Part 1: The Fibonacci ratios · Chapter 3 of 14 · 8 min

Extensions and projections: where the leg goes after the correction

How a leg is projected (C + 1 × AB, 1.272 × AB, 1.618 × AB) and what happens on gold: the ratio CD/AB has a median of 0.70 after corrections up to 100%, falls near 1, 1.272 or 1.618 in 10% of cases, and price reaches a projection less and less often the farther it is, with the Fibonacci levels on the same curve as their neighbors.

The extensions

After a leg AB and a correction to C, the idea of projections is that the next leg (from C to D) has a length tied to AB: equal (100%, the "AB = CD" case), or 127.2%, or 161.8% of AB. The expected price of D is C ± e × |AB|, in the direction of AB.

An example with real numbers: the down leg from 4,373.53 $ (A, 22 September 2026 at 16:00) to 4,249.11 $ (B, 24 September at 08:00) is 124.42 $ long. The correction climbs to C = 4,318.68 $ (25 September at 08:00). The projections are 4,318.68 − e × 124.42: at 100% 4,194.26 $, at 127.2% 4,160.42 $, at 161.8% 4,117.37 $.

Where the leg really goes

On the 210 sequences A → B → C → D (H4 swings of at least 2 ATR, December 2025 - October 2026) the ratio CD/AB is measured. Taking only the corrections that do not take back the whole leg (r up to 1: 104 cases), the median of CD/AB is 0.70 (25th percentile 0.46, 75th percentile 1.10). With all the sequences the median is 0.95. By band, in the 104 cases with r up to 1:

Band of CD/ABCases
0 - 0.38215
0.382 - 0.61831
0.618 - 0.78613
0.786 - 117
1 - 1.2726
1.272 - 1.6184
1.618 - 2.61815
above 2.6183

In most cases the leg CD is shorter than AB: 76 out of 104 are below 1. Within ±0.05 of 1, 1.272 or 1.618 fall 24 sequences out of 210 (11.4%, ±4.3 points): 12 near 1, 6 near 1.272, 6 near 1.618. With only the corrections up to 100% it is 10 out of 104 (9.6%, ±5.7 points). As in chapter 1, we compare with the same three levels shifted all together (from ±0.12 to ±0.30): 5 shifts out of 10 get at least 24 hits (median 25.5) and, on the corrections up to 100%, 5 out of 10 get at least 10 (median 8.5). The Fibonacci levels are in the middle of the others.

Figure 1 shows an almost perfect "AB = CD" case: the leg from 4,540.03 to 4,889.61 $ (AB = 349.58 $), the correction to 4,755.59 $, then the high of 5,110.47 $ on 26 January: CD = 354.88 $, ratio 1.015, 5.30 $ from the 100% projection (5,105.17 $). Figure 2 shows a 161.8% case: AB = 124.42 $ as above, D at 4,116.60 $ on 28 September at 12:00, CD = 202.08 $, ratio 1.624, 77 cents from the projection (4,117.37 $). Figure 3 shows the other extreme: the leg from 4,370.58 $ (28 May at 04:00) to 4,594.83 $ (29 May at 12:00), AB = 224.25 $, correction to 4,451.52 $ and a rise to 4,543.07 $ on 2 June: CD = 91.55 $, ratio 0.408, far from all the projections (the first, at 100%, was at 4,675.77 $).

Figure 1. AB = CD: leg from 4,540.03 to 4,889.61 (349.58), correction to 4,755.59 and high at 5,110.47 (354.88, ratio 1.015, 5.30 from the 100% projection).Data: gold, XAUUSDT perpetual future (real candles) · 4 hours (H4) · from 15 Jan 2026 20:00 UTC to 27 Jan 2026 16:00 UTC.
Read the chart in words

H4 gold chart from 15 to 27 January 2026 with the broken line A, B, C, D and the projections at 100%, 127.2% and 161.8% of AB from C.

WhenWhatPrice
21 Jan 2026 16:00 UTC100%5,105.17
21 Jan 2026 16:00 UTC127.2%5,200.26
21 Jan 2026 16:00 UTC161.8%5,321.21
16 Jan 2026 12:00 UTCPoint A4,540.03
21 Jan 2026 04:00 UTCPoint B4,889.61
21 Jan 2026 16:00 UTCPoint C4,755.59
26 Jan 2026 04:00 UTCPoint D5,110.47

Times in UTC.

