PRODUCT / INTERNAL BUILD

Portfolio analysis platform — Indian share-market portfolio platform

One FastAPI + Next.js system for live quotes, holdings P&L, strategies, alerts, public CMS and role-aware admin — built around Indian equity and derivatives workflows. Not a broker. Not investment advice.

FastAPIPostgreSQLSQLAlchemyAlembicNext.js App RouterTailwindWebSocketsJWT / Google OAuthIndian market APIsKite ConnectPaytm (creds stored)OpenAI (optional)
Industry
Indian equity & derivatives software
Type
Portfolio, market, strategy & SaaS admin
Role
Architecture + full-stack + realtime
Realtime
Customer quotes/book · admin ops snapshots
Stack
FastAPI · PostgreSQL · Next.js
Proof
Product UI: dashboard, P&L, Kite-ready orders, templates
Portfolio analysis platform dashboard with portfolio value, unrealized vs realized P&L, Sync from Kite and a live WebSocket status

Screenshots are from a local product session. Account emails are redacted. Rupee figures are that session’s book — not a published track record.

Product walkthrough

The same product on video — public landing through portfolio and strategy surfaces. It is a walkthrough, not a published return and not a broker pitch.

Open the dedicated demo page → · Watch on YouTube →

The problem

Indian investors already have brokers, screeners, Excel and chat groups. What they usually do not have is one system of record for the book, the live tape, the rules they said they would follow, and the operator who has to run subscriptions and support.

Constraints

Delivery stayed inside the client's stack, tenancy, compliance and media-ownership boundaries. Where a choice was forced (suite vs owned plane, BYOC vs CPaaS, local vs cloud models), the architecture section below records the trade-off rather than a marketing rewrite.

  • Holdings in one app, charts in another, P&L in a spreadsheet that is already stale
  • Decisions on delayed prices and manual MTM
  • No structured entry/exit or risk limits — only intent
  • Weak visibility of exposure, drawdown and what actually made or lost money
  • Public site, blog, legal pages and support sitting outside the product

A chatbot that “picks stocks” is the wrong product. The hard parts are identity, live quotes that match the book, strategy state you can audit, and an admin plane that can store Paytm and Google credentials without leaking them back to the UI.

Business requirements

One path for a customer: sign in, see portfolio value and P&L, watch instruments live, attach alerts, run a rule-based strategy through a backtest, and ask support without leaving the product. One path for ops: users, plans, subscriptions, market sync, blog, legal pages and chats. Public pages stay SEO-shaped so the product can explain itself.

This page does not publish invented AUM, win rates or “boosted returns.” Those are sales claims. The engineering claim is a complete surface map.

The solution

FastAPI owns REST, WebSockets, ORM models and integrations. PostgreSQL is the system of record. Next.js (App Router, Tailwind) renders public marketing, customer workspaces and admin. JWT covers email/password sessions; Google OAuth uses client id/secret stored through admin (masked on read). An Indian stock-market API client feeds instruments and quotes. Zerodha Kite Connect is the advertised broker-shaped integration for holdings, positions and orders. OpenAI is optional and sits behind FAQ matching on support chat.

Three audiences, one codebase:

  • Customers — dashboard, portfolio, market, strategies, analytics, alerts
  • Admins — users/roles, logs, subscriptions, customers, payments, plans, Paytm/Google creds, market sync, blog, legal, support, FAQs
  • Public visitors — landing, features, analytics, strategies, pricing, about, blog, support, legal pages

The Next.js UI is role-aware (admin vs customer). Backend role enforcement is designed to tighten on the current-user contract — this page does not pretend every admin route is already a fortress.

The public shell is a marketing site, not a second product: Features, Analytics, Strategies, Pricing, About, with Sign In, a separate Admin entry, and Get Started. Login is email/password plus Continue with Google, and a distinct admin login path.

