How It Works

How to Interpret
the RegimeForecast Dashboard

RegimeForecast combines ensemble Hidden Markov Models with multi-factor market data to produce regime classifications, probability-weighted signals, and forward-looking forecasts. Here's how to read each panel.

Markets don't move randomly — they cycle through distinct behavioral states. Trends persist, volatility clusters, and risk-off periods tend to follow recognizable patterns. RegimeForecast uses a class of statistical models called Hidden Markov Models (HMMs) to identify which state the market is currently in, how confident the model is, and where it's likely headed over the next 7 days.

You don't need to understand the math to use the output. But if you want to — it's all here.

The Math to Calculate Hidden Markov Models

Hidden Markov Model mathematical definition showing discrete-time and continuous-time stochastic processes, emission probability, and Borel set formulation

Panel 1

7-Day Regime Forecast

7-day market regime forecast showing regime probability evolution from T+1 to T+7 using HMM transition matrix projections

This chart shows the probability distribution of regimes over the next 7 trading days (T+1 to T+7). The stacked bars represent how confident the model is in each regime state.

How to Read It

  • Strong green dominance — avoid defensive positioning; the macro backdrop is constructive
  • Gradual increase in red — tighten risk controls later in the week
  • Sharp red increase — regime transition risk is elevated; volatility hedges become more important

Slightly rising red probability into later days suggests mild uncertainty creeping in — not a regime flip, but a reminder that forecast confidence decays over time.

Answers: "Is the macro backdrop likely to change soon?"
7-day market regime forecast showing regime probability evolution from T+1 to T+7 using HMM transition matrix projections

Panel 2

Asset Signals

Multi-asset regime detection signals showing probability-weighted scores for SPY, QQQ, RSP, VTI, DIA, IWM, and TLT

This table translates regime probabilities into asset-level scores. The "Score" represents the probability-weighted expected win rate — similar to Rho-T logic, applied per asset.

Score Thresholds

  • > 0.60 — Bullish edge. Consider pressing directional exposure.
  • 0.40 – 0.60 — Neutral. No strong statistical edge.
  • < 0.40 — Bearish edge. Defensive posture warranted.

When cross-asset dispersion is low (all scores clustered), the macro regime is balanced. This environment often favors carry trades, short-vol strategies, or neutral structures rather than directional bets.

Answers: "Which assets have a regime-adjusted probability edge right now?"
Multi-asset regime detection signals showing probability-weighted scores for SPY, QQQ, RSP, VTI, DIA, IWM, and TLT

Panel 3

Model Inputs

Quantitative model inputs table showing VIX level, VIX term structure, VRP, MOVE index, credit spread ratio, breadth ratio, rotation ratio, and commodity trend indicators

This is your diagnostic dashboard — the raw data driving the model's regime classification. Understanding these inputs helps you see why the model is forecasting what it is.

Key Indicators

  • VIX Level & Term Structure — Equity volatility and its forward curve shape
  • VRP (Volatility Risk Premium) — Gap between implied and realized vol; higher = options are overpriced
  • MOVE Index — Rates/bond market volatility
  • Credit Spread Ratio — Credit stress indicator; stable values = no distress
  • Breadth & Rotation Ratios — Market participation width and sector leadership rotation
  • Commodity Trends — Oil, gold, silver trend signals for macro context

Example Pattern

Falling VIX + declining VRP + stable credit spreads + weak breadth = grind-up price action with compression. This environment is typically short-vol friendly but narrow in leadership.

Answers: "Why is the model forecasting what it is?"
Quantitative model inputs table showing VIX level, VIX term structure, VRP, MOVE index, credit spread ratio, breadth ratio, rotation ratio, and commodity trend indicators

Panel 4

Rho-T Win Rate

Rho-T probability-weighted win rate gauge showing 53.4 percent for SPY — a neutral reading indicating no clear directional edge

Rho-T is the probability-weighted expected win rate. It is not the regime state — it's the expected win probability given all regime probabilities combined.

