Discussing the article: "Statistical Arbitrage Through Cointegrated Stocks (Part 6): Scoring System"

 

Check out the new article: Statistical Arbitrage Through Cointegrated Stocks (Part 6): Scoring System.

In this article, we propose a scoring system for mean-reversion strategies based on statistical arbitrage of cointegrated stocks. The article suggests criteria that go from liquidity and transaction costs to the number of cointegration ranks and time to mean-reversion, while taking into account the strategic criteria of data frequency (timeframe) and the lookback period for cointegration tests, which are evaluated before the score ranking properly. The files required for the reproduction of the backtest are provided, and their results are commented on as well.

With the screening workflow in mind, we are now ready to define a scoring system and backtest it. The previous article was left with an outline of the possible criteria for it. 

  • Strength of Cointegration 
  • Number of Cointegrating Vectors (Rank) 
  • Stability of portfolio weights 
  • Reasonable spreads 
  • Time to reversion 
  • Liquidity 
  • Transaction costs 

Besides these criteria, we may add two other not mentioned in that outline:

  • Timeframe (Data frequency)
  • Lookback window


    Author: Jocimar Lopes