Part 4: Trading and the system · Chapter 14 of 14 · 19 min
A complete day, mistakes, reference, glossary and final test
A real day (17 March 2026) with all the levels and six trades out of six on target, read next to the ten-month average (−0.03 R per trade), the most common mistakes in chapter order, the quick reference of rules and measurements, the glossary of terms and a 15-question final test.
A day read with the whole book
Figure 1 shows Tuesday 17 March 2026 (trading day from 18:00 New York time on the 16th to 17:00, UTC hours). The levels, computed at 18:00 New York time as in chapter 3:
- yesterday's high, 5,037.72 $, and yesterday's close, 5,007.24 $ (yesterday's open is at 5,008.75 $);
- the previous week's low, 5,013.91 $ (chapter 4);
- a confirmed H4 swing, at 5,004.60 $ (chapter 2);
- the round number 5,000 (chapter 5).
Four of these levels are within 14 $ of each other, between 5,000 and 5,014: a confluence of four families (chapter 10). The day has a range of 65.4 $, with an average ATR of 19 $.
The price oscillates all day between 5,000 and 5,038: 13 contacts with levels, 6 holds and 7 breaks (in the chapter 1 sense: the first target at 1 ATR from the contact point). With rule A of chapter 13 (bounce at the close of the contact candle, stop half an ATR beyond the zone, target 1 R) the day gives six trades, all on target, +5.9 R: a sale at 5,013.59 $ at 23:00 on the 16th, one at 5,036.92 $ at 04:00 (on yesterday's high), a purchase at 5,010.14 $ at 06:00, one at 5,001.75 $ at 10:00 (on the round number), a sale at 5,016.71 $ at 13:00 and one at 5,003.70 $ at 17:00.
It is the day that convinces. A trader who had traded only this day would conclude that levels work, that a confluence of four levels is a strong point, that round numbers hold. And they would be wrong: over ten months the same rule (552 trades) gives −0.03 ± 0.08 R per trade, like random direction (−0.04), and 17 March is a day of oscillation between 5,000 and 5,038 $ in which the bounce worked six times in a row: a coincidence, not a property of levels.
Read the chart in words
H1 gold chart of 17 March 2026 with five levels marked and six numbered trades.
| When | What | Price |
|---|---|---|
| 16 Mar 2026 22:00 UTC | yesterday's high 5037.72 | 5,037.72 |
| 16 Mar 2026 22:00 UTC | previous week's low 5013.91 | 5,013.91 |
| 16 Mar 2026 22:00 UTC | yesterday's close 5007.24 | 5,007.24 |
| 16 Mar 2026 22:00 UTC | H4 swing 5004.60 | 5,004.60 |
| 16 Mar 2026 22:00 UTC | round number 5000 | 5,000.00 |
| 16 Mar 2026 23:00 UTC | Point 1 | 5,015.35 |
| 17 Mar 2026 04:00 UTC | Point 2 | 5,040.74 |
| 17 Mar 2026 06:00 UTC | Point 3 | 5,005.34 |
| 17 Mar 2026 10:00 UTC | Point 4 | 4,993.88 |
| 17 Mar 2026 13:00 UTC | Point 5 | 5,030.30 |
| 17 Mar 2026 17:00 UTC | Point 6 | 5,010.99 |
Times in UTC.
The most common mistakes
| Mistake | Chapter |
|---|---|
| Drawing levels on a finished chart and counting how many "held"; remembering the bounces and forgetting the breaks | 1 |
| Using a thin line; trying types of levels without a list fixed beforehand; drawing the swing before confirmation | 2 |
| Choosing the day when yesterday's levels held five times out of six; attributing to the level a difference that belongs to the hour | 3 |
| Comparing holds of different time bands without a control in the same band; drawing conclusions from month levels with 35 contacts | 4 |
| Telling the story of a high that fell on a round number as proof that round numbers count, ignoring the 400 extremes that do not | 5 |
| Saying "tested four times, so it is solid"; counting levels broken at the first touch as weak | 6 |
| Saying "breaks are almost always false (or true)" without saying with which definition | 7 |
| Telling the stories of retests that held; putting the stop a few dollars from the old level | 8 |
| Saying trendlines do not hold after deleting them at the break; using the "third touch" as confirmation | 9 |
| Telling the story of the three-level confluence and the bounce, without the cases where the price went through; presenting the best of 19 pairs | 10 |
| Presenting the best combination of stop and target among 24; looking at total R without error or random direction | 11 |
| Putting the stop "just beyond the level"; believing in the "stop hunt" because the price came back | 12 |
| Presenting the best variant of four and forgetting the drawdown and that the variants were born from the same data | 13 |
Quick reference: the rules
| Concept | Book rule | Ch. |
|---|---|---|
| Level and zone | A price with a zone of ±0.25 ATR (18.6 $ of median ATR: ±4.7 $) | 1, 2 |
| Contact | First H1 candle (market-open) that enters the zone from outside | 1 |
| Hold, break | From the contact point, the first target at 1 ATR (H1 close) within 24 candles | 1 |
| Comparison | Levels shifted by random 2-6 ATR with the same rule; for time bands, in the same band | 1, 3 |
| Swing | Pivot confirmed when the price moves away by 1.5 ATR; valid about 10 days | 2 |
| Day, week, month, Asia | High, low, close, open of the finished period (trading day 18:00-17:00 New York) | 3, 4 |
| True/false break | First close beyond the zone; true if then a close at 1 ATR beyond the zone, false if it comes back | 7 |
