Nadaraya-Watson: Multi-Filter

The "Nadaraya" indicator models a curve fitted to the bars using the Rational Quadratic Kernel function - based on the script pl.tradingview.com/v/AWNvbPRM/ with additional filters that help plot the trend directly on the price chart.
The following filters are used:
- ALMA curve logic to smooth the Watson Nadaraya regression curve -Additionally, ALMA has a "volume-weighted" option, which may be important when there is little data or small price fluctuations - it helps stabilize the bar price
- ATR logic to smooth local data based on the assumed window and multiplier
- Local data deviation (fluctuations within the local window) logic to smooth the Watson nadaraya regression curve

The basic data is optimized for BTC on a 1D (daily) timeframe to demonstrate the indicator's capabilities.

Due to the relatively complex process of optimizing parameters for any timeframe, it is recommended to start with ATR and %. After optimization for a given interval, the indicator is very precise, although it is recommended to use it for very liquid assets with a large amount of data (sampling) - this is aimed at creating a smooth curve with an accurate indication of the change in the trend direction.