Part 4: A system, mistakes and test · Chapter 13 of 14 · 9 min
An Elliott and Fibonacci system under test
Entry on wave 2 at 50% or 61.8% of wave 1, stop below the start, target at the end of wave 1 or wave 3 at 100% or 161.8%: 6 variants on 211 gold legs, mean R from −0.28 to −0.03, always below random entries with the same geometry (from +0.01 to +0.14). With hindsight the same legs give +1.36 R.
The system
The only way to use Elliott in real time is to enter on wave 2 to ride wave 3, so:
- Wave 1: a leg A → B (H4 swings of at least 2 ATR, B confirmed two candles later).
- Entry (wave 2): limit order at the level e of wave 1 (50% or 61.8%), in the direction of wave 1. Valid for 40 candles; it lapses if a candle closes beyond A.
- Stop: beyond A by 0.25 ATR (rule 1 says that below A the count is refuted). Risk between 0.5 and 3 ATR, otherwise the trade is skipped.
- Target (wave 3): the end of wave 1 (B), or wave 3 = 100% or 161.8% of wave 1, measured from the entry.
- Exit: 60 candles after the fill. Cost 0.30 $ per ounce.
The 211 H4 gold legs (December 2025 - October 2026) give, for the 61.8%, 144 trades (44 orders never filled, 23 skipped for risk).
The result
Mean R with the 95% interval, and the comparison with random entries: same stop and target distance in dollars, but random candle and direction (2,000 trials).
| Level and target | Trades | Mean R | Winners | Target (in R) | Random entries: median (5th-95th) | Random trials with R at least equal |
|---|---|---|---|---|---|---|
| 50%, end of wave 1 | 129 | −0.25 ± 0.16 | 40% | 0.86 | −0.03 (−0.16 / +0.10) | 100% |
| 50%, wave 3 = 100% | 129 | −0.28 ± 0.20 | 28% | 1.72 | +0.05 (−0.13 / +0.24) | 100% |
| 50%, wave 3 = 161.8% | 129 | −0.21 ± 0.25 | 25% | 2.79 | +0.11 (−0.12 / +0.34) | 99% |
| 61.8%, end of wave 1 | 144 | −0.12 ± 0.19 | 38% | 1.35 | +0.01 (−0.14 / +0.17) | 92% |
| 61.8%, wave 3 = 100% | 144 | −0.17 ± 0.22 | 27% | 2.19 | +0.07 (−0.13 / +0.27) | 98% |
| 61.8%, wave 3 = 161.8% | 144 | −0.03 ± 0.29 | 24% | 3.55 | +0.14 (−0.12 / +0.39) | 85% |
None of the 6 variants is positive: the best (61.8% with a 161.8% target) makes −0.03 ± 0.29. Random entries with the same geometry give between −0.03 and +0.14 R, and the system sits below most of the trials (from 85% to 100%). Part of the gap comes from the mechanism of the limit order: the order fills when price is moving against the position, and the same effect is in books 2-4. The 161.8% of wave 1 corresponds on average to 3.55 times the risk: 22% winners would be needed to break even, the winners are 24% and the result stays at −0.03 R.
Splitting the period, for the 61.8%: with target B the first half gives −0.10 ± 0.27 (71 trades) and the second −0.15 ± 0.26 (73); with wave 3 = 100% −0.23 ± 0.31 and −0.12 ± 0.33; with 161.8% −0.05 ± 0.41 and 0.00 ± 0.42.
Figure 1 shows a winning trade with a 161.8% target: wave 1 from 4,601.50 to 4,703.87 $ (6 April 2026), order at 61.8% at 4,640.61 $ filled on 7 April at 00:00, stop at 4,597.20 $, target at 4,806.24 $ (3.82 times the risk) reached on 7 April at 20:00: +3.81 R. Figure 2 shows a loss: wave 1 from 4,652.74 to 4,505.00 $, order at 4,596.30 $ filled on 5 May at 20:00, stop at 4,657.91 $ hit on 6 May at 04:00: −1.00 R (target at 4,357.26 $, 3.88 times the risk).
