Measure the Risk, Not the Guess
The Four Horsemen suite forecasts tail risk, volatility regime shifts, and expansion probability with accuracy published per asset. Not a direction call - a risk read you can verify against the open methodology.
The Four Horsemen
Four independent classifiers, each on a distinct math principle, each with accuracy published per asset.
Tail Risk
Will the next bar move further than its recent 90th-percentile move?
Regime State
Is the next bar in a high-vol regime vs. the 50-bar median?
Regime Change
Is the next bar the first high-vol bar after a calm stretch?
Regime Emergence
Is short-horizon (10-bar) vol spiking vs. the 30-bar median?
BTC-USD daily, single best-performing model per signal. None call price direction. Full per-asset ranges & methodology →
Built for Backtesters
Our data structures are built to drop directly into frameworks like Lumibot, Backtrader, or custom Python engines. Use the Four Horsemen to size risk and filter entries - pair them with your own directional logic.
- Up to 11 years of daily historical data
- Standardized JSON optimized for Pandas DataFrames
# Size down ahead of a likely regime change (risk cue, not a direction call)
if regime_change_conf > 0.85:
reduce_position_size()
# Widen stops when tail risk is elevated
if tail_risk_conf > 0.90:
widen_stop_loss()
Comprehensive Risk Payloads
{
"message": "Hourly data for BTC-USD with 40 candles.",
"data": [
{
"ticker": "BTC-USD",
"datetime": "2026-02-10T21:00:00",
"open": 68724.52,
"high": 68949.48,
"low": 68329.99,
"close": 68582.77,
"rsi_14": 42.2705,
"macd_line": -247.6883,
"atr_14": 667.331,
"future_vol": 0.0184,
"expansion_probability": 0.62,
"tail_risk_conf": 0.18,
"high_vol_conf": 0.74,
"regime_change_conf": 0.92,
"regime_emergence_conf": 0.41,
"breakout_conf": 0.55,
"market_regime_score": 58.1
},
.....
]
}
The Four Horsemen fields ship in every response - no separate endpoint, no extra call.
Ready to Measure Risk Instead of Guessing?
Join quants who size positions and manage risk on published, per-asset model accuracy - not a black-box score. Start your free trial today.