Part 4: The system under test · Chapter 14 of 14 · 16 min
A complete example, mistakes, reference, glossary and final test
A signal of version B read step by step from the bias to the exit, the most common mistakes in chapter order, the book's quick reference of rules and measurements, the glossary of terms and a 15-question final test.
A signal read step by step
Figure 1 puts the system together on H4, from 31 March to 7 April 2026, with a single signal of version B.
- Step 1, bias. The D1 trend is bearish (two falling significant highs and two falling lows): supply is aligned.
- Step 3, half of the range. The last H4 range runs from $4,482.84 (31 March at 00:00) to 4,794.47 (1 April at 20:00): the 50% is 4,638.66. The supply FVG sits above: premium.
- Level taken. On 1 April at 08:00 price goes past the swing high of 4,582.63 (30 March).
- Step 4, POI. On 2 April at 04:00 a supply FVG completes between 4,681.63 and 4,748.01 (0.91 ATR, ATR of 72.9 dollars).
- Step 2, structure. No break: the step is not satisfied (version B does not ask for it).
- Step 5, session. The order fills in the 12:00 candle (New York).
- Step 6, order. Sell limit at 4,681.63, stop at 4,766.24 (risk 84.61 dollars, 1.16 ATR), 2R target 4,512.41. With 100 dollars of risk the size is 100 ÷ 84.61 = 1.2 ounces.
- Outcome. Price touches neither the stop nor the target in 24 candles: time exit on 6 April at 12:00, at +0.16 R.
That day's journal reads: POI supply FVG, trend yes, half yes, take yes, break no, order at 4,681.63, time exit +0.16 R. It is neither a success nor a mistake: it is one case of B's sample.
Read the chart in words
H4 gold chart from 31 March to 7 April 2026. Two black lines mark the range high, 4,794.47 dollars, and low, 4,482.84, and a blue line the 50% at 4,638.66. A dashed pink line at 4,582.63 marks the swing high that was taken. A rectangle marks the supply FVG between 4,681.63 and 4,748.01. Horizontal lines mark the entry, the stop at 4,766.24 and the target at 4,512.41. A label marks the time exit.
| When | What | Price |
|---|---|---|
| 31 Mar 2026 12:00 UTC | High taken | 4,582.63 |
| 1 Apr 2026 08:00 UTC | take | 4,759.02 |
| 1 Apr 2026 20:00 UTC | Range high | 4,794.47 |
| 31 Mar 2026 12:00 UTC | Range low | 4,482.84 |
| 31 Mar 2026 12:00 UTC | Range 50% | 4,638.66 |
| 2 Apr 2026 00:00 UTC | Supply FVG | 4,681.63 – 4,748.01 |
| 2 Apr 2026 12:00 UTC | Entry | 4,681.63 |
| 2 Apr 2026 12:00 UTC | Stop | 4,766.24 |
| 2 Apr 2026 12:00 UTC | Target 2R | 4,512.41 |
| 6 Apr 2026 12:00 UTC | time exit | 4,652.60 |
Times in UTC.
