ML tools show mixed wins
Phoenix AI’s V4 TWAP engine reported an average slippage of $4.68 on a test of ~15 BTC trades, benchmarking DL-driven execution (x.com). At the same time, community projects are proliferating: Alter Ego published five Polymarket trading scripts including an ML model with 208 indicators, and Drip2Rip open‑sourced FLOOP Pro (Random Forest + XGBoost + LightGBM) claiming an 83%+ win rate on NQ futures ( ).
Phoenix presents itself as an institutional‑grade AI trading platform that bundles proprietary execution engines and an AlphaNet AI DEX offering on its public product pages. (phoenix.global) Phoenix's published implementation notes on dynamic TWAP optimization report simulated execution‑cost reductions of roughly 10–15% versus static TWAP benchmarks in their comparative tests. (alphanet.phoenix.global) The Phoenix project publicly promoted AlphaNet and PhoenixONE agent tooling during a product rollout cycle in October 2025, highlighting mobile beta availability and deep‑research AI agents as part of its roadmap. (telegram.me) An open‑source Polymarket copy‑trading bot repository attributed to an Alter Ego account exists on GitHub and includes multiple utility and automation scripts for Polymarket trade replication. (github.com) Polymarket operates via Polygon and is commonly analyzed with on‑chain subgraphs and APIs such as Bitquery, workflows that community trading scripts and backtests rely on for trade history and market liquidity data. (docs.bitquery.io) TradingView lists FLOOP Pro by TheRealDrip2Rip as an open‑source indicator and describes it as an ATR‑adaptive range filter optimized with an ensemble of Random Forest, XGBoost and LightGBM models trained on roughly 8,000 bars across three market regimes. (tradingview.com) The FLOOP author maintains a presence on GitHub and also distributes community products via platforms like Whop, showing a mix of public code publishing and paid community channels for the same developer identity. (github.com)
