开源 Jev 交易 bot：新闻喂策略，代码管下单
有人用 Jev 把 4.9 万条新闻分类成资产、情绪、重要性和监管风险，再让算法调整入场条件，还公开了分层回测，结果并未稳定跑赢无 AI 版本。
this open-source Jev trading bot is wild. the builder processed 49,496 news texts for around $2.50. not to ask a chatbot “should i buy BTC?” to plug news directly into a trading strategy. here’s what’s inside: → BTC, ETH and SOL tracked on 4h candles → news from crypto media, the Fed, SEC, Trump and Binance → Jev classifies the asset, sentiment, materiality and regulatory risk → the algorithm adjusts its entry conditions or blocks new buys → a dashboard shows the paper portfolio and historical replay Jev reads the headlines. code controls the trades. you can start with a simulated 1,000 USDC portfolio without connecting an exchange account. a TypeSafe API key is required for Jev. but the part worth stealing is the testing setup: 1. run the strategy without AI news analysis 2. add Jev’s sentiment bias 3. test the regulatory veto separately 4. combine both 5. change the trading fees and see whether the result survives and here’s the uncomfortable bit: the builder’s published tests don’t show a consistent return improvement from adding Jev. no magic money printer. an open repo that lets you test whether the AI actually helps or just makes your bot more expensive. that’s more useful than another “autonomous trading agent” with a green screenshot and no code. save this if you’re building trading agents.
这条是引用着另一条写的：x.com/sector_sol/status/2093769463065702779