Powered by Zerodha-Kite and Gemini
If you invest directly in Indian stocks, keeping up with them can become a job of its own. stock-tracker helps with that.
It connects to your Zerodha account, follows the stocks you own and the ones you're watching, reads the latest news, checks company fundamentals, looks at disclosed activity from promoters and large investors, and puts the important things together for you.
Every morning, you get a simple Today brief instead of having to figure out where to start.
It doesn't place trades on its own. You decide what to do.
Before the market opens, stock-tracker goes through your portfolio and watchlist. For each stock, it:
It collects recent news and classifies it as positive, negative or neutral. The goal isn't to replace reading the news, but to help you quickly identify what deserves your attention.
It looks at things like valuation, growth and financial health, helping you understand whether something has materially changed.
It tracks publicly disclosed activity from promoters, insiders, institutions and other large investors to see whether there are meaningful changes in ownership or buying and selling activity.
It finds relevant peer companies and compares the stock against them, rather than looking at the company in isolation.
All of this gets condensed into a Today brief with things like:
The idea is simple: you shouldn't have to spend your morning figuring out what happened overnight.
Add the stocks you're interested in and stock-tracker keeps the information you usually have to collect from several places together.
Market prices from your broker along with valuation, growth and financial-health metrics — making it easier to spot when a stock has become unusually expensive, when growth is slowing, or when something else in the numbers deserves a closer look.
Recent news is collected automatically and classified as positive, negative or neutral. Instead of opening twenty different articles every morning, you can start with the stories that are most relevant to the companies you own.
For each stock, stock-tracker finds a set of relevant competitors so you can compare the company with its closest alternatives. The peer discovery is designed for the Indian market — it accounts for PSUs vs private companies, conglomerates, and businesses with very few direct competitors.
The app looks at publicly disclosed transactions and ownership changes involving promoters, insiders, institutions and other large investors — turned into a simpler view of whether there are signs of accumulation, distribution or potential red flags.
There isn't a secret source of investor activity. Most of the information comes from public disclosures. stock-tracker collects things like:
It then looks at three broad signals.
These signals are combined into three simple outputs:
There are signals worth investigating further.
There are signs that ownership is moving out.
There are red flags that warrant caution.
These aren't buy or sell recommendations. They are signals to help decide where you should spend your research time.
And when there isn't enough data to make a reasonable assessment, the system says NO SIGNAL instead of making something up.
The app pulls information from a number of public and connected sources.
| Information | Source |
|---|---|
| Your holdings and market prices | Zerodha Kite Connect |
| Company news | Google News, Hindu Business Line |
| Insider and large transactions | Public NSE / BSE disclosures |
| Shareholding patterns | NSE / BSE disclosures |
| Mutual-fund activity | AMFI, mfdata.in |
| Company fundamentals | yfinance, with Tickertape used where needed |
| Index levels | Zerodha Kite, with Moneycontrol used for some indices |
The data is cached so the same pages aren't repeatedly fetched throughout the day. Data collection also runs independently for different sources, so if one source changes its website or becomes unavailable, it doesn't necessarily stop the entire morning analysis.
Running a large AI model over every piece of data every day would be expensive and unnecessary. So stock-tracker uses two levels of analysis.
A smaller, cheaper model goes through the stocks in your watchlist and portfolio every day. It looks at the available information and identifies which stocks actually deserve a deeper look.
Only the shortlisted stocks are sent for a more detailed analysis using a more capable model. This keeps the daily analysis relatively inexpensive while still allowing more reasoning where it's actually useful. There is also a daily budget limit, and paid deep-analysis calls require explicit user action.
This started as a personal project because I didn't have the time to follow all the stocks I owned as closely as I wanted to.
I wanted something that could do the first pass for me every morning and tell me where I should pay attention. That's still the goal.
It's not meant to replace your judgement, and it doesn't make trades for you. It's just meant to make the research part of investing a little less painful.
Personal project for educational use. This is not investment advice. The app never places trades without your explicit action.
↓ Drag the bar at the bottom up to see how it works technically.
Next.js · React · TypeScript · Tailwind
Handles the user interface and communicates with the backend through authenticated API requests.
FastAPI · Python · SQLAlchemy
FastAPI acts as the main gateway between the frontend, database, Zerodha, AI providers and Redis.
TimescaleDB / PostgreSQL
Postgres stores users, watchlists, portfolio data, decisions and the smart-money dataset. TimescaleDB is used for time-series market data.
Celery + Redis
Long-running tasks such as the morning analysis, news collection, instrument synchronisation and smart-money ingestion run as background jobs.
Zerodha Kite Connect
Used for market data, quotes, historical candles and portfolio information. The app does not place trades automatically.
Gemini 2.5 Flash-Lite + pluggable providers
Gemini Flash-Lite handles the bulk screening. The deeper analysis layer is provider-agnostic, with Gemini as the current default and the architecture prepared for other providers.
| Layer | Technology |
|---|---|
| Backend | FastAPI, Python 3.12, SQLAlchemy 2.0, Alembic |
| Frontend | Next.js 16, React 19, TypeScript, Tailwind v4 |
| Database | TimescaleDB / PostgreSQL 16 |
| Cache & broker | Redis 7 |
| Background jobs | Celery |
| Broker | Zerodha Kite Connect |
| AI | Gemini 2.5 Flash-Lite + pluggable deep-analysis provider |
| Security | JWT, Fernet encryption, rate limiting, risk engine, kill switch |