Regime Persistence

How long has the current perp regime held, and how unusual is that historically? Returns duration_hours, historical_percentile, and typical_remaining_hours derived from perp_history.db. BTC, ETH, or SOL via ?currency=. Ed25519 signed. $0.15. 30s cache.

Disclaimer: This is a market regime classification derived from quantitative signals, not financial advice. Past regimes do not predict future outcomes. Always conduct your own analysis before making financial decisions.

What It Does

Reads perp_history.db directly — no upstream HTTP calls — to compute how long the current regime run has lasted and compare it against all historical runs of the same regime. The historical_percentile tells you where this run sits in the distribution: 90th percentile means this run has lasted longer than 90% of historical runs of the same regime.

Regime classifications are accumulating from live data (started S90, June 2026). The percentile calculations become more statistically robust as history grows — expect meaningful distributions after 30+ days, reliable after 90+ days.

Endpoint

EndpointDescriptionPriceCachePreview
/oracle/intel/regime/persistenceCurrent regime duration and historical percentile. Defaults to BTC.$0.1530s/preview ↗
/oracle/intel/regime/persistence?currency=ETHETH persistence$0.1530s
/oracle/intel/regime/persistence?currency=SOLSOL persistence$0.1530s

Response Format

json
{
  "service": "regime_persistence",
  "layer": "Oracle Intelligence",
  "currency": "BTC",
  "current_regime": "NEUTRAL_CHOPPY",
  "duration_hours": 0.5,
  "historical_percentile": 45.0,
  "typical_remaining_hours": 0.3,
  "historical_runs": 199,
  "median_duration_hours": 0.4,
  "disclaimer": "This is a market regime classification derived from quantitative signals, not financial advice...",
  "signed": true,
  "fetched_at": 1781100000
}

Preview Endpoint

Returns all fields — unsigned, up to 24h stale. Note: the 24h stale preview means the percentile and duration may reflect a previous regime entirely. Use the paid endpoint for real-time persistence data.

bash
curl https://api.myceliasignal.com/oracle/intel/regime/persistence/preview
curl "https://api.myceliasignal.com/oracle/intel/regime/persistence/preview?currency=ETH"

Interpreting the Fields

FieldInterpretation
duration_hoursHow long the current regime has been active
historical_percentile% of historical runs of this regime that ended sooner. 90th = unusually long run.
typical_remaining_hoursEstimated hours until regime change based on historical median. Not a prediction.
historical_runsSample size. Below 30: treat as indicative only. Above 100: statistically meaningful.
median_duration_hoursThe median duration of all historical runs of this regime

Agent Use Patterns

Entry timing: If historical_percentile is above 80th and the regime is favorable (e.g. BULLISH_MOMENTUM), the run is unusually long — regime change may be imminent. Consider waiting for a new regime to establish before entering.

Hold duration: Use typical_remaining_hours as a rough guide for trade duration in regime-following strategies. A NEUTRAL_CHOPPY regime with 0.3h typical remaining is not a basis for a multi-hour hold.

Regime freshness: A regime with duration_hours below 0.5 and low historical percentile is fresh — the regime just established and historically has more time to run.

Data maturity note: Regime Persistence accuracy improves with history. During the first 30 days of live data, historical_runs will be low for rare regimes (BULLISH_MOMENTUM, BEARISH_FUNDING_EXTREME). Treat percentiles with caution until historical_runs exceeds 50 for a given regime.