We build the journal in-house, which means it changes every month and you get to read the changelog. It also means we owe you the reason, not just the release note. Every item below started as a coaching problem: something a mentor kept saying out loud on calls, or a number a student could see but could not act on. If we cannot name that problem, we do not build the thing.
That test is stricter than it sounds. A journal can grow features forever, and most do: more charts, more metrics, more badges. Ours has one job, which is to make you consistent, and anything that does not change what you write or what you decide on Sunday is weight rather than value.
What shipped in July
| What shipped | What it does | Who it's for |
|---|---|---|
| R-distribution chart | Plots the shape of your demo outcomes instead of collapsing them to an average | Anyone past a fortnight of logged entries |
| Week-score breakdown | Tap any 0–100 week score to see which entries moved it, and which way | Students who saw the score but not the cause |
| Cohort feed | Weekly anonymised winning setups — chart, entry, quality rating. Never P&L, never names | Every stage, heaviest use in Stage 1 |
| Faster mentor marking | The queue runs roughly twice as quickly, so feedback lands nearer the entry | Everyone waiting on a mentor mark |
| MT5 import fix | Swap and commission handled correctly on partial closes | Students importing demo history from MT5 |
| Streak fix | Journalling streaks no longer break on market-holiday weekdays | Anyone about to lose a streak unfairly |
The three that change how you review
R-distribution first, because it changes a habit rather than a screen. An average tells you almost nothing about a month of demo entries: the same average comes from twenty small, plan-following entries and from a fortnight of noise rescued by one outsized result. Those are opposite months. The distribution shows the shape — where the cluster sits, how long the left tail runs, whether the entries that worked are the size your plan says they should be. It is the chart a mentor would sketch for you on a call.
Second, the week-score breakdown, which is an admission with a feature attached. We shipped a 0–100 weekly score that gave you a number and would not tell you why — the exact thing we teach students never to accept from their own charts. Now it is tappable: every entry that moved it, and which way. A score you cannot decompose is a mood ring. A score you can decompose is a to-do list.
Third, the cohort feed. The obvious version shows the numbers, because numbers are what people screenshot. We built the other one deliberately: chart, entry and quality rating, anonymised, no P&L and no names. A feed with money on it turns a study group into a scoreboard, and the students who need it most stop posting the week they have a setback. What you should take from your cohort is a decision, not a figure.
The three that were simply wrong
Faster mentor marking is not really a feature, it is us fixing latency we had quietly learned to live with. Feedback that arrives days after an entry is history. Feedback that arrives while you still remember why you clicked is coaching. The queue now runs roughly twice as fast, which is a floor rather than a finish line.
The MT5 import was mishandling swap and commission on partial closes, which distorted the R values in every affected import — including, uncomfortably, the distributions the new chart draws. An analytics page is only ever as honest as its import. The streak bug was smaller in size and worse in spirit: we had built something that punished a student for not journalling on a day the market was shut, rewarding attendance over judgement in a tool meant to teach judgement. Streaks now skip market holidays.
What this changes about your Sunday
The weekly review, after this month's changes
Open the distribution before the calendar
Start with the shape of the month, not the sequence of days. One question: is the cluster where your plan says it should be? A 1:1 strategy that keeps producing outcomes at a fraction of its target is not a psychology problem yet. It is an early-close problem — you are cutting entries before the target, and the shape says so before you read a single entry.
Take the two entries furthest from the cluster
Not your best and worst weeks — the two entries furthest from where everything else lives, because outliers are where process breaks in both directions. Read your written reason on each. If an entry that worked has no plan reference, it is a red process entry whatever the number says.
Tap the week score and read the movers
Now see which entries actually moved it. Half the time the number you were unhappy about was driven by two rows you had already forgotten. That is the point of the breakdown: it replaces a feeling about the week with a list of decisions you can name.
Write one change, not five
Finish with a single sentence naming one thing you will do differently next week, and put it where next week's review will read it back to you. One you actually run beats five you abandon by Wednesday.
What we changed our mind on
Two things, honestly. The first is the average. We led with it because it is the number every journal leads with, and it flattered quiet months and hid loud ones equally well. Replacing a headline metric is expensive and we put it off too long. If a number cannot be acted on, promoting it to the top of the page is a design decision, and it was the wrong one.
The second is scoring at all. The honest case against a weekly score is strong: any single number invites you to optimise for the number instead of the behaviour behind it. We made it decomposable rather than deleting it, and we treat that as a decision under review, not a settled one. If the score starts driving behaviour more than the entries do, it goes.
- Do I need to do anything to get these?
- No — the journal is part of the programme at every stage and updates arrive with it. One exception: if you imported MT5 history before this month, ask a mentor whether your affected imports need re-running so the corrected swap and commission handling applies.
- Is any of this graded automatically?
- No. Your entries are marked by a mentor, which is why the fix was making the queue faster rather than replacing the person in it. The analytics draw what you logged; the judgement about whether a decision was sound stays human.
- Why publish a changelog at all?
- Because you are trusting your trading education to software we write, and the reasonable response to that is scepticism. A monthly public list of what changed, what broke and what we reversed is how you hold us to it.
The reason we publish this is not transparency theatre. A journal is a promise about attention — yours to your trading, ours to your journal — and a promise you cannot audit is worth little. Next month's list is already forming, and the fastest way onto it is to tell a mentor which screen is not earning its place.
The Traderess coaching team
Written the way we coach — direct, structured, no shortcuts.