Curve shape

No current reading

Curve shape is derived from the Gavel curve, and no auction has settled on mainnet yet — so there is no reading to publish. It is computed and appears here automatically once the market produces the underlying data.

What it measures

Geometric classification of the yield curve: NORMAL / INVERTED / HUMPED / FLAT.

Protocol-native signals derived from the canonical curve and credit surface fits.

How it is computed

From the log-quadratic regression rate = β₀ + β₁·ln(d) + β₂·ln(d)² fitted to bin means. NORMAL when β₂ ≈ 0 and β₁ > 0; INVERTED when β₁ < 0; HUMPED when β₂ < 0 with interior maximum (vertex_days reported); FLAT when both coefficients near zero. This reads the curvature of the fitted curve; the retired classifier compared observed bin endpoints instead, which is why the two could disagree about one fit.

Available windows: snapshot. Default: snapshot. The window is how far back the metric is computed over — "90d" means 90 days of history, not a 90-day instrument.

How to read it

NORMAL is the default — lenders demand more for longer commitments. INVERTED is rare and suggests near-term volatility expectations exceeding long-term risk. HUMPED shows where peak rate sensitivity lives along the curve.

Use it programmatically

This indicator is available over the REST API and to AI agents over the Model Context Protocol. The methodology above is the whole of it — there is no proprietary adjustment layer between the inputs and the number.