Capstone: One Method, Start to Finish

Before you read: The Weekly Review: Process Before Outcome

What you will learn

• How the plan, test, journal and review lessons fit together as one process.
• How to hold back sessions before testing, so a later re-test is honest.
• How one rule change is chosen, written and re-tested.
• Why small samples rarely settle anything, and what to do instead of deciding too early.

Why follow one method all the way through?

Each lesson in this module covers one step. What happens when you join them up?

Earlier lessons covered building a plan, honest backtesting, journaling, the weekly review and reviewing your own statistics. This lesson follows one fictional trader through all of them in order, with one method, from first draft to a re-test.

The point is the process, not the method. Every number below is invented, and none of it says whether this idea works in any real market.

Fictional example. John, Harbor Index futures, the prices, times, costs and results are all invented for teaching. This shows how a written plan can be built and tested. It is not a recommendation, and nothing here says how often this idea works.

John’s market moves in ticks of 0.05, worth $5 on the standard contract, so 1 point is $100. A fictional micro contract is one-tenth the size: 1 point is $10. His costs are $5 per standard contract and $1 per micro contract, round trip, plus 1 tick of slippage on market and stop orders.

Step 1: Write the plan, version 1

What has to be written down before the first replay?

John starts with the value-edge idea from the volume profile lesson: price opens below yesterday’s value area and comes back in.

He also writes down the test design before running it: which sessions, what costs, and which sessions he won’t look at yet.

Step 2: Replay 50 sessions, and hold 40 back

Why decide which sessions not to look at before you start?

John collects 90 past sessions that opened below the prior value area. Before replaying any of them, he splits them. Sessions 1–50 are for testing now (in-sample). Sessions 51–90 stay unopened for a later re-test (out-of-sample). Once he has looked at a session, it can’t be “unseen” again.

He replays sessions 1–50 bar by bar with bar replay, with costs included. Invented numbers: this shows how to count, not how the method performs.

Expectancy per trade = +$40 ÷ 24 = about +$1.67. That’s close to zero, and 24 trades is a small sample. John writes in his notes: “Not evidence of anything yet. Next step tests whether I can follow the plan live, not whether it makes money.”

Step 3: Trade it at small size

If the replay can’t prove anything, why trade it live at all?

Replays can’t test execution: hesitation, missed orders, real fills. John trades version 1 with 2 micro contracts instead of 2 standard ones, for 20 sessions. That’s a tenth of the dollar risk, with the same rules.

He also writes a pause rule before starting: “If the live record falls $60 below where it started, stop live trading and go back to replay.”

Step 4: Journal every day

What does a daily entry record when nothing happened?

He writes one entry per session, including the no-trade days. Two examples:

Day 7. Open 100.40, value area low 101.20, so 0.80 below. Two closes above at 09:30 and 10:00, bought 2 micro at 101.55. Half at the point of control, the rest stopped at entry. Followed the plan. Net +$7.00.

Day 12. Open 99.90, value area low 101.20, so 1.30 below. Two closes above by 11:00, bought at 101.40. Closed back below at 11:30. Followed the plan. Net −$12.00. Note: the open was a long way below value today.

Step 5: Review every week

What does a weekly review count besides money?

Each Friday, John reviews the week (the weekly review). He looks at process first, then results.

The pause rule wasn’t triggered (−$15.30 is well inside −$60). One rule break in 20 sessions goes on his process list.

In week 4, John sorts the live trades by how far below value the market opened. He chooses 1 point as the dividing line because it’s a round number, and he writes that down so he can test other lines later.

Step 6: Make one change, backed by evidence

If you only change one thing, which one, and how do you write it?

Eight live trades are too few to decide anything. So John returns to the replay sessions he has already tested (1–50) and splits them the same way:

Both records point the same way. He notes the weakness himself: he chose the split after seeing the live results, so it may fit these sessions by chance (overfitting). That’s why step 7 exists.

He writes version 2, with exactly one change and a dated note:

Version 2 (one change). Added to “Not trading if”: the open is more than 1 point below the value area low. Why: far-below opens lost in both the replay (8 trades) and live (3 trades). What would undo it: if the held-back sessions show far-below opens doing no worse than near ones. Not changed: entry, exits, management, size, costs.

Version 2 is the plan shown in the volume profile lesson.

Step 7: Re-test on sessions he hasn’t seen

What makes a re-test honest?

John opens sessions 51–90 for the first time and replays them once, with version 2 and the same costs. He also logs what version 1 would have done on the far-below days, for comparison. Invented numbers: this shows how to count, not how the method performs.

The direction matches the change. But 13 and 5 trades can’t separate a real difference from luck. John’s decision is written, like everything else:

Keep version 2. Stay at micro size. Collect 40 more held-back sessions before any further change. Next test: does the result hold if the line is 0.75 or 1.25 points instead of 1?

He doesn’t call the method proven, and he doesn’t increase size. The process stays the same: change one thing, then re-test on sessions he hasn’t seen.

Seven-steps diagram

Illustrative diagram constructed for teaching. Not real market data.

A horizontal timeline with seven numbered steps: written plan v1 → 50 replayed sessions (40 locked, with a padlock icon) → small live size → daily journal → weekly reviews → one change (v2) → re-test on the 40 locked sessions. A loop arrow runs from the re-test back to “collect more sessions”. A caption reads “Change one thing, then re-test on sessions you haven’t seen.”

What this does not tell you

• Whether the value-edge idea works in any real market, or how often.
• That a 1-point filter, or any filter, improves a real method. In John’s record it rests on 16 trades and 8 trades.
• How many sessions are “enough”. That depends on how different the results are and how much they vary.
• Nothing here suggests trading any market, size or plan.

Check yourself

Reflect on these; there is no score.

1. Why did John split his 90 sessions before replaying any of them? Hint: a session you’ve looked at can’t be unseen.
2. Why did he trade at micro size after a replay result close to zero? Hint: what can a replay not test?
3. What was weak about the evidence for the 1-point rule, and which step dealt with it? Hint: when was the dividing line chosen?
4. Why did he decide not to increase size after the re-test? Hint: 13 and 5 trades.
5. What would make John undo version 2? Hint: it’s written in his note.

Last reviewed November 15, 2026. Spotted a mistake? Report an error.