Probability of a sharp 24-hour crypto move. Higher score = higher probability.
Realized vol moderate, OI calm, liquidation pulse elevated, funding mixed, options pricing in moderate stress.
Bi-directional: a high reading means a sharp move is more likely — not which way.
What COIL Index measures
Crypto markets don't simply move — they coil, then strike. Leverage stacks invisibly, volatility compresses, until something forces the unwind. Coil Index measures the build-up — the loaded spring of leverage, volatility, liquidations, funding pressure, and options market stress that precedes sharp moves in either direction.
A high reading means conditions are coiled: leverage is crowded, volatility is shifting, order books are setting up for a sharp move. A low reading means the market is genuinely calm — leverage is light, vol is low, dislocations are unlikely.
COIL is direction-neutral. A high reading means a sharp move is more likely. It does not say which way. Use COIL to size risk, not to decide long vs short.
COIL Index — Daily Peak
Daily peak hourly score. Bars colored by tier — anything above 70 is the alert zone. Hover an alert bar to see the BTC and ETH 24-hour move that followed. The Coil Index measures fragility, not direction — high readings precede sharp moves either way.
How it's built
Coil Index v1.0-EXT is a two-stage model. Each stage runs every hour automatically.
Coil Index measures fragility through five direct signals. Realized volatility — the market's actual movement intensity, the most documented predictor in financial econometrics since Engle's 1982 ARCH work. Open interest velocity — leverage building across major venues. Liquidation pulse — cascade fuel measured at the source. Funding-price divergence — when leveraged positioning runs counter to price action, signaling crowded trades approaching forced unwind in either direction. Options market stress — implied volatility and put-skew premiums pricing in tail risk before it surfaces in spot. Each component is independently meaningful.
The base reading isn't read in isolation. A second layer learns from six years of historical patterns — how a 70 reading behaves differently before a CPI release vs. on a quiet weekend, how weekend liquidity dynamics amplify fragility, how BTC-ETH correlation breaks signal regime shifts. Sixteen contextual features feed into the refinement, producing the final score.
The final score is mapped to six tiers from Quiet to Extreme. Methodology is versioned, tested against pre-committed predictions, and shipped only when it clears gates locked in writing before the backtest runs.
COIL Index — Calibration
11,154 hourly out-of-sample predictions · 16-month HOLDOUT (2025–2026)
Daily risk gauge — leveraged positions begin facing stress
| Score | Hours flagged | % of all hours | Hit rate | 95% CI | Lift vs base |
|---|---|---|---|---|---|
| 70+ | 1,513 | 13.56% | 85.1% | 83.2–86.8 | 1.82× |
| 75+ | 1,052 | 9.43% | 91.9% | 90.1–93.4 | 1.97× |
| 80+ | 693 | 6.21% | 95.4% | 93.6–96.7 | 2.04× |
| 85+ | 415 | 3.72% | 95.7% | 93.2–97.2 | 2.05× |
| 90+ | 196 | 1.76% | 98.5% | 95.6–99.5 | 2.11× |
| 95+ | 133 | 1.19% | 100.0% | 97.2–100.0 | 2.14× |
Wilson 95% confidence intervals · Out-of-sample HOLDOUT (Jan 2025 → Apr 2026), no look-ahead · v1.0-EXT model trained on data through Dec 2024 · Same model evaluated against 3%, 4%, and 5% move thresholds — same score, three operational yardsticks