Five tabs, one truth
| Tab | What it holds |
|---|---|
| Positions | Live net per instrument and product, with mark-to-market, day P&L and the exit controls beside each row. |
| Orders | Working, completed and rejected — including the rejections, which are recorded as durable rows rather than a toast that vanishes. |
| Holdings | Settled delivery, with the deliverable quantity distinguished from pledged collateral, so an exit is never sized against shares you cannot sell. |
| GTT | Every broker-side standing trigger, including the trailing stops the app manages, with the frozen/managing state of each stated plainly. |
| Journal | Recorded FIFO fill pairs with modeled differences and estimated charges. Opening inventory and complete capture remain unverified. |
The journal shows its charge model
Recorded fills are paired FIFO within broker, exchange, instrument and product — a token held under two products is two independent positions and never cross-matches — and each supported leg is estimated with the configured Indian charge model: brokerage, STT/CTT, exchange transaction charge, SEBI fee, stamp duty, GST, and the flat DP (demat debit) fee on a delivery sell, which on a small delivery exit is most of the bill and is the head most tools quietly omit.
Brokerage is charged per order, not per fill. An order that filled in six trades pays one ₹20, not six — the journal pro-rates a single order-level fee across the slices that consume it, so partial fills do not multiply that modeled fee. Estimates are not a reconciliation to your contract note.
Export the whole thing as CSV, per round-trip, with each charge head as its own column and a totals block underneath. The columns sum to the total — that is enforced, not assumed.
Trade X-ray — explore recorded-fill models
The X-ray pairs recorded fills and shows modeled positive-pair percentages, net differences and expectancy (average modeled rupees per pair) by time of day, weekday, instrument class and symbol, plus the period after a modeled loss. Opening positions and complete capture are unverified: earlier missing fills can reverse a result. These diagnostics do not establish actual account profit or personal win rates. Every bucket is N-gated — below the sample threshold it says “insufficient history” rather than presenting a number built on four trades.
- The hour buckets follow your data, so an evening MCX or CDS close lands in a real bucket instead of nowhere.
- Unpriceable segments abstain. If a round-trip is on an exchange the charge model has no rate table for, the drag renders “—” and names how many trips it could not price — never a computed-looking ₹0.00 with net equal to gross.
- Ghost fills replay a saved plan on recorded tape as a paper model. They do not establish achievable alternate execution or profits.
The session timeline underneath it
An append-only, on-device event spine records the session as one ordered stream: orders placed, modified and cancelled with the market context stamped on them, tickets you staged but never sent (hesitations), breaker trips, regime flips, analytics cards. One JSONL file per IST day, on your machine. “What was the tape doing when I did that?” is a single range query, and every analytics surface above reads from it.
