Composite read on the current Hyperliquid market regime. Aggregates breadth, volatility, and funding across the top-100 HL universe to classify the state as Risk-on, Mixed, or Risk-off.
Higher = more assets up over the period
Mean cohort vol percentile; higher = more volatile environment
Positive = longs crowded across cohort
The regime classification combines three signals across the top-100 HL perpetuals on the 4h timeframe:
The classification logic:
Risk-on: breadth > 60% AND aggFunding < 0 bps
Risk-off: breadth < 40% AND aggFunding > 5 bps
Mixed: anything in betweenThe boundaries are intentionally conservative — most days are “Mixed.” The pure Risk-on / Risk-off labels fire only when both breadth and positioning agree, which is the configuration most predictive of regime persistence.
Not financial advice. The regime label is a signal among many. Backtest any strategy across regimes before assuming a single regime is best for your edge — Keel exposes a regime tag in every backtest run for exactly this purpose.
A composite of three cohort-wide signals from the top-100 HL universe: (1) breadth — what % of assets are positive on the period, (2) aggregate volatility — mean of cohort NATR percentiles, (3) aggregate funding — mean funding rate across the cohort in basis points. Combined, they characterize whether the market is broadly bid (Risk-on), broadly stressed (Risk-off), or somewhere in between (Mixed).
Risk-on: breadth >60% AND aggregate funding <0bps — most assets up, longs not crowded, traders willing to take risk. Risk-off: breadth <40% AND aggregate funding >5bps — most assets down with longs paying heavy funding, signaling stress and crowded positioning. Mixed: anything in between — the most common state, indicates indecision.
Refreshes every page request (server-rendered). The underlying data refreshes hourly on the 4h-timeframe snapshot. So the regime label shifts only when the underlying composite shifts, typically once per hour at the boundaries.
Not financial advice, but a common framework: in Risk-on, trend-following and breakout strategies tend to work better; in Risk-off, mean-reversion and short-bias setups can outperform; Mixed regimes are the hardest — most strategies underperform. Backtest your strategy across regimes (Keel exposes a regime tag in every backtest run) before assuming a single regime is "best" for your edge.
The current readout uses the live snapshot; producing a 90-day history requires a historical-snapshot data path that lab-api doesn't yet expose. The chart is a Sprint 4 candidate — see projects/seo-tools/sprints/sprint-3.md §6 deferred enhancements for the trigger.
Top-30 HL perps by current funding rate, annualized. Drives the aggregate-funding input to this regime indicator.
Most-active HL perps by relative volume right now. Complements regime read with flow data.
Build a custom cross-sectional screen by momentum, funding, volume, breakout, regime.