Phoenix AI posts TWAP benchmarks
Phoenix AI published V4 TWAP execution benchmarks on AlphaNet showing average slippage of $4.68 on ~15 BTC trades over 30 seconds using tick‑level data—evidence that specialized, tick‑level execution engines can materially cut slippage in short‑horizon trades. The numbers offer a concrete data point for builders of execution stacks. (x.com)
AlphaNet’s public litepaper says its real‑time data pipeline processes spot trades at tick level and futures at 100ms granularity, framing short‑horizon execution on true tick‑level feeds rather than minute bars. (phoenix.global: ) The AlphaNet user guide lists “Dynamic TWAP” as a live algo that adapts execution policy using orderbook, price and volume inputs in real time. (alphanet-userguide.phoenix.global: ) A technical whitepaper titled “Dynamic TWAP Optimization” dated October 28, 2025 compares reinforcement‑learning and Almgren–Chriss approaches, reporting empirical simulation results that show roughly 10–15% reductions in expected execution cost versus static benchmarks in their tests. (alphanet.phoenix.global: ) Phoenix has publicly framed AlphaNet’s execution stack as an “order execution edge,” earlier claiming saved execution costs of roughly 5 bps on large‑cap and up to 30–40 bps on mid/small caps in marketing posts around its AlphaNet launch. (binance.com: ) The AlphaNet Minima/Maxima model tutorial explicitly states it derives signals from exchange tick‑level trade and limit‑order‑book features, confirming the platform’s execution and insight models operate on high‑frequency orderflow. (alphanet-doc.phoenix.global: ) Phoenix has been rolling AlphaNet into partner integrations — Bella Protocol announced a strategic partnership to use AlphaNet’s AI trading capabilities and WOO X similarly announced collaboration to expose AlphaNet signals to its liquidity network. (prnewswire.com: ) (woox.io: )