The most common mistakes
| Mistake | Chapter |
|---|---|
| Believing a list of concepts is a system, with no written rules and no count | 1 |
| Stacking filters until the cases win: with four filters 4 POI remain | 2 |
| Believing the D1 bias shows the direction of the following days | 3 |
| Forgetting the bias delay: the swings can be weeks old | 3 |
| Treating the structure break as confirmation | 4 |
| Believing price in discount rises and in premium falls | 5 |
| Preferring a kind of POI for the result of one sample | 6 |
| Counting an overlapping order block and FVG as two cases | 6 |
| Picking the best time band after seeing the result | 7 |
| Entering deeper to raise the potential R | 8 |
| Comparing entry models without looking at how many POI enter | 8 |
| Believing break-even or the partial exit make the system safer | 9 |
| Computing the size from the target instead of from the risk | 10 |
| Choosing the risk per trade by looking only at the hoped-for result | 10 |
| Presenting the version with the highest result as "the system" | 11 |
| Presenting the best variant of a grid as a result | 12 |
| Taking a good month as proof | 13 |
| Keeping the journal only of the trades taken | 13 |
Quick reference: rules and thresholds
| Concept | Rule of the book | Ch. |
|---|---|---|
| System | Six steps with written rules: bias, structure, half of the range, POI, session and confirmation, order. A step enters only if the sample justifies it. | 1, 2 |
| D1 bias | Last two significant highs and lows (swings at 1.5 ATR) all rising = bullish, falling = bearish, otherwise undefined. | 3 |
| Break | H4 BOS or CHoCH in the direction of the POI, produced by the middle candle, the third or the one right after. | 4 |
| Half of the range | Last significant H4 range: demand below the 50%, supply above. | 5 |
| POI | Engine order block or FVG; overlapping if on the same side, born less than 10 candles apart and sharing price. | 6 |
| UTC bands | Asia 00-08, London 08-12, New York 12-20, evening 20-24 (opening hour of the H4 candle). | 7 |
| Entry models | Limit (near edge, 50%, far edge), market at the close of the first candle that touches the zone, H1 confirmation. | 8 |
| Order | Stop = far side + 0.25 ATR, risk 0.5-3 ATR, target at R, exit at 24 candles, cost 0.30 dollars. | 9 |
| Size | Risk in dollars ÷ stop distance in dollars per ounce. | 10 |
| Versions | A all POI, B trend and half, C B plus take, D at least two filters out of four, E at least three out of four. | 11 |
| Robustness | Grid of variants, random filters (1,000 trials), two halves of the period, different thresholds. | 12 |
| Routine | 00:00 levels and bias; 07:00 Asian range; at every H4 close POI filters; evening journal of all POI; week: review without changing the rules. | 13 |
Quick reference: the measurements
| Measurement | Result | Ch. |
|---|---|---|
| POI and baseline | 291 in 9.6 months; 216 trades, −0.10 ± 0.09 R (order blocks −0.20 ± 0.14, FVGs −0.04 ± 0.12) | 1 |
| Filter funnel | 291, 118 (trend), 37 (also half), 21 (also take), 4 (also break) | 2 |
| D1 bias in agreement | 56-58%; "always bearish" 54-56%; aligned POI −0.07 ± 0.15, counter-trend −0.09 ± 0.19 | 3 |
| H4 breaks | 94: carry on 46% (±10) against 36% up first for any candle; POI with a break −0.04 ± 0.17, without −0.13 ± 0.11 | 4 |
| Position in the range | 0-25%: up first 34%, down first 45%; above the high 50% against 30%; right half −0.11 ± 0.14, wrong half −0.10 ± 0.12 | 5 |
| Kind of POI | Overlapping +0.05 ± 0.18, not overlapping −0.16 ± 0.11; fake zones −0.12 | 6 |
| Bands of the fill | London +0.25 ± 0.26 (33), New York −0.25 ± 0.13 (97); more than 20 cuts | 7 |
| Entry models | Limit −0.10, 50% −0.23, far edge −0.38; market +0.05 ± 0.11; confirmation +0.11 ± 0.16 (67); control −0.12 ± 0.11 | 8 |
| Target and management | 1R −0.19, 2R −0.10, 3R −0.02; break-even −0.08; partial −0.14; level as target +0.05 ± 0.12 | 9 |
| Risk | Median maximum drawdown 16%, 30%, 52%, 68% at 0.5%, 1%, 2%, 3%; median losing streak 12 | 10 |
| System versions | B +0.35 ± 0.30 (25 trades), C +0.46 ± 0.42 (13), D −0.06 ± 0.14 (104), E +0.26 ± 0.31 (22) | 11 |
| Robustness | Grid 96/54; filters together beyond chance in 1-3% of trials; walk-forward +0.95 and 0.00; thresholds: B positive in 13 cells out of 16 | 12 |
| September 2026 | B 10 trades +4.82 R; baseline 27 trades −6.46 R; B in the month weighs more than half of its result | 13 |
All the measurements come from ten months of a single instrument (December 2025 - October 2026), with wide margins of error. None proves an operating edge.