Portfolio analysis platform public landing: Advanced Trading Analytics Platform for Professionals, Start Free Trial and Explore Features

How the system flows

Portfolio data flows from ingestion through analytics to investor-facing dashboards — API-first with role-based access.

flowchart LR
  Ingest[Market and portfolio ingest] --> Store[Normalized store]
  Store --> Analytics[Analytics engine]
  Analytics --> Dash[Investor dashboards]
  Dash --> Alert[Alerts and reports]
Investor data journey.

Architecture

The book and the tape are services. The LLM is not the book.

flowchart TB
  UI[Web dashboards admin investor]
  API[Backend APIs auth]
  DB[(Database analytics store)]
  Jobs[Scheduled jobs ETL]
  Ext[External market feeds]
  UI --> API
  Ext --> Jobs
  Jobs --> DB
  API --> DB
Portfolio analysis platform platform architecture.

Key components

Backend

FastAPI, SQLAlchemy, Alembic. Auth, market, portfolio, orders/trades, strategies, alerts, admin, blog, legal, support. WebSocket handlers close DB sessions promptly.

Frontend

Next.js App Router, Tailwind, JWT helper, WebSocket hook, chat widget on public and app pages. Sidebar/topbar change with role.

Realtime

Customer channel for portfolio, positions and quotes. Admin channel for subscriptions, customers, payments and a seats/MRR-style snapshot. Polling remains as a fallback.

Integrations

Indian market data client. Kite Connect as the advertised broker sync. Paytm and Google OAuth secrets stored in api_keys and masked on read. OpenAI only if configured.

What the product actually covers

Customer workspace

The signed-in shell is a left nav, not a pile of unrelated apps: Dashboard; Portfolio (Holdings, Positions, P&L Breakdown); Market (Watchlist, Instruments, Live Price Board); Trading (Place Order, Orders Book, Trades); Strategies (List, Builder, Backtesting Lab); Analytics; Alerts (including Alerts Center); Settings. A floating chat widget sits on these screens.

  • Dashboard — portfolio value, open positions, gainers/losers, margins, unrealized vs realized P&L, a positions heatmap panel, prediction insight, and Sync from Kite. This session showed WebSocket status in the chrome. Heatmap and prediction still showed connect-later placeholders — the panels exist; they were not fully live here.
  • Portfolio — holdings and positions, plus P&L breakdown: portfolio value, unrealized vs realized, daily (30 days) and monthly (12 months) charts. This session had no monthly series yet and showed Live WS disabled with a Refresh action.
  • Market — watchlist, instrument search, live price board
  • Trading — unified broker submission marked Kite-ready: portfolio, provider (Zerodha Kite), symbol, side, type, qty, limit price. Copy on the form: credentials must be configured in integrations. Orders book and trades sit beside it.
  • Strategies — list, builder, backtesting lab, plus templates (RSI Oversold Bounce, EMA Crossover, Breakout, Mean Reversion, Momentum) with when-to-use / when-not-to-use notes
  • Analytics & alerts — performance and indicator views; condition-based alerts with email / SMS / webhook as product channels
Portfolio analysis platform portfolio P&L breakdown with unrealized versus realized P&L and daily/monthly charts
Portfolio analysis platform place-order form: personal portfolio, Zerodha Kite provider, RELIANCE market buy, Kite-ready credentials note
Portfolio analysis platform strategy templates: RSI oversold bounce, EMA crossover, breakout, mean reversion and momentum

Admin and public

  • Admin — users and roles, logs, subscriptions, customers, payments, plans, market sync, docs, Paytm and Google credential screens
  • Public — landing, features, analytics, strategies, pricing, about, blog listing/detail, support, privacy / terms / cookies / security. Feature cards on the marketing site: real-time portfolio tracking, analytics, strategy builder and backtesting, alerts, risk management, live market data. The public Analytics page shows sample Sharpe / CAGR / drawdown / win-rate tiles — those are marketing fixtures, not a verified live book.
Portfolio analysis platform public features: real-time portfolio tracking, analytics, strategy builder, alerts, risk management and live market data

Quantity auto-sizing from ATR and “recommendation: Hold/Buy/Caution” are UI contracts on computed fields. They are not a promise of profit. The landing “14-day free trial” line is public CTA copy — this case study still does not claim a finished Paytm checkout.

