DeepAR How can context length capture seasonality?

[aihiangng]

It is often advised to set context length = prediction length, as shown on the documentation here: https://docs.aws.amazon.com/sagemaker/latest/dg/deepar_how-it-works.html

However, if I am trying to predict for 1 week ahead, and my data contains yearly seasonality, how can I set my context length to only 1 week? Im struggling to understand why we encourage context length = prediction length.

Thanks!