Part 3: Elliott waves · Chapter 11 of 14 · 9 min
From count to forecast: what happens after wave 2, 4 and 5
A count says what price should do: after wave 2 exceed the end of wave 1, after wave 5 correct. On gold, after wave 2 price reaches the end of wave 1 before breaking the start in 79% of cases (101), as after any correction; after wave 5 of the valid impulses the correction is no deeper than the one that follows other moves.
From count to forecast
A count has operating value only if it says what price will do and when it is wrong. Elliott's best-known forecasts are three:
- After wave 2: price exceeds the end of wave 1 (wave 3 is the strongest). The count is refuted if it breaks the start of wave 1.
- After wave 4: price exceeds the end of wave 3. The count is refuted if it enters the territory of wave 1.
- After wave 5: a correction starts that takes back at least part of the impulse (usually 38.2%-61.8%).
Each forecast is tested in real time: the swing P2 (or P4, P5) exists two candles after its extreme, and from there price is followed for 60 H4 candles (10 days).
After wave 2
Cases: the triples of swings P0, P1, P2 (H4 swings of at least 2 ATR, alternating) in which wave 2 does not break the start of wave 1. There are 101 (December 2025 - October 2026, gold). Price reaches the end of wave 1 before breaking the start, within 60 candles, in 78 cases out of 99 (79%, ±8 points); in the other 2 neither happens.
By depth of wave 2:
| Wave 2 / wave 1 | Cases | Reaches the end of wave 1 | Margin |
|---|---|---|---|
| up to 0.382 | 15 | 15 (100%) | |
| 0.382 - 0.5 | 12 | 8 (67%) | ±27 points |
| 0.5 - 0.618 | 14 | 11 (79%) | ±21 points |
| 0.618 - 1 | 58 | 44 (76%) | ±11 points |
Short corrections do better (15 out of 15), but they are 15 cases and the other bands sit between 67% and 79% with margins of 11-27 points. The Fibonacci value, between 0.5 and 0.618, gives 79%: like the whole. The more important point is another: this test is, by construction, the same as the benchmark (any leg with a correction that does not take back the whole leg reaches B before breaking A): 78 out of 99. A "counted" wave 2 is an ordinary retracement, and the forecast adds nothing to the structure that contains it.
Figure 1 shows a successful case: P0 at 5,133.05 $ (26 February 2026 at 12:00), P1 at 5,396.44 $, P2 at 5,275.08 $ (wave 2 retraces 46% of wave 1, confirmed on 1 March at 08:00). On 2 March at 04:00 price exceeds 5,396.44 $. Figure 2 shows a failed case: P0 at 4,370.58 $ (28 May at 04:00), P1 at 4,594.83 $, P2 at 4,451.52 $ (64% of wave 1, confirmed on 1 June at 20:00). Price does not reach 4,594.83 $: on 5 June at 12:00 it breaks 4,370.58 $ and the count falls.
Read the chart in words
H4 gold chart from 26 February to 2 March 2026 with the three swings numbered 0, 1, 2, the line of the end of wave 1 (target) and that of the start of wave 1 (invalidates) and the outcome label.
| When | What | Price |
|---|---|---|
| 26 Feb 2026 12:00 UTC | Point 0 | 5,133.05 |
| 28 Feb 2026 08:00 UTC | Point 1 | 5,396.44 |
| 1 Mar 2026 00:00 UTC | Point 2 | 5,275.08 |
| 1 Mar 2026 08:00 UTC | end of wave 1: target | 5,396.44 |
| 1 Mar 2026 08:00 UTC | start of wave 1: invalidates | 5,133.05 |
| 2 Mar 2026 04:00 UTC | reached | 5,420.00 |
Times in UTC.
Read the chart in words
H4 gold chart from 27 May to 6 June 2026 with the three swings numbered 0, 1, 2, the line of the end of wave 1 (target) and that of the start of wave 1 (invalidates) and the outcome label.
| When | What | Price |
|---|---|---|
| 28 May 2026 04:00 UTC | Point 0 | 4,370.58 |
| 29 May 2026 12:00 UTC | Point 1 | 4,594.83 |
| 1 Jun 2026 12:00 UTC | Point 2 | 4,451.52 |
| 1 Jun 2026 20:00 UTC | end of wave 1: target | 4,594.83 |
| 1 Jun 2026 20:00 UTC | start of wave 1: invalidates | 4,370.58 |
| 5 Jun 2026 12:00 UTC | invalidated | 4,336.00 |
Times in UTC.