Realtime

Two WebSocket surfaces, not one kitchen-sink socket:

  • Customer — portfolio, positions, quotes (and related book events)
  • Admin — subscriptions, customers, payments plus an MRR/seats snapshot

The client hook registers handlers for portfolio, quotes, positions, alerts and admin summary. If the socket drops, periodic polling still refreshes the screens. That is how a trading UI survives a flaky network without pretending the tape is a REST list. The dashboard session showed a connected WebSocket; the P&L screen in the same run showed Live WS disabled plus Refresh — both states are part of the product, not a contradiction hidden off the page.

Auth, payments and OAuth

  • Auth — email/password JWT; Continue with Google; separate Sign In vs Admin entries on the public nav. Google client id/secret is stored through admin and masked on read
  • Paytm — credentials stored masked in the database. Initiate / callback / status are designed to be extended. This case study does not claim a live public checkout
  • Secrets — Paytm and Google payloads live in api_keys; reads return masks. Encryption-at-rest can wrap that column; this page does not claim a HSM

Plans, subscriptions, customer accounts and payment transactions are first-class rows so ops can see seats and status without opening a spreadsheet.

Content, FAQ and support chat

The same API that serves the book also serves trust surfaces.

  • Blog — admin CRUD, categories, featured image, SEO fields, public listing and slug pages
  • Legal — privacy, terms, cookies, security as CMS pages, not hardcoded HTML only
  • FAQs — keyword matching, view/helpfulness counters, common FAQs in the widget
  • Support — floating chat on pages plus a full support route. Flow: FAQ match → wait if an admin just replied → optional OpenAI → fallback copy. Admin can reply and close chats

OpenAI is optional. If the key is absent, FAQ matching and fallback messages still work. That is the correct failure mode for a support widget.

Data model (what matters)

Identity: users, roles, permissions, API keys. Commercial: customer accounts, plans, subscriptions, payment transactions. Book: portfolios, positions, orders, trades, margin snapshots. Tape: quotes, ticks, indicators cache. Discretion: strategies, alerts, alert events. CMS: blog posts, legal pages, support chats/messages, FAQs.

Relationships are ordinary and therefore auditable: a portfolio owns positions; orders own trades; alerts own events; a chat owns messages with cascade delete; plans feed subscriptions. Alembic owns schema change. Seed data exists for local bring-up — credentials for that seed are not published here.

Engineering challenges

  • Stale book vs live tape — WebSocket push plus polling, not a single GET that looks live
  • Role split — customer must not see admin billing; admin summary is a different channel
  • Secrets in the product — Paytm and Google must be configurable without landing in git or in JSON responses
  • AI after data — predictions and support copy consume computed or FAQ context; they do not invent quantity
  • Payments honesty — storing Paytm credentials is not the same as a finished checkout

How those were solved

Market and portfolio services write to PostgreSQL; sockets publish snapshots. Admin credential screens mask on read. Support tries FAQ before the model. Payment tables exist so checkout can be completed without redesigning billing. Docker Compose plus Alembic is the documented deploy path; this page does not paste env files.

My role

Architecture + full-stack + realtime. API and schema, market/portfolio/strategy surfaces, WebSocket channels, CMS and support, admin credential storage, Next.js role-aware UI. Live client books stay off this page.

ArchitectureFastAPIPostgreSQL schemaWebSocketsNext.js dashboardsMarket dataCMS / supportAdmin SaaS

Technology stack

FastAPISQLAlchemyPostgreSQLAlembicNext.js App RouterReactTailwind CSSWebSocketsJWTGoogle OAuthIndian stock APIKite ConnectPaytm (stored creds)OpenAI (optional)

These technologies are the documented stack. This page does not add brokers, exchanges or certifications that were not part of the work.

