trophy

Long Only, ETF Alpha Streams Competition with $27,500 Prize Pool

Back

Documentation discussion key-concepts/research-guide

Update Backtest







0

The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by QuantConnect. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. QuantConnect makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances. All investments involve risk, including loss of principal. You should consult with an investment professional before making any investment decisions.


The automatic parameter detection is a nice idea in theory, but at the moment I don't think it works very well yet. That's because it tends to detect things linke basic array indexing offsets as relevant strategy parameters when they are clearly not.

Like in this example:

# nested loop to create combinations (i.e. pairs) of symbols
for i in range(len(self.symbols)):
for j in range(i+1,len(self.symbols)): # here, '1' is detected as a parameter that contributes to overfitting
# ...

Is there any way around this?

It would really find it useful to be able to distinguish between actual strategy parameters, which affect performance, and trivial programming patterns such as "iterator plus one".

 

 

1

Thanks for the feedback @Bert, we're constantly improving the parameter detection system and will keep your comments in mind. There are some cases where the index is an implicit parameter, such as choosing the Nth item in an array vs the N-2th item. We may lean towards only presenting the parameters once in a while vs all the time to prevent the "noise" while coding. 

1

The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by QuantConnect. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. QuantConnect makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances. All investments involve risk, including loss of principal. You should consult with an investment professional before making any investment decisions.


Fair enough. As long as you don't plan to do Alpha scoring based on the automatically detected parameter count, the occasional spurious detection is not really a problem.

1

Update Backtest





0

The material on this website is provided for informational purposes only and does not constitute an offer to sell, a solicitation to buy, or a recommendation or endorsement for any security or strategy, nor does it constitute an offer to provide investment advisory services by QuantConnect. In addition, the material offers no opinion with respect to the suitability of any security or specific investment. QuantConnect makes no guarantees as to the accuracy or completeness of the views expressed in the website. The views are subject to change, and may have become unreliable for various reasons, including changes in market conditions or economic circumstances. All investments involve risk, including loss of principal. You should consult with an investment professional before making any investment decisions.


Loading...

This discussion is closed