This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend
What I Built
For the past few months, my friend has been discussing investing with me, but he hasn’t managed to start. I wanted to make researching his first stocks feel less overwhelming.
PSXHelper brings Pakistan Stock Exchange company information into one dashboard: prices, financial statements, ratios, dividends, and a cash preview showing how many units a budget could buy.
The goal isn’t to choose stocks for him. It’s to give him enough knowledge to make his own first decision with less friction.
The core AI layer uses Ollama with Gemma 4 E2B or E4B, depending on the hardware running it. Gemma’s role is to turn company data, financial terms, and calculation results into explanations my friend can understand.
This is a demo, not investment advice. Its scope grew as I built it, and I may continue developing it after October.
PS: Gemma is failing and I'll have to fix it later but since its already very late. I'll enter anyway
Demo
Code
PSXHelper
A beginner-friendly Pakistan Stock Exchange research demo, built for a friend.
My friend has been discussing investing with me for months but hasn't managed to start. PSXHelper brings company information and plain-language explanations into one place, helping him research his first stocks with less friction, not choosing stocks for him.
Built for the Hacktoberfest Weekend Challenge: Build for a Friend This is a demo. Its scope grew during development, and work may continue after October.
Features
- Provider-validated ticker lookup; data availability varies by company.
- Company snapshot, prices, share counts and price/market-cap history.
- Income statements, balance sheets, cash flow and provider-reported ratios.
- Yearly dividend summaries and payout observations.
- Compact K/M/B figures, chart hover values and keyboard observation selection.
- Cash preview showing affordable whole units and remaining PKR cash.
- Local Gemma explanations with a separate panel of sourced evidence.
- Visible missing-data states, source references and coverage caveats.
Local AI is
…Data access currently uses a private, unofficial SDK. I won’t publicly release the data-fetching APIs until I obtain full permission. Access to a source does not establish redistribution rights.
How I Built It
I used OpenCode as my AI coding agent, with Python, FastAPI, pandas, and a lightweight HTML/CSS/JavaScript frontend. Financial calculations use Python Decimal; missing information stays missing rather than becoming zero.
Why Does Open Innovation Matter?
Open tooling lets me inspect, adapt, and keep developing the project beyond this weekend.
Open-weight models make the explanation layer adaptable to the hardware I have. I can run Gemma locally through Ollama, choose E2B or E4B, and control what information reaches the model without depending on a hosted inference API.
The most important part being privacy, no data is being stored, nothing is sent anywhere. This makes it perfect for small personal apps like this.
## Prize Categories
- Best Use of Gemma — the core model behind the beginner-focused explanation layer.
Top comments (0)