After wave 4
Cases: windows of five swings in which rules 1 and 3 are already met. There are 13. Price reaches the end of wave 3 before entering the territory of wave 1, within 60 candles, in 10 cases out of 12 (83%, ±21 points; in one case neither happens). The benchmark (any leg with a correction that does not take it all back) is 79%: the difference is 4 points with a margin of 21.
After wave 5
Cases: the 13 valid impulses followed by at least 60 candles of data. We measure how much price retraces at most of the whole impulse (from P0 to P5) in the 60 candles after the confirmation of P5. The right benchmark is made with similar moves that do not meet the rules: the 137 windows of six swings that break at least one and end beyond the start.
| Maximum retracement of the whole move | Valid impulses (13) | Windows that break the rules (137) |
|---|---|---|
| At least 38.2% | 11 (85%) | 124 (91%) |
| At least 50% | 9 (69%) | 117 (85%) |
| At least 61.8% | 7 (54%) | 104 (76%) |
| Median | 0.89 | 1.06 |
After wave 5 the correction is almost always there (85% at least at 38.2%), but it is also there after moves that are not impulses (91%), and a little deeper. No margin lets you say that the complete impulse foretells a more pronounced correction. Figure 3 shows a case: the impulse from 24 February 2026 (5,101.20 $) to 2 March (5,420.00 $, 318.80 $). After the confirmation of P5 price falls to 5,004.60 $ on 3 March at 12:00: 130% of the impulse. The correction does not stop at a Fibonacci level, it takes back the whole impulse and goes beyond.
Read the chart in words
H4 gold chart from 24 February to 6 March 2026 with the six swings of the impulse numbered 0 to 5, the 38.2%, 50% and 61.8% levels of the whole impulse and the label of the maximum retracement.
| When | What | Price |
|---|---|---|
| 24 Feb 2026 12:00 UTC | Point 0 | 5,101.20 |
| 25 Feb 2026 16:00 UTC | Point 1 | 5,218.80 |
| 26 Feb 2026 12:00 UTC | Point 2 | 5,133.05 |
| 28 Feb 2026 08:00 UTC | Point 3 | 5,396.44 |
| 1 Mar 2026 00:00 UTC | Point 4 | 5,275.08 |
| 2 Mar 2026 04:00 UTC | Point 5 | 5,420.00 |
| 3 Mar 2026 12:00 UTC | maximum retracement: 130% | 5,004.60 |
Times in UTC.
Procedure in four steps
- Write the forecast with a target and a level that refutes it.
- Start from the confirmation of the swing and count all the cases, including refuted ones and those with no outcome.
- Compare with the same event without the count (any retracement, moves that break the rules).
- Give the margin of error and how many cases you have.
Exercises
Chapter exercises
1. On the chart the four letters sit on P0, P1, P2 and on the outcome candle. Which letter marks the level that, if broken, refutes the wave 2 count?
Read the chart in words
| When | What | Price |
|---|---|---|
| 26 Feb 2026 12:00 UTC | Point A | 5,133.05 |
| 28 Feb 2026 08:00 UTC | Point B | 5,396.44 |
| 1 Mar 2026 00:00 UTC | Point C | 5,275.08 |
| 2 Mar 2026 04:00 UTC | Point D | 5,420.00 |
Times in UTC.
2. A wave 1 goes from 1,000 to 1,100 and wave 2 falls to 1,050 (hypothetical numbers). Which level refutes the count if broken?
3. After wave 2 price reaches the end of wave 1 before breaking the start in 78 cases out of 99 (79%), and after any retracement it is the same 79%. What does it mean?
4. After wave 5 the correction reaches at least 61.8% of the impulse in 54% of the valid impulses (13) and in 76% of the windows that break the rules (137). 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.