Production considerations

  • Market data and P&L disclaimers — not a broker/depository product.
  • User portfolio isolation and encrypted credentials storage.

Engineering outcomes

No invented AUM, CAGR or “users made X%.” What this system actually established:

  • A customer path from login → live book and tape → strategies/alerts → analytics, with the signed-in nav matching the screenshots on this page
  • Kite-shaped order ticket and Sync from Kite on the dashboard — credentials still configured in integrations, not implied as a live broker session
  • An admin path for users, plans, subscriptions, market sync, CMS and support
  • Public SEO pages fed from the same CMS tables
  • FAQ-first support with optional OpenAI and human takeover
  • Masked storage for Paytm and Google OAuth, with payment initiation left extendable

Screenshots and diagrams

Portfolio analysis platform dashboard: portfolio value, unrealized vs realized P&L, Sync from Kite
Dashboard. Session book, not a published return. Heatmap and prediction panels still showed connect-later placeholders in this run.
Portfolio analysis platform P&L breakdown with daily and monthly chart slots
P&L breakdown. Live WS was disabled here; monthly series was empty.
Place order form with Zerodha Kite as provider
Place order — Kite-ready. The form itself says credentials must be configured in integrations.
Strategy template cards including RSI, EMA crossover and breakout
Strategy list with templates, builder and backtesting lab in the same nav.
Portfolio analysis platform public landing hero
Public landing. Trial CTA is marketing copy, not a claim that Paytm checkout is finished.
Public features grid
Public features grid: book, analytics, strategies, alerts, risk, live data.
Public analytics marketing section with sample Sharpe and CAGR tiles
Public analytics marketing. Sample Sharpe / CAGR / drawdown / win-rate tiles are fixtures — not this practice’s verified performance.
Public strategies section: visual builder, backtesting, automated execution
Public strategies story: builder, backtest, broker execution as advertised.

Architecture on this page is the HTML diagram above. The walkthrough is also on the Portfolio analysis platform demo page and in How AI is changing the way India invests in the share market.

Architectural insight

A wealth product is a book, a tape, a rule engine and an ops console. If the LLM owns the positions, you do not have a product you can defend. If PostgreSQL owns the book and the model only comments or answers FAQs, you do.

Disclaimer: Portfolio analysis platform is software for analysis and operations. It is not a broker, PMS, investment adviser or exchange. Nothing here is a recommendation to buy or sell securities. Vendor capabilities (Kite Connect, Paytm, Google OAuth, OpenAI) are described as advertised integrations in this build.

FAQ — buyer & architecture questions

Ten common questions about this case study, fit, and engagement.

Documented features focus on research, alerts or portfolio UX—not investment advice on this page.

Market data licensing and disclaimers are operator responsibilities; we build the software layer.

See screenshots for delivered surfaces.

Some include screening or research assistants; scope is documented per project.

Via APIs scoped in discovery—no generic promise here.

Send jurisdiction, data sources and user roles via contact.

Share the current stack (PBX, CRM, cloud), the failure or goal in one paragraph, peak call volume, carriers, and any deadline. Screenshots, a pcap, or a short Loom beat a 40-page RFP. Use the project brief or email hello@unifiedpbx.in.

Yes. Delivery is remote-first from Delhi with scheduled overlap for EU, UK, US and APAC stand-ups. Production changes use written runbooks, rollback steps and agreed maintenance windows.

Yes. Many engagements begin with a 1–2 week SIP trace review, tenant-isolation audit, or architecture assessment. If the fit is good, scope expands from evidence—not from a generic sales deck.

Founders, CTOs, telecom leads, MSPs and product teams building or fixing UCaaS, CCaaS, CRM+voice, AI voice, or vertical SaaS—not buyers who only need seats on a mass-market suite.

Building a market-data or portfolio product?

Share the book, the tape, the roles and whether checkout is in scope. The first reply is whether the gap is realtime, broker sync, strategy rules or admin SaaS.

Discuss Your Project