🧠 Regime Detection Intelligence
HMM & Multi-Factor Ensemble Classification System
The Problem of Market Regimes
Markets operate in fundamentally different behavioral states. A strategy optimized for trending markets will hemorrhage capital in mean-reverting conditions. Academic research (Hamilton, 1989; Ang & Bekaert, 2002) has established that regime-switching models significantly outperform static approaches.
ChartingQueen's regime detection system addresses this by continuously classifying market state across multiple dimensions, enabling dynamic strategy allocation that adapts to the prevailing market environment in real-time.
3-State Gaussian Hidden Markov Model
Core regime classification engine
3 hidden states mapping to Bull, Transition, Bear regimes
6-dimensional observation vector: log returns, rolling volatility, VIX level, SMA distance, volume ratio, yield curve slope
Walk-forward methodology with 12-month rolling window, retrained monthly via Expectation-Maximization
Out-of-sample log-likelihood scoring, state persistence analysis, regime duration statistics
Posterior probability distribution P(State | Observations) updated daily for each market
87% correct identification of major regime transitions since 2010 (S&P 500 drawdown > 10% as bear ground truth)
5-Brain Ensemble Voting System
Multi-factor regime consensus
Gaussian HMM posterior probabilities. Highest weight due to superior performance in capturing non-linear regime dynamics and hidden state transitions.
Volatility regime classification: VIX < 15 = complacency (bullish), 15–20 = normal, 20–30 = fear (transition), > 30 = panic (bear). Based on CBOE VIX research.
Price relative to 200-day SMA. Above = bullish structural trend, below = bearish. One of the most reliable long-term trend indicators (Faber, 2007).
10Y–2Y Treasury spread. Inversion historically precedes recessions with 12–18 month lead time. Provides macro-economic regime context.
Advance/decline ratios, new highs vs new lows. Divergence from price = leading indicator of regime change. Captures internal market health.
Ensemble Consensus Protocol
Transition logic & hysteresis
Weighted Vote Aggregation
Confidence-adjusted thresholds with proportional brain weights summing to 100%
Sustained Signal Requirement
Regime transition requires 3+ consecutive days of sustained signal to prevent whipsaws
Early Warning System
Tracks voting trend direction before official regime transition — provides advance notice
Hysteresis Mechanism
Higher threshold to exit bullish regime (0.65) than to enter (0.55) — reduces false transitions
Per-Asset Regime Intelligence
Individual security regime scoring
In addition to market-wide HMM classification, ChartingQueen computes per-asset technical regimes using a 6-factor scoring model. This enables differentiated positioning — for example, TSLA may be in a bear regime while QQQ overall remains bullish.
Trend Alignment
30%
Momentum RSI
20%
MACD Direction
15%
Volatility State
15%
VIX Context
10%
Volume Confirmation
10%