Backtesting my strategy with 14 years worth of historical data

I managed to rewrote a subset of my actual algorithm into Quantopian using Python. It is only a subset because I didn't manage to code the macroeconomic and sector analysis part which was written wholly in C++. Nonetheless, the performance is quite impressive even without the support of macroeconomic and sector analysis. One thing to take note is that the maximum drawdown is -40.6% which is huge. But this is the nature of the algorithm. High risk for abnormal returns...