Glossary
The book's terms are in the site glossary, where each one has an example with numbers.
| Term | Definition |
|---|---|
| SMC (Smart Money Concepts) | A way of reading the chart with three ideas: structure (highs and lows), liquidity (resting orders above highs and below lows) and imbalance (order blocks and FVGs), with the hypothesis that large operators leave traces. |
| SMC system in six steps | A sequence of steps, each with a written rule: bias, structure, half of the range, point of interest, session and confirmation, order (entry, stop, target, size). |
| Point of interest (POI) | A zone where an order is considered: in the book, an engine order block or an FVG, even overlapping. |
| Daily chart bias | The direction of the D1 trend: bullish if the last two significant highs and lows (swings at 1.5 ATR) are rising, bearish if they are falling, otherwise undefined. |
| Entry models | The ways of entering a POI: limit order (at the near edge, at the 50% or at the far edge), market at the close of the first candle that touches the zone, H1 confirmation. |
| H1 confirmation | Entry at the close of the first H1 candle that breaks structure in favour after the first touch of the POI, with the stop beyond the extreme touched between the touch and the confirmation. |
| Break-even | Moving the stop to the entry price when the trade has reached a certain profit (in the book +1R). |
| Partial exit | Closing part of the position at a preset profit (in the book half at +1R) and leaving the rest with the stop at the entry and the original target. |
| Losing streak | A number of consecutive losses. With a 29-33% target probability at 2R, the median streak over 216 trades is 11-13. |
| Walk-forward | A robustness test: you choose the best variant on one part of the data and test it on the other, without tweaking it. |
| Random-filter comparison (permutation) | The filters are assigned to POI at random and the same analysis is repeated (1,000 times) to see what chance finds with the same grid of variants. |
| Out-of-sample data | Data not used to choose the rules: the only way to know whether a result holds. |
| Signal journal | A record of every POI completed, including those discarded by the filter, with the filters, the decision and the outcome (real or hypothetical) in R. |
What to do now
- Keep a journal of all POI with the filters of the version you choose, written in advance. Your sample is the only one that concerns you.
- Count up to a hundred trades before changing a rule, and use a risk per trade that lets you bear a streak of twelve losses.
- If you want to retest B, use it on new data: more months, and at least one period with a rising D1 trend.
- For the foundations: the «Market structure» book, the «Order block from A to Z» book and the «FVG and liquidity» book. For risk and discipline: the risk and psychology course.
Final test
Fifteen questions on the book. You pass the test with at least 12 correct answers.
Final test of the book
1. In this book, what is an SMC system?
2. On the chart four candles carry a letter. Which one takes the swing high (the liquidity) before the supply FVG?
Read the chart in words
| When | What | Price |
|---|---|---|
| 6 Apr 2026 12:00 UTC | Point A | 4,694.80 |
| 1 Apr 2026 08:00 UTC | Point B | 4,759.02 |
| 2 Apr 2026 00:00 UTC | Point C | 4,794.27 |
| 2 Apr 2026 12:00 UTC | Point D | 4,694.23 |
Times in UTC.
3. Under the book's rule, when is the D1 trend bearish?
4. The swing bias agreed on 58% of days and "always bearish" on 56%. What do you conclude?
5. After a structure break price carried on 46% of the time, against 36-40% for any candle. What do you conclude?
6. In the lowest quarter of the range price rose first 34% of the time and fell first 45%. What do you conclude?
7. Why are 83 overlapping POI worth less than 83 independent cases?
8. Orders filled in London gave +0.25 R and those in New York −0.25 R, among more than twenty cuts of the same sample. What do you conclude?
9. On the chart the letters sit on four lines of a sell limit order: stop, target, the 50% of the range and entry. Which letter is the 50% of the range?
Read the chart in words
| When | What | Price |
|---|---|---|
| 6 Apr 2026 16:00 UTC | Point A | 4,766.24 |
| 6 Apr 2026 16:00 UTC | Point B | 4,512.41 |
| 6 Apr 2026 16:00 UTC | Point C | 4,638.66 |
| 6 Apr 2026 16:00 UTC | Point D | 4,681.63 |
Times in UTC.
10. Why does the limit order at the far edge give a worse result than the one at the near edge?
11. What does break-even do to the expected value?
12. The stop is 40 dollars from the entry (hypothetical numbers) and you accept risking 120 dollars. How many ounces do you open?
13. With 1% risk per trade and the system's real R values, the median maximum drawdown is 30%. What does it mean?
14. Version B gave +0.35 ± 0.30 R over 25 trades, and random subsets of 37 POI do at least as well 4% of the time. What do you conclude?
15. The choice made in the first half gave +0.95 R in the second and the one made in the second 0.00 R in the first. 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.