Part 1: How a candle is tested · Chapter 2 of 14 · 8 min

Candle anatomy

Body and wicks must be measured in ATR: the same 10 $ body is worth 4.1 ATR on a calm night and 0.1 ATR in a crisis. The marubozu (body ≥ 90% of the range) appears 179 times on H1 gold, more than the 124 of a random series, but does not continue: 37% against 37% of the control and −0.17 ± 0.15 R.

The proportions of a candle

On H1 gold the body takes up a median 46% of the range; the upper wick 22% and the lower wick 23%. The book classifies every candle by the size of the body in ATR:

BodyShare of candles
Small (under 0.25 ATR)38%
Medium (0.25-0.75 ATR)41%
Large (0.75-1.5 ATR)16%
Huge (1.5 ATR and more)5%

The same figure in dollars can fall in opposite classes. On 1 August 2026 at 19:00 a candle has a body of 9.53 $ with an ATR of 2.33 $: it is worth 4.1 ATR. On 31 January 2026 at 02:00 another has a body of 9.66 $ with an ATR of 79.34 $: it is worth 0.1 ATR. Among the 203 candles with a body of about 10 $ the ATR goes from 2.33 to 79.34 $ (median 16.60 $). That is why the rules of the book never use dollars.

Figure 1. A body of 9.53 $ on a calm night (ATR 2.33 $): 4.1 ATR, a huge candle (1 August 2026, 19:00).Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 1 Aug 2026 07:00 UTC to 2 Aug 2026 06:00 UTC.
Read the chart in words

H1 gold chart from 1 to 2 August 2026 with the 19:00 candle and the measure of its body in ATR.

WhenWhatPrice
1 Aug 2026 19:00 UTCbody $9.53 = 4.1 ATR4,058.90

Times in UTC.

Figure 2. A body of 9.66 $ in a crisis (ATR 79.34 $): 0.1 ATR, a small candle (31 January 2026, 02:00).Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 30 Jan 2026 14:00 UTC to 31 Jan 2026 13:00 UTC.
Read the chart in words

H1 gold chart from 30 to 31 January 2026 with the 02:00 candle and the measure of its body in ATR.

WhenWhatPrice
31 Jan 2026 02:00 UTCbody $9.66 = 0.1 ATR4,884.30

Times in UTC.

What follows a large candle

After a candle of each class, the next candle has the same colour in 46%-50% of cases: no class gives a more predictable next candle. Within 10 candles price touches a ±1.5 ATR threshold more often after large candles, but in both directions.

After a candleContinuesReversesNo threshold
Small bullish32%32%36%
Large bullish43%39%18%
Huge bullish44%46%10%
Small bearish32%34%35%
Large bearish42%42%16%
Huge bearish42%44%15%

The control (any candle) is 37% up, 36% down and 27% no threshold. After a huge candle the threshold is touched in 85%-90% of cases, but the direction is a coin toss: size announces movement, not direction.

The marubozu

A marubozu is a candle with almost no wicks. Rule of the book: body of at least 90% of the range and range of at least 1 ATR. It is bullish if it closes above the open, bearish otherwise. Theory reads it as continuation: the direction of the candle should go on.

On H1 gold it appears 179 times in 9.7 months (101 bullish and 78 bearish). A random series with the same candles has a median of 124 (5th-95th percentile 117-133): fewer. But shuffling in blocks of 6 candles breaks up periods of high volatility and a sudden explosion weighs more against a still low ATR. With blocks of 24 candles (closer to the reality of agitated days) the random series have 156 (from 145 to 165): the difference shrinks but stays. Of the 179, 66 have a range of at least 2 ATR relative to the ATR of the previous candle.

After the marubozu price continues in 37% of cases (±7 points) and reverses in 44%, no threshold in 19%; the control is 37%/36%/27% and the random series 38%. The 164 valid trades (15 skipped for risk outside 0.5-4 ATR) give −0.17 ± 0.15 R: 67 targets, 85 stops, 12 time exits. The random series give a median −0.06 R (from −0.16 to +0.07): gold is worse than 96% of them. Bullish 92 trades at −0.17 ± 0.20 R, bearish 72 at −0.18 ± 0.22 R; first half −0.20 R, second −0.15 R. The result is consistent with price pulling back after an explosive candle (the stop sits on the extreme of the candle), but it is one of the 29 formations the book tests and has to be retested on new data.

