Discovering market...

The baseline model. Assumes log-returns are normally distributed (GBM) and estimates volatility in real-time using an Exponentially Weighted Moving Average.

Same probability formula, but volatility estimated via GARCH(1,1) that adapts to volatility clustering. Auto-calibrates every ~60 seconds.

Replaces normal distribution with Student-t (ν=7), assigning higher probability to extreme moves.