| Retest | First contact from the new side within 72 candles of the true break | 8 |
| Trendline | Line through two consecutive confirmed swings, with no overshoot beyond 0.25 ATR | 9 |
| Confluence | Families of different levels within 0.5 ATR of the contact | 10 |
| Trade | Entry at the close, stop at 0.5 ATR beyond the zone (bounce, retest) or 0.25 ATR inside (break), target 1 R, 24 candles, cost 0.3 $, one at a time | 11 |
| Size | Dollar risk (1% of the account) divided by the stop distance | 12 |
Quick reference: the measurements
| Ch. | Result |
|---|---|
| 1 | 1,715 contacts with time-based levels: hold 48% (± 2), break 51%, random 47-51%; round numbers 44%; resistances 47%, supports 50% |
| 2 | H1, H4, D1 swings: hold 48-49% (random 47-49%); the zone changes contacts (310-436), not the outcome; 33% of contacts do not reach the line; 2.8 levels within 1 ATR on average |
| 3 | Yesterday's levels: hold 48% for all four; high and low touched on 50% of days, close on 84%; inside days 14%, outside 11% |
| 4 | Previous week 442 contacts at 47% (random 48%); week's open touched again in 98% of weeks, close 83%, high 50%, low 43%; month: 20 levels and 35 contacts |
| 5 | Multiples of 50 $ 44% against 49% for shifted grids (46-51%); 20% of days' extremes within 5 $ of a multiple of 50, as by chance |
| 6 | Hold by touch 47%, 50%, 48%, 50%, 55%; after a hold the next touch holds 49%, after a break 51% |
| 7 | False breaks 42% (one close), 31% (two), 26% (0.5 ATR), 12% (1 ATR); random equal |
| 8 | After a true break the retest arrives in 75% of cases within 3 days (median 10 candles) and holds 50% (± 4) |
| 9 | Trendlines 47-48% (random 46-48%); third touch 42% (random 45%); channel parallels 50% |
| 10 | Confluence in 85% of contacts (random 86%); hold 50%, 48%, 47%, 46% with 0, 1, 2, 3 or more levels; level with a round number nearby 44% against 51% |
| 11 | Nine entry rules: R from −0.17 to +0.06, errors ± 0.08-0.15; random direction −0.02/−0.04; bounce at round numbers −0.17 ± 0.09 |
| 12 | Push beyond the level in holds: median 0.26 ATR; stop at 0.25, 0.5, 1 ATR hit in 81%, 76%, 67%; return after the stop 52%, 45%, 36% (random 54%, 47%, 36%) |
| 13 | Variants −0.05, −0.00, +0.03 (errors ± 0.08-0.10); the best of four exceeded by 32% of random trials; drawdown 17.1 R; about 1,500 trades for ± 0.05 |
All measurements come from ten months of a single instrument (December 2025 - October 2026, Binance gold perpetual future, prices only), with wide margins of error. None demonstrates an operating advantage.
Glossary
The book's terms are in the site glossary, where each has an example with the numbers.
| Term | Definition |
|---|---|
| Level zone | A level is a price with a margin around it, not a thin line. In the book the zone is a quarter of an ATR wide on each side. |
| Contact with a level | The first H1 candle that enters the level's zone from outside, that is with the previous close outside the zone. |
| Hold and break of a level | From the contact point, hold = the price moves away by 1 ATR to the starting side with an H1 close; break = it goes 1 ATR the other way; the first of the two counts, within 24 candles. |
| Random comparison levels | For each real level, the same level shifted by a random number of ATR (from 2 to 6, above or below) and measured with the same rule. |
| Comparison by time band | Splitting contacts by time of day (Asia, London, American morning, afternoon) and comparing the hold between bands. |
| Inside and outside day | An inside day stays within the previous day's range; an outside day exceeds both yesterday's high and low. |
| Round numbers as levels | Prices ending in 00 or 50 (4,000, 4,500, 4,550 $) that many traders use as "psychological" levels. |
| Third touch and number of touches | The idea that a level touched several times is stronger (or weaker) and that a trendline is confirmed at the third touch. |
| True and false break of a level | Break: first H1 close beyond the zone on the side opposite to the one of arrival. True: then a close at 1 ATR beyond the zone; false: a close that comes back to the starting side beyond the zone. |
| Retest and polarity | After a true break, the retest is the first contact with the zone from the new side; polarity is the idea that an old support becomes a resistance (and vice versa). |
| Trendline through two confirmed swings | A line through two consecutive swings of the same type (rising lows or falling highs), each confirmed by a 1.5 ATR move away, with no candle passing it by more than 0.25 ATR between the two points. |
| Measured confluence of levels | The number of levels of a different family (day, week, month, Asia, H4 swing, round numbers, trendline) within half an ATR of the price of a contact. |
| Push beyond the level | How far the price passes the level (in ATR) before turning, when the level holds. |
| Measured stop hunt | The idea that the price is pushed beyond a level to collect the stops and then comes back. |
What to do now
- Treat levels as references, not predictions: they serve to measure distances, to set stops and targets and to know where the price has already been.
