As I was recently looking around at other quantitative firms, I happened across Quantpedia. In the past we’ve spoken to Quantpedia and are actually featured on their website. However, this time around I was taking a look in their screener section and came across the “F-score and Equity short-term reversals”. Being who I am I decided to put this strategy to the test.
The investment thesis here is that a company with good fundamentals that performed poorly over the past month is likely to outperform next month, and companies that performed well last month and have bad fundamentals will underperform. To capitalize on this we will long companies that did poorly and short companies that did well.
To set this up we need two lines in the editor –
Next we set up the short.
To do this we create a “position_weights” tab and if the “short” criteria is true then the system shorts that stock, else, it longs that stock.
On top of that, if you take on a short position you need to have cash in your account as well –
I went through a few different variations between 1.4 and 2.5. The required minimum is 1.4 and that provided the best return so we went with that.
Backtesting this since 2000 on a monthly rebalance, how does it do?
Over the previous 18 years this strategy had an annual return for 8.99% for a total return of 649.34%. In that time a total of 85,000 trades were made. The primary concern is the 60% drop in the first year of the strategy. Since we are including a short strategy within this backtest it very well could have put us into the realm of losing more than what was originally in our account. Thankfully our portfolio bounced back, but it appears that returns dwindled away over time.
In years two and three we absolutely crushed the market returning over 75% each year. After that we beat the market every so often – specifically winning in 2008. (The times that this strategy absolutely crushes the market are in recession times) The problem that arises with this is that it hasn’t beaten the market by any significant amount since 2011. Seven years of mediocre returns puts me on edge, but if you believe another recession is on the way this could be a good place to protect yourself.
Comparatively, the Quantopian strategy has an annual standard deviation of 20%. Thanks to the amount of cash held in our portfolio we were able to keep the standard deviation lower at 16%. We have an average holding period of just over a month, and we were able to successfully get a sharpe ratio of 1.221.
At the end of the day, we can’t perfectly replicate the Quantopian strategy as we don’t have the same systems as them, but we got as close as we could. They have over 60+ strategies that you can see for free, and the way they calculate everything is actually incredibly interesting.