Hi Drt, really sorry about the delay, to be honest, I didn't see the comment notification previously.
The ensemble mean is mostly a diagnostic tool. The idea is to check whether the models make their mistakes at the same time or not. The rolling window simply smooths the noise so you can see the underlying pattern more clearly.If the average stays stable while the individual curves move around, it means the errors are not strongly correlated.
On its own this metric does not do much. It simply tells you that if you use these signals inside a strategy, applying the model across multiple assets could deliver a more attractive risk-return profile than relying on a single asset.
Thank you very much Lucas.
I was wondering what would be the practical use of the ensemble mean of rolling directional accuracy. How would you use it in to improve a model?
Hi Drt, really sorry about the delay, to be honest, I didn't see the comment notification previously.
The ensemble mean is mostly a diagnostic tool. The idea is to check whether the models make their mistakes at the same time or not. The rolling window simply smooths the noise so you can see the underlying pattern more clearly.If the average stays stable while the individual curves move around, it means the errors are not strongly correlated.
On its own this metric does not do much. It simply tells you that if you use these signals inside a strategy, applying the model across multiple assets could deliver a more attractive risk-return profile than relying on a single asset.