Intelligence Objects

Arvio normalizes everything it tracks into one of seven types, so different lenses compose instead of living in separate silos:

An opportunity and the narrative driving it are separate objects, linked — so you can see both the specific pick and the broader thesis behind it.

Objects link when they share tokens or tags. A narrative drives an opportunity; an opportunity is exposed to a risk; a signal signals for an opportunity; capital rotation tracks a narrative. Each relationship carries a human-readable label, so the graph reads as sentences, not just edges.

This means an insight on one object surfaces its context: a risk object shows which opportunities it threatens, a narrative shows which tokens it's driving.

A pattern is an empirical market regularity mined from resolved prediction outcomes — not a one-off event, but a recurring rule ("volume acceleration on Solana memecoins preceded positive 7-day returns in X of Y samples"). Patterns are mined with resolved outcomes as ground truth.

Each pattern carries a conviction history across runs. When new evidence confirms it, conviction rises; when contradicted, it falls; when invalidated, it's marked faded. A pattern visibly self-improves as evidence accrues — or honestly fades when the market stops agreeing.

Every object carries a rarity read — how unusual its current conviction is against its own history. Arvio computes a z-score over the object's conviction history and bins it:

A narrative sitting at "Very Unusual" is doing something it almost never does — which is often where the alpha (or the trap) lives.

Beyond raw scores, each object is enriched with the questions a good analyst actually asks:

Signed evidence (+/−) shows what's supporting and what's undermining each object, so conviction is never a black box.