Candl3.tech

• 

  Sometimes the trend is up...

• 

  Sometimes the trend is down! Can you spot the difference?

Inspiration

We are a team of young quant developers, heavily inspired by skills based platforms like LeetCode and ZetaMac.

We realized that no existing platform takes this rapid learning approach and applies it to building quant trading skills. Welcome Candl3.tech !!

What it does

Candl3.tech is a skills development platform that helps young traders quickly practice sighting charts. We query 30 years of real stock data from the SP500 and ask users to predict whether a frame should be held long or short. After the user makes a call, we reveal the next week of data and update their bankroll.

Users are given 50 candles of past data (hence the name) and have to enter a hold position for the next 5 candles. As users train on our platform, they start to recognize patterns for good trade setups. Can you tell when the market is going up or down?

The humiliation of sending the bankroll into the negatives is a stark contrast to the rush of watching your money 10x over a handful of trades. The overall user-experience is captivating to say the least. We designed the website so that users can easily place 20-30 trades per minute. Spend a few hours with us and you'll be a pro!

How we built it

We are running a Next/React front-end with a Flask/MongoDB Atlas back-end.

The database was created by scraping Yahoo Finance into a dataframe on a Jupyter Notebook. After some gentle data engineering, the info for the candles is processed and sent to MongoDB atlas in the form of JSON documents. We designated one document per candle, and each "level" queries about 50 documents. This JSON data is then served on our flask backend endpoint.

We built a special React element using ApexChart.js to display the candles on a dynamic chart. The React element then calls a query to the back-end to populate the chart with data from the json. We manage the user's bankroll in a shared state that our React components have access to. As each round progresses, we update and display the bankroll as appropriate.

Challenges we ran into

• Designing a system that is simple and elegant
• Setting up SSH protocol on guest wifi
• Standardizing production environment (thank you mentors!)
• Connecting to MongoDB atlas server
• Loading data into the database (very time intensive)
• Running both Next.Js and Flask concurrently
• Styling our front-end components
• Adding JS logic and shared states
• Hosting our project on Vercel

Accomplishments that we're proud of

• Solid product vision
• Clear value proposition
• Working MVP in 24 hours
• Excellent user experience

What we learned

• Streamlining a tech stack
• Using Docker to manage environments
• Deploying a cloud database using MongoDB Atlas
• Building React Components using Apex Chart
• Setting endpoints using Flask
• Managing revision history in GitHub

What's next for Candl3.tech

Phase 1 - MVP (Completed)

• Single user interface
• Data from SP500
• Grading system

Phase 2 - User Connectivity

• Add Google/Github OAuth
• Add collection for users in MongoDB
• Track and display user statistics over time
• Add FAQ/Getting started page
• Add accessibly features like high-contrast and Google Translate

Phase 3 - Network Effect

• Give users a public homepage
• Allow users to compete in events
• Create a global leaderboard

Phase 4 - Advanced Features

• Develop ML pattern matching for common trading strategies
• Help users learn from good/bad trades
• Gather data from a higher frequency
• Allow users to train on other assets
• Add indicators for fed moves and breaking news

Built With

• flask
• javascript
• mongodb
• next
• python
• react
• tailwind