Nadaraya-Watson Kernel Regression

Category: trend

A non-parametric smoothing filter that uses a Gaussian kernel to calculate a weighted moving average. It provides a smoother curve than a standard SMA while remaining causal (non-repainting).

Formula

NW_t = \frac{\sum_{i=0}^{N-1} w_i \cdot x_{t-i}}{\sum_{i=0}^{N-1} w_i} \text{ where } w_i = e^{-\frac{i^2}{2h^2}} \text{ and } h = \text{bandwidth}

Inputs

• Period (default: 64)
• Bandwidth (h) (default: 8.0)
• Source (default: close)

See signal primitives and every published strategy that uses Nadaraya-Watson Kernel Regression on WOBR StrategyVerse.