For each lookback window — 10, 30, 60, 90, 180 days — you plot the minimum, maximum and percentiles of realised volatility over history. The result narrows as the horizon lengthens, hence the cone shape, because long-window averages vary less than short ones.
Overlaying today's implied volatilities on the cone answers a better question than iv-rank alone: not just whether implied volatility is high for this name, but whether it is high relative to what this name has actually been capable of delivering at that horizon.
Example: XYZ 30-day realised volatility has ranged from 12% to 68% over five years, with a median of 22%. Today's 30-day implied is 25% — barely above median, and nowhere near the levels that would make selling obviously attractive despite an iv-rank reading of 70.
Related: realized-volatility, iv-rank, iv-percentile, implied-vs-realized