Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
sortino 5.34

I was wondering what a "good" range was for sortino ratio. I ran an algorithm for past two years and got 5.34 is this decent?

1 response

It is probably too good to be true.

When you compute downside deviation, you begin by summing (r_i - r_0)^2 for all r_i < r_0, where r_i are your daily returns and r_0 is the risk-free rate. Did you then divide by the total number of days, or the total number of days when r_i < r_0? In the former case, you will get Sortino ratios about 40% larger on average than the latter case. People do it both ways.

Even with the 40% boost, 5.34 is not a credible ex ante Sortino ratio for an unlevered low or moderate frequency equity strategy. My guesses are it's a calculation error, or it was data-mined, or it uses unrealistic execution assumptions.

It's easy to find strategies with ex post 5.34 and higher Sortino ratios over two years. Individual stocks often have them.

To answer your question, a solid long-term backtest of an unlevered, low frequency equity strategy showing a Sortino ratio above one, after adjusting for data mining, and based on solid economic fundamentals, would be considered very good. A two year Sortino of any number, especially without adjustment or economics, would not be impressive.