Journal & analytics

This page covers the review surface of Cortiq: trade journals, session journals, analytics, cohort comparison, and the raw conversations view. By the end you’ll know which screen answers which review question and the loop that ties them together.

What this is

Cortiq is not only about opening trades — it’s also about understanding what happened and whether the workflow is improving. The review surface is built around that understanding.

Three journal types capture different scopes:

Trade journal — one trade, with entry, management, exit, and the AI’s reasoning attached.
Session journal — one session over a broader window, with cycle-by-cycle behavior.
Run journal — one start-to-stop run, useful for comparing executions of the same configuration.

Analytics computes the metrics on top of those journals; cohorts let you compare sessions side-by-side; conversations gives you the raw AI dialogue when the rendered journal isn’t enough. Session reviews close the loop by turning a finished session’s journal into durable, instrument-scoped lessons that the next session reads automatically.

How it fits into Cortiq

flowchart LR
  T(Closed trade) --> S[(Session trades<br/>store)]
  C(AI conversation) --> S
  S --> J(Journal page)
  S --> A(Analytics page)
  S --> R(Scheduled<br/>Telegram report)

Every closed trade and every AI cycle lands in one store. The Journal, Analytics, and scheduled-report views are different lenses on the same data.

The review surface lives across five screens:

Library → Journal — trade journal and session journal.
Library → Dashboard — performance and risk overview.
Library → Session Reviews — AI reviews and the instrument-scoped lessons they produce.
Library → Cohorts — side-by-side session comparison.
Library → Conversations — raw AI conversations across sessions.

How to use it

Establish a review rhythm

Don’t open the journal only when something goes wrong. The loop that produces compounding improvement looks like this:

Make one strategy change (playbook section, data package input, or risk limit).
Run the updated logic in a controlled scope — one symbol, one session.
Review the session journal end-to-end, not only the trade list.
Check whether the metric you intended to move actually moved.
Adjust before changing anything else.

Changing two things at once turns the review into guesswork.

Read the right journal for the question

Question	Journal to read
Did this single trade follow the playbook?	Trade journal.
Did this session stay disciplined over the day?	Session journal.
Did the AI’s reasoning change after I edited the playbook?	Conversations view + session journal.
How does this run compare to last week’s?	Run journal + analytics.
Which configuration is winning across multiple sessions?	Cohorts.
What should I change for the next run on this symbol?	Session review.

Compare configurations with cohorts

Open Library → Cohorts and group sessions you want to compare. Cohorts work best when the sessions actually match on most variables — same symbol, same provider, same risk profile — and differ only on the lever you’re testing. Comparing two sessions that differ on five variables tells you nothing.

Analytics page with P/L chart, win rate, best/worst day stats visible

Close the learning loop with session reviews

Reading the journal tells you what happened. A session review turns that reading into something the next session acts on, so improvement compounds instead of living only in your head.

After a session ends, run a one-click AI review. It reads the session’s trade statistics, the AI’s per-trade reasoning, a price-path digest of what the market actually printed, and the cycles where the AI chose to do nothing. It produces two channels:

Suggestions — scored, evidence-backed proposed changes to a playbook, data package, skill, instrument profile, or preparation. Each card has Apply, Discuss, or Dismiss. Nothing changes until you approve it.
Lessons — durable, instrument-scoped observations. Cortiq injects active lessons for that symbol into the next session on the same instrument as a reference block. They tilt the AI’s conviction and flag known pitfalls; they never add rules the playbook does not authorize.

This is the loop: journal → review → lesson → next session reads the lesson automatically. The Session Reviews screen (/reviews) lists every review and shows the active lessons per instrument. The full mechanism, including the command/reference authority split, is on the Session review page.

Schedule a recurring summary

Set up a daily or weekly Telegram report under Settings → Reports. The summary is the same data the workspace shows, but lands in your phone without you opening the desktop. See Execution modes & notifications for the configuration steps.

Reference

Metrics computed by the analytics layer

Metric	What it tells you
Daily P/L	Today’s net profit or loss across the configured scope.
Weekly P/L	The current week’s net result.
Total P/L	Cumulative across all included sessions.
Win rate	Percentage of trades that closed in profit.
Best and worst day	The single days with the highest gain and loss.
Best and worst trade	Outlier trades on either side.
Average win / average loss	The expected size of each side; the ratio matters more than either number alone.
Total trade count	Activity volume; useful for spotting overtrading.

Journal scopes

Scope	Records	Useful for
Trade	One trade’s entry, management, exit, and AI reasoning.	Single-decision review.
Session	Every cycle in a session window.	Strategy-level review.
Run	One start-to-stop execution.	Comparing executions of the same configuration.

Common questions

My win rate dropped after a playbook change — should I roll back? Not yet. Open the session journal and check whether losing trades are losing for the same reason. A higher loss rate on disciplined trades means the playbook now passes on setups the old version took — that may be the desired change. A higher loss rate on improvised trades means the playbook lost structure.

Cohort comparison shows two sessions are very different — what’s it actually telling me? Cohorts surface differences; they don’t explain them. Use cohorts to find the gap, then open the session journals on both sides to read the cycles where the divergence appeared.

How long is journal data kept? For the lifetime of the local Cortiq install. Journals live in the local SQLite database; they don’t expire. Back up cortiq.db with the rest of your machine if you want offsite history.