How to Think About It

Think of Rho-T as answering: "If I take a long position today, what is my probabilistic edge?"

  • > 60% — Bullish edge. Statistical support for directional exposure.
  • 40% – 60% — Neutral. No clear regime advantage.
  • < 40% — Bearish edge. Risk reduction warranted.

At 53%, the edge is modest. This favors risk-defined structures, short-vol premium harvesting, and avoiding oversized directional bets.

Answers: "Is there an actionable statistical edge right now?"
Rho-T probability-weighted win rate gauge showing 53.4 percent for SPY — a neutral reading indicating no clear directional edge

Panel 5

Simple vs Comprehensive Model

Simple vs Comprehensive Hidden Markov Model regime comparison showing 15-year equity curve with green risk-on and red risk-off regime classifications

These two charts compare a simpler regime model against the multi-factor comprehensive model. Green dots represent risk-on states; red dots represent risk-off states.

Model Agreement

  • Both models agree — High conviction regime classification
  • Models diverge — Regime uncertainty; be cautious with sizing

The comprehensive model produces more refined signals, avoids some false flips, and typically reacts faster to macro deterioration. When both models show consistent green in recent periods, it indicates robust macro stability.

Answers: "How reliable is the current regime classification?"
Simple vs Comprehensive Hidden Markov Model regime comparison showing 15-year equity curve with green risk-on and red risk-off regime classifications

Deep Dive

The Quantitative Details

Backtested performance, regime shift detection, out-of-sample validation, and multi-model divergence analysis — the evidence behind the framework.

Result 1

Comprehensive Performance Results vs SPY

VVIX strategy performance table vs SPY benchmark showing Sharpe ratio 1.08, max drawdown -21.2%, and 2022 crisis alpha of +10.6%

This table compares the VIX Strategy to the SPY benchmark across full-cycle performance metrics. While the strategy underperformed SPY slightly in total return and CAGR, it delivered materially stronger risk-adjusted results.

Key Metrics

  • Sharpe Ratio: 1.08 vs 0.84 — significantly higher risk-adjusted returns
  • Max Drawdown: -21.2% vs -33.7% — meaningfully lower peak-to-trough loss
  • 2022 Crisis Return: -8.4% vs -19.0% — generating +10.6% alpha during the bear market

This is not a pure return-maximization model. It is a risk-managed regime strategy that sacrifices some upside to materially reduce drawdowns and improve risk efficiency. The objective is smoother compounding across cycles.

VVIX strategy performance table vs SPY benchmark showing Sharpe ratio 1.08, max drawdown -21.2%, and 2022 crisis alpha of +10.6%

Result 2

Model Shift Detection (High Volatility Regime)

HMM regime shift detection showing State 3 High Volatility with 95.9% persistence probability, rhoT scores, and short volatility risk trigger recommendation

The model identifies shifts into State 3: High Volatility / Cash with high probability. Implied annual volatility of ~29% confirms a statistically unstable environment. RhoT scores remain slightly above 0.50, meaning daily win probability is marginally positive but not high confidence.

Key Insight

  • This is not a confirmed crash regime — it is a chop / whiplash environment
  • Risk-adjusted returns deteriorate in this state
  • For short-volatility strategies, this acts as an active risk trigger

Exposure should be reduced or neutralized until regime stability improves.

HMM regime shift detection showing State 3 High Volatility with 95.9% persistence probability, rhoT scores, and short volatility risk trigger recommendation

Result 3

Performance Report (2024–2026 Test Set)

HMM Ensemble out-of-sample performance report 2024-2026 showing +50% total return matching SPY benchmark with zero false bear signals

In the 2024–2026 out-of-sample test window, the HMM Ensemble matched SPY's total return (+50%) with identical volatility and Sharpe metrics.

Structural Takeaway

  • The model captured the entire 2024–2025 bull market
  • It avoided premature exits
  • It did not generate false bear signals during the rally

This confirms regime persistence detection works effectively in trending markets. The model does not overreact to noise — a critical flaw in many tactical systems.