- Use the base rates for targets (yesterday's high is touched on 50% of days, the close on 84%, the week's open on 98%) and the push beyond the level for the stop (median 0.26 ATR in holds).
- Write the list of levels and the contact rules before looking at the chart, and always compare with random levels in the same time band.
- To size, start from the stop: dollar risk divided by the stop distance.
- If you want to test a system, run a forward test with written rules: you need hundreds of trades, and years for an error of ± 0.05.
- For the basics: the book "Market structure", the book "Order blocks from A to Z", the book "FVG and liquidity", the book "SMC: the whole system", the book "Indicators from A to Z", the book "Elliott and Fibonacci", the book "Chart patterns", the book "Candlesticks" and the book "Hours, sessions and news". 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. On the chart the grey box is the level and the letters A-D are on four H1 candles of 25-27 February 2026 (UTC hours). Which letter is the contact, the first candle that enters the zone from the side of arrival?
Read the chart in words
| When | What | Price |
|---|---|---|
| 25 Feb 2026 20:00 UTC | level 5167.56 | 5,162.75 – 5,172.37 |
| 26 Feb 2026 03:00 UTC | Point A | 5,192.96 |
| 26 Feb 2026 08:00 UTC | Point B | 5,189.86 |
| 26 Feb 2026 20:00 UTC | Point C | 5,199.98 |
| 27 Feb 2026 02:00 UTC | Point D | 5,198.30 |
Times in UTC.
2. Why do we compare real levels with levels shifted by 2-6 ATR with the same rule?
3. With a ±0.25 ATR zone, 33% of contacts with yesterday's high and low do not reach the exact line. What does it mean?
4. The next day touches yesterday's high on 50% of days and the close on 84%. What does it tell you about a target?
5. The evening hold is 44% and the American morning 57% on yesterday's levels, but random levels with a contact at the same hour give 44% and 50%. What does it mean?
6. Multiples of 50 $ hold 44% (± 4), the nine grids shifted by 5-45 $ on average 49% (46-51%), and 20% of the days' highs and lows fall within 5 $ of a multiple of 50. How do you read it?
7. After a hold the next touch holds 49%, after a break 51%. What does it mean?
8. On the chart the grey box is a level and the letters A-D are on four H1 candles of 20-21 April 2026 (UTC hours). Which letter is the break candle, the first that closes beyond the zone on the side opposite to the starting one?
Read the chart in words
| When | What | Price |
|---|---|---|
| 20 Apr 2026 04:00 UTC | level 4799.00 | 4,794.93 – 4,803.07 |
| 20 Apr 2026 11:00 UTC | Point A | 4,811.88 |
| 20 Apr 2026 16:00 UTC | Point B | 4,807.89 |
| 20 Apr 2026 20:00 UTC | Point C | 4,821.72 |
| 21 Apr 2026 02:00 UTC | Point D | 4,811.35 |
Times in UTC.
9. With one close beyond the zone false breaks are 42%, with one close at 1 ATR beyond the zone 12%, and random levels give the same values. What do you take from it?
10. After a true break the price comes back to touch the level in 75% of cases within three days, and the retest holds 50%; random levels give 74% and 50%. What does it mean?
11. The "third touch" of a trendline holds 42% and random lines at first contact give 45%; deleting the line at the break drops the hold to 41% and to 44% for random ones. What do you take from it?
12. 85% of contacts have another level of a different family within half an ATR (random 86%) and the hold is 50%, 48%, 47%, 46% with 0, 1, 2, 3 or more levels. How do you read it?
13. The bounce at time-based levels gives −0.04 ± 0.08 R and random direction on the same entries −0.04 R. What does it mean?
14. A stop half an ATR beyond the zone is hit in 76% of contacts; after it is hit the price comes back by half an ATR in 45% of cases (random levels 47%). What do you take from it?
15. Four system variants give R from −0.05 to +0.03 and the best is exceeded by 32% of trials with random levels; to measure R with ± 0.05 about 1,500 trades are needed. How do you treat them?
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.