Market Regimes

TRADEOS.tech operates with a regime-aware signal pipeline. Before any signal reaches feasibility, the system classifies the current market environment into one of five regime types and suppresses strategies that historically underperform in that environment.

The five regimes

1. Trending

Characterized by sustained directional price movement with above-average volume. Momentum signals and trend-following strategies are favored. Mean-reversion signals are suppressed.

Typical indicators: rising ADX, price above long-term moving averages, increasing directional volume, positive funding rates for the dominant direction.

2. Mean-Reverting

Price oscillates within a range; breakouts fail. Mean-reversion and range-trading strategies are active. Momentum signals are suppressed.

Typical indicators: flat or declining ADX, price bounded between support/resistance, Hurst exponent below 0.5, oscillating funding rates.

3. High-Volatility Breakout

Volatility expanding sharply, often following a compression period. Breakout strategies are active; standard momentum and mean-reversion signals are gated. Position sizing is reduced across all strategies due to elevated uncertainty.

Typical indicators: ATR expansion, volume spike, funding rate volatility, increasing VPIN.

4. Low-Volatility Compression

Volatility contracting toward multi-week lows. A broad range of signal types are permitted since directional risk is lower. Carry strategies and funding-rate arb perform well here.

Typical indicators: declining ATR, narrow bid-ask spreads, low funding rate variance, compressed Bollinger Bands.

5. Crisis / Correlation Spike

All assets moving together in a risk-off regime. Carry strategies are closed, directional exposure is minimized, and only funding rate arb is permitted for new entries. Existing positions are evaluated for protective exits.

Typical indicators: cross-asset correlation spike (crypto + equities + bonds moving together), VPIN above emergency threshold, significant negative funding rates, accelerating drawdown.

How regime detection works

The regime classifier runs continuously and updates on each new candle. It synthesizes signals from multiple indicators:

• ADX and directional movement
• Hurst exponent (mean-reversion vs. trending)
• Realized vs. implied volatility ratio
• Cross-asset correlation
• Order flow toxicity (VPIN)
• Funding rate regime

No single indicator determines the regime. The classifier produces a probability distribution across the five regime types and uses the highest-confidence assignment. When confidence is low (multiple regimes plausible), the system defaults to the most conservative applicable behavior.

Effect on signal gating

Each signal type has a list of regimes in which it is permitted. This is configured in the policy profiles under regime_overrides.

Example from the balanced profile:

regime_overrides:
  crisis_correlation_spike:
    allowed_signal_types:
      - funding_rate_arb  # Only carry allowed in crisis
  high_vol_breakout:
    allowed_signal_types:
      - momentum
      - volatility_mean_reversion
      - intraday_vol_breakout
    disabled_signal_types:
      - mean_reversion_short_term

Signals that arrive for a disabled strategy type in the current regime are dropped before they reach feasibility. This suppression is logged and visible in the dashboard's regime activity feed.

Regime transitions

When the regime classifier changes its dominant classification, the system:

1. Logs the transition with timestamp and confidence
2. Updates the allowed signal set for the new regime
3. Evaluates all open positions to check if they should be held or flagged for exit
4. Notifies the dashboard (regime badge updates in real time)

Regime transitions are recorded in the audit trail and are visible in the dashboard's signal history timeline.