Each row is a state bin of the current Daily_Slope (quantile bins from very low to very high).

Each column is a forecast horizon H (days ahead).
Each pixel color is the Jensen–Shannon divergence between:
the future distribution conditioned on being in that state today
vs
the unconditional future distribution at the same horizon.
So each cell answers:
“If Bitcoin is in this normalized-growth state today, how different is the distribution of normalized growth H days in the future from baseline?”
This is a state-conditional memory map of the normalized dynamics.
This map independently supports:
✔ Scaling law is the dominant organizing principle
✔ Deviations are structured noise with short memory
✔ No persistent state-conditioned dynamics
✔ No regime breaks in normalized coordinates
✔ Institutions damp extremes but don’t alter universality class