Inside ASCEND-CX: The AI Crypto Momentum Engine
Momentum is only tradeable when funding, breadth and depth agree. The crypto engine weights all three before it will rank a rotation.
Momentum, normalised for volatility
Cross-Sectional Momentum carries 20% of the composite and is volatility-normalised across 7, 30 and 90-day lookbacks. Raw trend in digital assets is dominated by whichever asset happens to be most volatile that week, which is why the unnormalised version of this factor is close to useless.
Market Breadth (15%) then asks how many majors confirm the same direction. A move one asset wide is a story; a move most of the complex confirms is a regime.
Funding and flow are the stress gauges
Perp Funding Stress (17%) reads funding rates and open-interest skew against their trailing mean, which is where crowded positioning shows up before price does. On-Chain Flow (13%) adds exchange net-flow, stablecoin supply and holder-cohort drift.
Implied versus Realised Vol (11%) reads the term-structure slope against realised volatility, and Beta to Risk Assets (10%) keeps the engine honest about how much of a move is simply risk appetite and the dollar.
Depth decides whether a ranking is actionable
Liquidity Depth (14%) measures order-book depth and expected slippage at the intended clip size. A ranking that ignores depth produces ideas that are unfillable at the quoted price, which is a slower way of being wrong.
As with every engine in the fleet, the composite maps to a calibrated probability and a fractional-Kelly weight, and the rotation publishes with the factors that produced it.