When I first started backtesting Mohanram G-Score, I decided to skip the first factor, return on assets. If you recall from my overview post on Mohanram’s G-Score, G1 is defined to equal 1 if a firm’s return on assets (ROA) is greater than the industry median and 0 otherwise. Basically, this G1 factor looks at whether a company is more profitable than the majority of its industry peers. You’d expect companies with higher than average returns on assets to outperform the market.
For this backtest, I’m going to break G1 into five quintiles based on each company’s rank by industry for ROA. The 20% of companies ranked lowest in their industry by ROA will be in the first quintile and the companies ranked in the top 20% in their industry by ROA will be in the fifth quintile.
Here are the backtest rules I ran on Portfolio123 to test ROA ranked by industry on stock performance:
The backtest results are in the table below.
Summary of Results for G1: ROA Ranked by Industry
* Average excess returns were analyzed starting in each month with 12-month holding periods (230 sample periods). This avoids the potential for seasonal reporting bias.
The above backtest table and average annual excess returns chart indicate that ROA does seem to correlate with 1-year stock returns. Annualized returns for each quintile seems to increase in a linear fashion. However, average excess returns increase from the first to fourth quintile, but then the top quintile underperforms. That might be an indicate that ROA exhibits diminishing returns at the high end. Maybe those high returns attract competition at the extreme. Overall, I was a bit surprised that ROA ranked by industry did not have higher percentage 1-year out-performances in the higher quintiles. The Sortino ratio does increase steadily from 0.43 in the lowest quintile to 0.67 in the highest quintile.
Mohanram G-Score: G1 ROA Ranked by Industry Returns (2002 – 2020)
It is a bit curious that the first quintile outperforms the S&P 500 equal weight benchmark in the 2002 to 2007 time period. I wonder if that is more of a data quality issue (missing ROA values ending up in the first quintile) or a strange market phenomenon at that time. What are your thoughts on this backtest? Please share them in the comments section below.