Perp markets and positioning

This section is the first application of intenthybrid AI. The agent and its privacy core are general. Here they are pointed at one domain: reading risk in perpetual futures markets. The pages that follow describe crowding, liquidations, risk reads, and the manager as that application.

This page sets up the vocabulary the rest of the documentation uses. If you already trade perps, it will be familiar. If you do not, it is enough to follow everything that comes after.

What a perp is

A perpetual future, or perp, is a contract that tracks the price of an asset without an expiry date. Traders take leveraged long or short positions, and a funding mechanism keeps the contract price tethered to the underlying.

The three numbers that matter for positioning

intenthybrid AI reads risk from positioning, and positioning is mostly visible through three public signals.

Price. Where the contract trades. On its own it tells you direction, not crowding.

Funding. A periodic payment between longs and shorts that nudges the perp price back toward the underlying. When funding is positive, longs pay shorts, which usually means the book is skewed long and the crowd is paying to hold that side. Negative funding is the mirror image.

Open interest. The total size of outstanding positions. Rising open interest into a one-sided market means more leverage is stacking on the same side, which is exactly the condition that produces a violent unwind when it breaks.

Why positioning beats price alone

Price tells you what happened. Positioning tells you how fragile the current state is. A market can grind higher on thin, balanced positioning and barely flinch, or it can sit at the same price on heavily one-sided, highly levered positioning and be one liquidation cascade away from a flush. The second state is the one that matters for risk, and you cannot see it on a price chart.

How intenthybrid AI uses these

The terminal normalizes price, funding, open interest, and liquidation data across markets, then combines them into a single crowding index per market and a view of liquidation flow. The model reads those to produce a risk read. The rest of this section walks through each of those ideas.