I'm new to the sit and have been experimenting with many different ways to quant-trade. I'm very impressed with the platform and am interested in seeing how my algorithms perform against the current market.
I'm new to the sit and have been experimenting with many different ways to quant-trade. I'm very impressed with the platform and am interested in seeing how my algorithms perform against the current market.
umm...not too often...but what is long term performance like?
Though definitely impressive (assuming beta/sharpe/vol/DD all reasonable), long term performance (in my humble opinion) is a lot more important. Once algo ported to Quantopian, you can either share the algo or the tearsheet (algo is not shared, but the stats are) to get more in-depth input from the community.
Here is a link to a PDF with the backtested results since 2007....
https://www.quantopian.com/posts/backtesting-to-6-months-of-live-trading-a-tearsheet-analysis
clone the notebook - change your backtest in the get_backtest("**57a8a47382591310058651e0**") and your set
too lazy to read your code, but massive respect if you did not over fit.. i recommend doing live trading with your algo to see if it works with brand new data
sorry Steve - the attached is probably a better tear sheet to use ... has all the stats on leverage, turnover etc. that can provide a more clear perspective on the algo's viability
... and then need to go to
Quantopian video on how to evaluate algo's ... lots of detail over the 50 or so minutes, but worth it...
[https://www.youtube.com/watch?v=-VmZAlBWUko][1]
Thank you Umar and Kristen. I plan on watching the video, learning about "fit" and other important characteristics of a good algo.
Can't tell much without a true backtest but looking at your PDF I'd say that your algo might have some potential. However, I would not try and trade this live due to the low sharpe ratio and high DD. You can see that nearly all of your returns happen after 2014. There are also a lot of transactions, did you factor in commissions and day trading restrictions (if using less than 25K starting capital)? The only other thing that I can think of that could derail this is hindsight bias or overfitting, how sensitive are your returns to your parameters?