Course Name: Trading with Machine Learning: Regression, Section No: 8, Unit No: 1, Unit type: Video
In the entire course, no place it has explained a python code for calculating R (co efficient of determination) for multi variate regression… is it possbile to get python code for this ?
Besides - can you please let me know what type of data would this strategy work on ?
I believe- If we have low R, that means that IVs are not influencing the DV substantially… correct ?
Is there any further guidance /notes on what type of data should be select and for what time frame to make this startegy more effective
Hi Ankit,
For the coefficient of determination, you may use this resource.
If R^2 is low, that means the line is not a good fit for the data and what you said about IV and DV hold true.
Any regression model is as good as the features you select. If you select good features, then the prediction would be fairly accurate. You should backtest with different combinations of features and select the ones best suited for your underlying on the basis of your backtest results.
This strategy involving regression is susceptible to regime changes, and if that happens you need to recalibrate your regression model. The performance would greatly vary in different regimes and you will need to do extensive backtests to identify, retrain and analyse the performance to finally narrow down your stock universe.