Read the chart in words
H4 gold chart from 5 to 8 April 2026 with points 0 and 1 of wave 1, the entry line, the stop line (4,597.20) and the target line (4,806.24).
| When | What | Price |
|---|---|---|
| 6 Apr 2026 00:00 UTC | Point 0 | 4,601.50 |
| 6 Apr 2026 08:00 UTC | Point 1 | 4,703.87 |
| 7 Apr 2026 00:00 UTC | Entry (61.8%) | 4,640.61 |
| 7 Apr 2026 00:00 UTC | Stop | 4,597.20 |
| 7 Apr 2026 00:00 UTC | Target 161.8% | 4,806.24 |
| 7 Apr 2026 20:00 UTC | target reached | 4,833.73 |
Times in UTC.
Read the chart in words
H4 gold chart from 3 to 6 May 2026 with points 0 and 1 of wave 1, the entry line, the stop line (4,657.91) and the target line (4,357.26).
| When | What | Price |
|---|---|---|
| 3 May 2026 20:00 UTC | Point 0 | 4,652.74 |
| 4 May 2026 16:00 UTC | Point 1 | 4,505.00 |
| 5 May 2026 20:00 UTC | Entry (61.8%) | 4,596.30 |
| 5 May 2026 20:00 UTC | Stop | 4,657.91 |
| 5 May 2026 20:00 UTC | Target 161.8% | 4,357.26 |
| 6 May 2026 04:00 UTC | stop hit | 4,682.64 |
Times in UTC.
The edge of hindsight
Of all the 211 legs, 14 were, afterwards, wave 1 of a valid impulse (chapter 7). If you take only those, with the order at 61.8%:
| Legs | Trades | Mean R (target B) | Mean R (wave 3 = 100%) |
|---|---|---|---|
| Wave 1 of a valid impulse (14, chosen with hindsight) | 8 | +1.36 ± 0.04 | +2.21 ± 0.06 |
| All the others (197) | 136 | −0.21 ± 0.19 | −0.31 ± 0.22 |
The result is splendid and worthless: an impulse is valid only after wave 5 is confirmed, five swings after the moment you should enter (in figure 3 the leg from 2 to 3 June 2026 gives an order at 4,498.18 $ filled on 4 June at 08:00, with the stop at 4,550.80 $ and the target at 4,308.03 $ reached on 6 June at 04:00, +3.61 R; the impulse it belongs to completes only on 8 June at 04:00, at 4,271.20 $). In real time you do not know which leg will become an impulse. Selecting trades from the final count is a forecasting error: it uses in the past information that exists only in the future.
Read the chart in words
H4 gold chart from 1 to 8 June 2026 with the broken line of the valid impulse in dashed grey, points 0 and 1 of wave 1, the entry line, the stop line (4,550.80) and the target line (4,308.03).
| When | What | Price |
|---|---|---|
| 2 Jun 2026 04:00 UTC | Point 0 | 4,543.07 |
| 3 Jun 2026 20:00 UTC | Point 1 | 4,425.55 |
| 4 Jun 2026 08:00 UTC | Entry (61.8%) | 4,498.18 |
| 4 Jun 2026 08:00 UTC | Stop | 4,550.80 |
| 4 Jun 2026 08:00 UTC | Target 161.8% | 4,308.03 |
| 6 Jun 2026 04:00 UTC | target reached | 4,304.34 |
Times in UTC.
Procedure in five steps
- Write the system with every rule applicable in real time: swing, entry, stop, target, exit, cost.
- Count all the legs, including orders never filled and trades skipped.
- Compare with random entries with the same stop and target geometry.
- Split the period and check that the two halves say the same thing.
- Distrust any filter that uses information available only after the trade.
Exercises
Chapter exercises
1. Wave 1 goes from 1,000 to 1,100 (hypothetical numbers). Order at 61.8% at 1,038.20, stop at 1,000 and target at 161.8% of wave 1 from the entry (1,200.00). How many R is the target worth?
2. The system (61.8%, 161.8% target) gives −0.03 ± 0.29 R, and random entries with the same geometry give +0.14 R (5th-95th: −0.12 / +0.39). What can be said?
3. The 14 legs that were wave 1 of a valid impulse give +1.36 R (8 trades) and the others −0.21 R. What is wrong with the comparison?
4. Why does the system sit below random entries with the same geometry?
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.