 UTC | Point 1 | 4,866.07 |
| 17 Feb 2026 06:00 UTC | Point 2 | 4,890.32 |
| 17 Feb 2026 06:00 UTC | Entry | 4,913.47 |
| 17 Feb 2026 06:00 UTC | Stop | 4,860.34 |
| 17 Feb 2026 06:00 UTC | Target | 4,966.60 |
| 17 Feb 2026 14:00 UTC | stop hit | 4,845.33 |
Times in UTC.
Read the chart in words
H1 gold chart of 4-5 January 2026 with the mother candle and the inside candle numbered and the lines of entry, stop and target.
| When | What | Price |
|---|---|---|
| 5 Jan 2026 01:00 UTC | Point 1 | 4,385.05 |
| 5 Jan 2026 02:00 UTC | Point 2 | 4,390.00 |
| 5 Jan 2026 02:00 UTC | Entry | 4,403.29 |
| 5 Jan 2026 02:00 UTC | Stop | 4,382.33 |
| 5 Jan 2026 02:00 UTC | Target | 4,424.25 |
| 5 Jan 2026 06:00 UTC | target reached | 4,432.82 |
Times in UTC.
Procedure in four steps
- Write the rules of the harami and of the inside bar (bodies, range, trend) before looking for them.
- Count how often they appear and compare with more than one random shuffle.
- Look at the break of the mother in all contexts, not only those that confirm the theory.
- If you add a confirmation, count the trades, the error and the random benchmark, and retest on new data.
Exercises
Chapter exercises
1. On the chart the letters A-D are on four candles of 29 December 2025. Which letter is the small candle inside the body of the previous one (the second candle of the harami)?
Read the chart in words
| When | What | Price |
|---|---|---|
| 29 Dec 2025 12:00 UTC | Point A | 4,477.54 |
| 29 Dec 2025 15:00 UTC | Point B | 4,383.73 |
| 29 Dec 2025 16:00 UTC | Point C | 4,360.00 |
| 29 Dec 2025 19:00 UTC | Point D | 4,348.65 |
Times in UTC.
2. A bearish candle has a body from 100 to 90 (ATR 8). The next, bullish, has a body from 92 to 96 (hypothetical numbers). Is it a bullish harami under the rule of the book?
3. After a rise the first break of the mother of an inside bar is downward in 50% of cases, and after a fall in 50%. What does it mean?
4. The harami with confirmation gives +0.10 ± 0.26 R (53 trades), above the 95th percentile of the random series (+0.04). 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.