Methodology

Each vault runs one delta-neutral basis trade: hold the asset spot and short the matching SoDEX perpetual in equal size. Price exposure nets to zero, so value barely moves when the asset does. The return is the funding rate the short collects each hour — paid by crowded longs, not by token inflation.

Funding APR is the live SoDEX mainnet funding rate annualized by the deterministic engine — the real economics of the trade. When funding is positive the short earns; when negative the short pays. The board ranks markets by funding so the trades that are paying right now lead, and labels each as “short earns” or “short pays.” Order execution runs on the SoDEX testnet sandbox.

SoSoValue daily spot-ETF flows are reduced to a regime per asset. A composite of the flow’s direction, streak length, magnitude versus the recent norm, and short-term trend yields a conviction (low / medium / high) and a position-size multiplier. A high-conviction sustained outflow escalates the vault straight to de-risk; inflows keep full size. Matched SoSoValue news supplies the grounded “why.”

Sizing, net delta, funding annualization, net asset value, and liquidation distance are computed in a pure, unit-tested core using fixed-point decimal math — never floating point for money, and never inside a language model. The neutrality simulator and the deposit preview both run on this same core, so the numbers you sign for are the numbers shown.

The one-line explanation on each vault only restates figures the engine already computed. It is constrained to a JSON schema, validated, and never produces a price prediction, a new number, or advice. If its key is unset it shows an explicit state rather than guessing.

BasisDesk is non-custodial: it only ever sees public addresses and signed intents, and signing happens entirely in your wallet. Every fund-moving action passes a pre-trade receipt restating size, fees, and worst case. This is testnet software and not financial advice; crypto markets carry risk, including total loss of capital.