zerokn0wledge

Paradex RPI explained: retail price improvement via flow segmentation

Quote223 words

RPI creates two parallel books: a standard API book for algos and an enhanced UI-only book where market makers quote tighter spreads to retail. Rate limits and an auth token separate retail from bots, producing crossed books.

Key claims
Retail profile limits: 10 orders/second and 1,000 orders/24h
RPI orders are maker-only, visible in UI but hidden from API
Retail can market buy below API ask prices
A key feature of @paradex's unique CLOB architecture and ambitious first-principles redesign of decentralized exchange infra is RPI. WTF is RPI you're wondering? Let me explain ↓ RPI a.k.a. "Retail Price Improvement" ends the unfair speed advantage bots have over retail traders by creating two parallel order books through protocol-level flow segmentation: ➤ Standard Book: API-accessible (algorithmic traders) ➤ Enhanced Book: UI-only, includes standard orders + RPI liquidity (retail only) RPI orders are maker-only, visible in UI but completely hidden from API. 1 ) Retail Profile (gets RPI liquidity): ➤ Requires special authentication token (proves you're using the UI, not API) ➤ Rate limits: ≤10 orders/second, ≤1,000 orders/24 hours 2 ) Standard Profile (default): ➤ No token OR exceeds limits = API book only What's the result? Retail users see "crossed" books (RPI bid > API ask), enabling: ➤ Market buys below API bot prices ➤ Market sells above API bot prices Market makers post tighter spreads in RPI book since they're protected from high-speed toxic flow (predatory algorithmic trading). Retail traders exit the latency arms race entirely by operating on a separate, better-priced playing field. Bottom line: Retail traders get better prices than bots. By design. If you want to dive deeper, check the @a1research__ article linked below. Want to experience this new era of DEX trading yourself? Check here ↓ app.paradex.trade/r/a1res
QUOTING @a1research__
t.co/WhOnKTYb4m

1 image on X

Open on X

What came next · since 04 Oct 2025

zerokn0wledge avatar
Working together

Research, advisory and builds for deep-tech teams.