Members: Lim Wen Bin (583), Siow Yin Sheng (800)


Alt text

Synopsis: U.S. Attorney Chuck Rhoades goes after hedge fund king Bobby "Axe" Axelrod in a battle between two powerful New York figures. IMDB (8.4/10)

  1. Billions is awesome. Go watch it.
  2. Its hard to open a brokerage account to start trading. High barriers to entry for beginners.
  3. People think they have an edge in predicting the markets but most don't. Need a systematic process to track performance.

What it does

A telegram bot that:

  • updates the user with a real-time, personalized news feed for stocks
  • displays historical stock price series
  • user can submit price predictions that are converted into trades
  • provides users with sentiment analysis on news articles
  • trains a custom sentiment lexicon per user using user predictions
  • keeps a track record over time

How we built it

  • IEX exchange api
  • python-telegram-bot
  • python data science stack: pandas, numpy
  • textblob NLP python package
  • custom implementation of a user-trained text sentiment dictionary

Challenges we ran into

  • lack of historical financial news data with accurate timestamps and entity tagging for training
  • intermittent internet

Accomplishments that we're proud of

  • incorporating multiple components into a project
  • having fun learning

What we learned

  • how to make a full functional telegram bot
  • how to implement a natural language processing pipeline

What's next for Bobby Axelrod

  • applying the user's custom model to trade algorithmically on a large number of stocks
  • decentralized investment predictions and track records on the blockchain
  • sharing trades with friends (inline telegram bot)
  • more predictive signals (times series analysis, digital signal processing)
  • include more assets (more stocks, futures, options, ETFs, forex)
Share this project: