Hi Maitreyi,

Alisa and I took a look at this for you. It looks to me like you have some discrepencies based on 2 things, 1) date range and 2) settings for slippage and transaction costs.

When we set identical settings for date range, set_commissions and set_slippage based on your simulated results (7/15-7/18 trading range) we were able to get very close agreement between the backtest results and what we saw in your simulation (something like -27% returns).

Those numbers don't really match your comments either here or in your email to feedback though, so its possible we aren't comparing the correct code. If you still see a discrepancy in your results after double checking the traded dates and settings for commissions and slippage please send us a note via [email protected] with links to the exact results you'd like us to review.

Keep in mind that if you do not explicitly set the values for slippage and commissions in your initialize() method the backtester will use default values (see docs here) as opposed to defaulting to zero commissions and slippage.

Best regards,
Jess

P.S. One other thing I noticed, you have the capital base set to $10,000 but the bet size set to $500,000. That's going to make your results look a bit strange because our backtester will allow you to take that ~10:1 leverage, and it will show your results based on % change from the original capital amount. It might make more sense if you set the capital base to cover the exposures the algo is going to take on - that will give you more intuitive results I think.