The purpose of this is to calculate the moving average with exponentially weighted method, a single exponential smoothing model where the moving average value depends on the window size, and the alpha ∈ (0,1]. alpha parameter defaults to 0.3. A smaller alpha value will produce more smoothing and more lag. The rest of the aggregation for the ewma model is as follows:

"weekly_moving_avg": {"moving_avg":{"buckets_path":"weekly_change","model":"ewma","settings":{"alpha":0.3},"window":4,"predict":4}}}}