Figure 2. A 161.8% leg: AB = 124.42, CD = 202.08 (ratio 1.624), D at 4,116.60 against a projection of 4,117.37.Data: gold, XAUUSDT perpetual future (real candles) · 4 hours (H4) · from 22 Sept 2026 08:00 UTC to 30 Sept 2026 04:00 UTC.
Read the chart in words

H4 gold chart from 22 to 30 September 2026 with the broken line A, B, C, D and the projections at 100%, 127.2% and 161.8% of AB from C.

WhenWhatPrice
25 Sept 2026 08:00 UTC100%4,194.26
25 Sept 2026 08:00 UTC127.2%4,160.42
25 Sept 2026 08:00 UTC161.8%4,117.37
22 Sept 2026 16:00 UTCPoint A4,373.53
24 Sept 2026 08:00 UTCPoint B4,249.11
25 Sept 2026 08:00 UTCPoint C4,318.68
28 Sept 2026 12:00 UTCPoint D4,116.60

Times in UTC.

Figure 3. A leg that stays short: AB = 224.25, CD = 91.55 (ratio 0.408), D at 4,543.07 against a 100% projection of 4,675.77.Data: gold, XAUUSDT perpetual future (real candles) · 4 hours (H4) · from 27 May 2026 20:00 UTC to 4 Jun 2026 16:00 UTC.
Read the chart in words

H4 gold chart from 27 May to 4 June 2026 with the broken line A, B, C, D and the projections at 100%, 127.2% and 161.8% of AB from C.

WhenWhatPrice
1 Jun 2026 12:00 UTC100%4,675.77
1 Jun 2026 12:00 UTC127.2%4,736.77
1 Jun 2026 12:00 UTC161.8%4,814.36
28 May 2026 04:00 UTCPoint A4,370.58
29 May 2026 12:00 UTCPoint B4,594.83
1 Jun 2026 12:00 UTCPoint C4,451.52
2 Jun 2026 04:00 UTCPoint D4,543.07

Times in UTC.

Reaching the projection, in real time

The measurement above uses D, which is known afterwards. The real-time test starts from the confirmation of C (two candles later) and follows price for 60 candles: is the projection C + e × AB touched before C is broken on the opposite side? If C is broken, the projection no longer holds. There are 101 corrections up to 100%. Alongside the Fibonacci levels we try nearby levels, which have no name:

Projection eReached before C breaksMargin
0.575%±8 points
0.61864%±9 points
0.7555%±10 points
140%±10 points
1.1538%±9 points
1.27233%±9 points
1.433%±9 points
1.533%±9 points
1.61829%±9 points
1.7527%±9 points
2.217%±7 points
2.61813%±7 points

The probability of reaching the projection falls steadily with distance, from 75% at 0.5 to 13% at 2.618. The Fibonacci levels (0.618, 1, 1.272, 1.618, 2.618) sit on the same curve as the unnamed levels: at 1.272 33%, at 1.4 and 1.5 33%; at 1.618 29%, at 1.75 27%. A price target at 1.618 is reached about once in three and a half times, like one at 1.75, and not because it sits at 161.8%.

Procedure in four steps

  1. Compute the projections C ± e × AB only after the confirmation of C, not after D.
  2. Count all the sequences and the distribution of CD/AB, not only the successful cases.
  3. Measure how often each projection is reached before C is broken.
  4. Compare with nearby levels: if the curve is smooth, the level has no effect of its own.

Exercises

Chapter exercises

1. On the chart the four lines are projections of AB from C, each with a letter. Which letter is the 161.8% projection of AB?

Chart exercise. Data: gold, XAUUSDT perpetual future (real candles) · 4 hours (H4) · from 22 Sept 2026 08:00 UTC to 30 Sept 2026 04:00 UTC.
Read the chart in words
WhenWhatPrice
25 Sept 2026 20:00 UTCPoint A4,160.42
26 Sept 2026 12:00 UTCPoint B4,117.37
27 Sept 2026 04:00 UTCPoint C4,194.26
27 Sept 2026 20:00 UTCPoint D4,241.79

Times in UTC.

2. A leg AB of 100 $ (hypothetical numbers) is followed by a correction to C. Where is the 161.8% projection if the leg goes up and C is 1,000?

3. Price reaches the 1.618 projection in 29% of cases, the 1.75 in 27% and the 1.5 in 33%. What does it mean?

4. Only 24 sequences out of 210 (11.4%) have CD/AB within ±0.05 of 1, 1.272 or 1.618, and moving the levels all together 5 shifts out of 10 do at least as well. 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.