project_plan.md

Project Plan

TODOS

Project 1

1. Explore trading strategies: basic understanding, existing implementation
   

   • First focus on pair trading (see example notebook in /src)

2. Get stock data (e.g. Yahoo Finance), implement & backtest:
   

   • Migrate the example implementation to a backtest framework (BackTrader or QuantConnect)
   • For backtest, refer to Chapter 3 of Ernest Chan's book

3. Benchmark/evaluate various strategies: e.g. return volatility, risk
   

   • See Performance Measurement, Chapter 3, Ernest Chan

4. Go paper trading with IKBR

   • BackTrader API support

We will maintain detailed explaination on our work in README.md

• Pairs Trading Python Implementation Explained: underlying maths covered!

--> Project 2

Resource

--- Forums ---

• QuantNet
• QuantStart
• Wilmott

--- Articles & Blogs ---

• Algorithmic Trading Strategies: Types, Steps, Modelling Ideas and Implementation

• Ernest Chan's blog: quantitative investment and trading ideas, research, and analysis

--- Tech Stack ---

• BackTrader: backtesting framework

• bt: MIT lisence

• Zipline: algorithmic trading simulator with paper and live trading capabilities; back testing

• Interactive Brokers: trading platform (paper trading)

• QuantConnect: algorithmic trading platform

--- Data ---

• Yahoo Finance
• Alpha Vantage API

--- General finance ---

• Finance Training Courses
  • Building an Investment Thesis

Project 1: Trading Strategy Implementation & Backtest

Explore various trading strategies:

• Dual class arbitrage
  Taking advantage of the price discrepency that exists between stocks that have dual listings. (e.g. GOOG & GOOGL)

• Bollinger band strategy
  Bollinger bands are calculated by taking the 20-day simple moving average (SMA) of a stock price and then calculating the upper and lower bounds as the SMA +- two standard deviations of the actual stock price.

• Sector-based pairs trading algorithm
  Two assets within the same sector/industry will likely have similar performances and therefore any observed significant deviations in the prices of these assets can be capitalized on in the form of buying the falling asset or shorting the rising asset.

  • Example 3.6: Pair Trading of GLD and GDX, Quantitative Trading by Ernest Chan

• Machine learning based trading algorithm

Project 2: Option Pricing Model

Potential pricing models:

• Monte Carlo
  Numerous random walks for the price of an underlying asset are generated, whereby each has its own associated payoff. These payoffs are averaged and discounted to today, ultimately revealing the price of the option.

• Binomial pricing model

• Black Scholes