# Abstract

reinforce.fi is building xUSD, an Arbitrum-native yield-bearing onchain dollar designed to help users earn variable yield on idle USDT without trading complexity or opaque offchain risk. The product is aimed first at the large base of practical stablecoin users on TRON and similar networks who want to keep control, stay onchain, and access better utility than basic lending. Under the hood, xUSD deploys capital into fully onchain delta-neutral spot-plus-perpetual strategies across the Arbitrum ecosystem, harvesting funding-rate and basis-like opportunities while minimizing directional market exposure. 

Reinforcement learning is used as an adaptive execution and risk-management layer that improves allocation, timing, venue selection, hedge management, and transaction efficiency under changing market conditions. Research and testing on the core strategy stack have produced approximately 10% APR with a 4.4 Sharpe ratio, but these figures should be understood as R\&D results rather than guaranteed live outcomes. This paper describes the product thesis, mechanism, architecture, risk framework, trust model, ecosystem design, and launch path for xUSD.

Implementation note: where launch parameters remain under finalization—such as bridge providers, the exact yield-accrual format, fee schedule, or any project-level token rights—this paper states the intended framework rather than fixed numeric terms.