Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
Another scoring bug? Different risk-free rates for each algo

Echoing Vladimir's point from earlier today, the Sharpe calculations for the June leaderboard use different risk-free rates for different algos, and some of those rates are even negative. This doesn't seem right. Q team, could you take a look?

2 responses

It could be related to my post yesterday

Yes, I think a change in the chosen risk-free rate could explain the change in your Sharpe.

After Friday's trading, your paper return was 18.64%, volatility 6.12%, Sharpe 2.40, implying a risk-free rate of 3.99%. [Following Vladimir: rate = return - (Sharpe * vol)]

After today's midday update, your return was 9.80%, volatility 5.97%, Sharpe 2.56, implying a risk-free rate of -5.50%. The negative rate is how your Sharpe could go up from 2.40 to 2.56 even though return went down and volatility stayed almost the same.