~/openarch/work/delta-// case study

Delta

An NSE F&O trading journal for Indian derivatives traders. Local-first capture, the full Indian cost stack applied to every trade, and pattern analysis over a trader's own history - never a recommendation.

OpenArch · in-house product·fintech · derivatives trading·2026·in development

openarch.in/work/delta

//the brief

What we built.

Most journals report gross P&L and quietly flatter the trader. Delta computes brokerage, STT, exchange transaction charge, SEBI fee, stamp duty and GST on every position, so the number on screen is what actually landed. It writes to SQLite first and syncs after, because a trader with bad signal still needs to log the trade.

// Net of costs. Always.

domainNSE F&O journalling

shapemobile, local-first

statusin development

instruments216 F&O stocks

stack

react native · expotypescriptsqlitesupabasegroq · llama-3.3-70b

//what we shipped

Trade journal

Entry and exit, lots, strike, expiry, notes and tags - a contract treated as an entity, not a string pasted into a field.

Indian cost stack

Brokerage, STT, exchange transaction charge, SEBI fee, stamp duty, and GST on the taxable slice. P&L is reported net, never gross.

Watchlists

All 216 NSE F&O stocks with correct lot sizes, versioned by contract month rather than hardcoded and left to rot.

Vault

Multi-book capital tracking, so separate pools stay separate and per-book performance is readable on its own terms.

AI Insights

Pattern analysis across the trader's own history - observational only. It will tell you Thursday expiries cost you; it will not tell you what to buy.

Offline by default

Every write hits local storage first and reconciles later. Full usability with no network is the baseline, not a degraded mode.

//data model

schema.sql5 tables

table fno_instruments

symbol, expiry, strike, option_type, lot_size, contract_month

table trades

contract_id, book_id, side, lots, entry_price, exit_price, opened_at

table charge_breakups

trade_id, brokerage, stt, txn_charge, sebi_fee, stamp_duty, gst

table books

user_id, name, capital, opened_at

table sync_log

entity, local_id, remote_id, op, synced_at

// local-first write path

01capture
02sqlite
03cost engine
04queue
05sync
06reconciled

//business logic

charges.tsCost stack

// All values in paise. Never floats for money.
function charges(t) {
  const brokerage = brokerageFor(t)
  const txn       = exchangeTxnCharge(t)
  const sebi      = sebiFee(t)
  const gst       = pct(brokerage + txn + sebi, 18)  // GST on fees only
  return brokerage + stt(t) + txn + sebi + stampDuty(t) + gst
}

contract-key.tsContract identity

// Expiry + strike + type is a composite key, not a label.
type ContractKey = {
  symbol:     string
  expiry:     ContractMonth
  strike:     Decimal
  optionType: 'CE' | 'PE' | 'FUT'
}
// lot size is looked up per contract month - never inlined

//on the roadmap

Honest edges.

→Broker statement import to reconcile against manually logged trades
→Per-book drawdown and exposure limits surfaced before entry
→Widening insight coverage without crossing into advisory territory

//result

216

NSE F&O stocks, correct lot sizes

charges in every P&L line

network calls required to log a trade

Delta reports P&L net of the full Indian cost stack - six separate charges applied per trade - and stays fully usable offline, syncing to Supabase once signal returns. Its AI layer is deliberately observational: pattern analysis over the trader's own history, with trade recommendations ruled out by design.

all_work // next buildLedgerlineUnderwrite with confidence.