Figure 3 shows a bullish marubozu that works: on 27 January 2026 at 20:00 a candle from 5,098.20 to 5,178.90 $ (high 5,185.67, low 5,096.03), a range of 3.4 ATR. Entry at 5,178.90 $, stop at 5,089.46 $ (89.44 $), target at 5,268.34 $ reached on 28 January at 05:00: +1.00 R. Figure 4 shows a bearish one that fails: on 24 January at 19:00 a candle from 5,018.45 to 5,002.03 $; entry at 5,002.03 $, stop at 5,021.91 $, target at 4,982.15 $; the stop is hit at 21:00: −1.02 R.

Figure 3. A bullish marubozu that works: 5,098.20 → 5,178.90, entry 5,178.90, stop 5,089.46, target 5,268.34 reached on 28 January 2026 at 05:00 (+1.00 R).Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 27 Jan 2026 12:00 UTC to 28 Jan 2026 11:00 UTC.
Read the chart in words

H1 gold chart from 27 to 28 January 2026 with the numbered candle and the lines of entry, stop and target.

WhenWhatPrice
27 Jan 2026 20:00 UTCPoint 15,096.03
27 Jan 2026 20:00 UTCEntry5,178.90
27 Jan 2026 20:00 UTCStop5,089.46
27 Jan 2026 20:00 UTCTarget5,268.34
28 Jan 2026 05:00 UTCtarget reached5,272.12

Times in UTC.

Figure 4. A bearish marubozu that fails: 5,018.45 → 5,002.03, entry 5,002.03, stop 5,021.91 hit on 24 January 2026 at 21:00 (−1.02 R).Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 24 Jan 2026 11:00 UTC to 25 Jan 2026 03:00 UTC.
Read the chart in words

H1 gold chart of 24 January 2026 with the numbered candle and the lines of entry, stop and target.

WhenWhatPrice
24 Jan 2026 19:00 UTCPoint 15,018.55
24 Jan 2026 19:00 UTCEntry5,002.03
24 Jan 2026 19:00 UTCStop5,021.91
24 Jan 2026 19:00 UTCTarget4,982.15
24 Jan 2026 21:00 UTCstop hit5,027.45

Times in UTC.

Procedure in four steps

  1. Measure body and wicks in ATR, not in dollars.
  2. Classify the candle (small, medium, large, huge) and remember that the class announces movement, not direction.
  3. For the marubozu, count how often it appears and compare with random series using more than one kind of shuffling.
  4. If you trade, put the stop beyond the extreme and compute the R with the cost.

Exercises

Chapter exercises

1. On the chart the letters A-D are on four candles of 27 January 2026. Which letter is the marubozu (body ≥ 90% of the range and range of at least 1 ATR)?

Chart exercise. Data: gold, XAUUSDT perpetual future (real candles) · 1 hour (H1) · from 27 Jan 2026 10:00 UTC to 28 Jan 2026 05:00 UTC.
Read the chart in words
WhenWhatPrice
27 Jan 2026 17:00 UTCPoint A5,090.00
27 Jan 2026 19:00 UTCPoint B5,102.99
27 Jan 2026 20:00 UTCPoint C5,185.67
27 Jan 2026 22:00 UTCPoint D5,195.00

Times in UTC.

2. A body of 10 $ is worth 4.1 ATR when the ATR is 2.33 $ and 0.1 ATR when it is 79.34 $. What does it teach?

3. A candle has a range of 20 (from 1,000 to 1,020) and a body of 19, with an ATR of 15 (hypothetical numbers). Is it a marubozu under the rule of the book?

4. After a marubozu price continues in 37% of cases, as in the control, and the trades give −0.17 ± 0.15 R against −0.06 of the random series. What do you conclude?

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.