Composite (0β100): Weighted average of all 6 components below. STRONG β₯70, MODERATE 50β69, WEAK 30β49, INEFFECTIVE <30.
Accuracy: How often the Markov direction prediction matched the actual next-day move.
Entropy: Low entropy = transitions are concentrated (predictable). High entropy = random-looking transitions (noisy).
Stability: How consistent the transition probabilities are over time β stable patterns are more trustworthy.
Return Sig: Statistical significance of the return differential between predicted states (t-statistic based).
Sample: Score reflecting depth of historical data β more transitions = more reliable statistics.