Hello all,
I am comparing the results of a backtest in the Quantopian IDE to the results of a pyfolio tear sheet, and am seeing a large discrepancy in the reported information ratios. The IR in the Q IDE is 0.95, while in the backtest it is 0.02. The returns, volatility, and max drawdown are all the same, while the Sharpe ratio in the Q IDE is 1.35 as compared to 0.95 in the pyfolio tear sheet. From what I've read on the forums the tear sheet is to be trusted, however the IR seems suspect since the algorithm outperforms the benchmark almost every year over the backtest period. Am I missing something, why the large discrepancy?
Regards,
Mark