Quantopian's community platform is shutting down. Please read this post for more information and download your code.
Back to Community
210% returns from the Mean Reversion Algorithm

Hello!

I wanted to hear some feedback on this version of the sample mean reversion algorithm.

I noticed after backtesting a year back it was beating the market, and then as I tested further back it continued to show positive returns.

It wouldn't let me test all the way back to 2002 for some reason and gave me an error.

The only thing I changed is to short the 15% best gainers and go long on only the 5% worst.

Thoughts, improvements, or suggestions?

  • Andrew
3 responses

Would you really trade an algorithm that was in a drawdown for over four years?

Here's a longer backtest. Most of the action is due to the Great Recession. So, if you can predict the next Great Recession, then you could fire up this algo and make a bundle. Otherwise, it is not so clear what the outcome would be if you were to launch the algo today.

Thanks guys!

It's good to have another pair of eyes. I felt for sure that something must have been wrong or unrealistic about the model. I definitely would not want to be a huge downtrend for four years, although it was cool to be able to work on the sample mean reversion algorithm!

Thanks,

Andrew