HMM Ensemble out-of-sample performance report 2024-2026 showing +50% total return matching SPY benchmark with zero false bear signals

Result 4

Calibrated rhoT Scores (Comprehensive HMM)

Calibrated rhoT probability-weighted win rate scores for SPY, QQQ, DIA, IWM, VTI, TLT, RSP showing neutral yellow signals with LightGBM vs Comprehensive HMM comparison

The rhoT score represents the probability-weighted likelihood of a positive return tomorrow based on the historical win rate of the current regime.

Score Interpretation

  • 0.50–0.55 — Slightly positive bias, coin-flip territory
  • Above 0.60 — High conviction bullish edge
  • Below 0.40 — Bearish edge

Scores in the 0.53–0.55 range confirm a statistically coin-flip positive environment — not a strong trend. This supports neutral or risk-defined positioning rather than aggressive directional exposure.

Calibrated rhoT probability-weighted win rate scores for SPY, QQQ, DIA, IWM, VTI, TLT, RSP showing neutral yellow signals with LightGBM vs Comprehensive HMM comparison

Result 5

HMM Ensemble Model (Advanced Feature Set)

HMM Ensemble Model with advanced feature set showing rhoT scores across SPY, QQQ, DIA, IWM, VTI, TLT, RSP in Transition/Neutral regime classification

This ensemble incorporates VIX and volatility term structure, credit spreads, breadth and rotation metrics, rates volatility (MOVE), commodities and FX, and cross-asset correlations.

Observation

  • Equities slightly positive (~0.54)
  • Bonds (TLT ~0.49) show mild downside vulnerability
  • Volatility remains elevated

Despite the broader feature set, the model classifies the current regime as "Transition / Neutral" across assets. This is not a clean bull trend, nor a crash — it is a late-cycle equilibrium with elevated fragility.

HMM Ensemble Model with advanced feature set showing rhoT scores across SPY, QQQ, DIA, IWM, VTI, TLT, RSP in Transition/Neutral regime classification

Result 6

7-Day Market Regime Forecast (Across Models)

7-day market regime forecast comparing Simple HMM bull signals, Comprehensive HMM high volatility, and LightGBM bearish cash signals showing major model divergence

There can be significant divergence between models. When simple trend models remain bullish but macro-sensitive models turn defensive, it often occurs near momentum exhaustion, macro stress build-up, or regime transition points.

Model Divergence Signal

  • Simple HMM: Strong Bull probabilities (73–95%)
  • Comprehensive HMM: High Volatility regime (75–96%)
  • LightGBM: Bearish/Cash bias

This disagreement is critical. Model divergence increases forward volatility expectations and signals elevated regime transition risk.

7-day market regime forecast comparing Simple HMM bull signals, Comprehensive HMM high volatility, and LightGBM bearish cash signals showing major model divergence

Result 7

7-Day Outlook Across Models

7-day bull probability outlook across Simple HMM, Comprehensive HMM, and LightGBM models showing widening forecast divergence and elevated regime transition risk

The forward probability table reveals the widening gap between model forecasts — surface trend strength masking underlying macro deterioration.

Divergence Pattern

  • Simple HMM: Sustained bull trend (>90%)
  • Comprehensive HMM: Rising bear probabilities (0% → 25%)
  • LightGBM: Persistent low bullish probability (~6%)

Historically, major regime shifts occur when trend models remain optimistic, macro models turn defensive, and supervised ML signals extreme asymmetry. This is a high-dispersion regime — not a high-conviction bull phase.

7-day bull probability outlook across Simple HMM, Comprehensive HMM, and LightGBM models showing widening forecast divergence and elevated regime transition risk

Ready to see your regime edge?

Start your 14-day free trial and get full access to all five dashboard panels.

Start 14-Day Free Trial

No charge until day 15. Cancel anytime.