wanghz / black-litterman-model

Implement Black-Litterman Model using Python. BL model uses a Bayesian approach to combine the subjective views of an investor for expected returns with the market equilibrium returns of assets. Implement back-test by 10 stocks listed in the US stock market.

Black-Litterman-Model

Brief Introduction

1. Implement Black-Litterman Model using Python.
2. Use 4 different kinds of view type to evaluate Black-Litterman Model.
3. Implement back-test by stock market. 4 Plot line charts which display accumulated return using BL model vs that using eqaul weight (comparative approach) for these 4 types.
4. Data: price of 10 stocks in the US stock market during the past ten years.

Data Source: Wind

Details

1. 4 different kinds of view type:

• Market value as view:
  It uses weights of 10 stocks' market value as weights of assets allocation.
• Arbitrary views:
  It measures the result when views are given arbitrarily and inaccurately.
• Reasonable views:
  It measures the result when views are given reasonably and accurately.
• Near period return as view:
  It measures the result when stock price and return of nearest T periods are used as views.

Results

1. These 4 kinds of view type show results as follows:

• Market value as views:
  • Nearly equal performance as Equal Weight method (comparative approach).
  • Market value weight can not predict future return of stock accurately.
• Arbitrary views:
  • Nearly equal performance as Equal Weight method (comparative approach).
  • It can not make money if no strong economic knowledge and efficient views even if using a complex model like BL.
• Reasonable views:
  • BL Model is really strong when the views are efficient and accurate.
  • It performs much better when whole market goes up largely. (e.g. 4 huge growth in year 2015)
  • But, it can not resist the large drop (e.g. 3 large drop in year 2015) within a short time.
• Near period return as views:
  • Views which generated from nearest data can response efficiently and quickly to huge change within a short time.
  • It performs well when the whole market goes down (e.g. Two large drops in year 2015).