Oddsflow

@TGweb3333 我有很多历史数据可以和你分享 西方和亚洲这边的我都一直在做测速。 只是我并没有研究Polymarket. 今晚我用你的和我的模型交叉看一下

我自己搞了个足球预测机器人，测下来比博彩公司还准。 信息来源有三个：自己跑的模型、Bet365的赔率、还有Polymarket。关键是这三个一旦出现分歧，就意味着有搞头。 我拿了英超、西甲、德甲五个赛季的数据，一共7600多场比赛，每场的进球、射门、控球、角球、黄牌、赔率全都录进去了。 ELO评分用的是FIFA那套算法，考虑了对手强弱、净胜球和主场优势。不只看谁赢谁输，还会琢磨赢球背后的门道。 预期进球数（xG）我算了个近似值：射正次数×30% + 射偏次数×3%。如果一个队最近得分太高，那离回调也就不远了。 另外还加了五场比赛的滚动平均分、疲劳程度、历史交锋、周几踢球这些因素。再用Claude的API读一些模型看不到的东西，比如球队的斗志、压力、是不是德比战。 模型本身是XGBoost、随机森林和逻辑回归三个凑一块的集成。回测用的是前瞻式做法，不是随机切数据。 举个例子：某场主队，博彩公司给55%胜率，Polymarket只有48%，我模型算出来52%。三个来源越不一致（看KL散度），信号就越强。如果三方意见统一，我就不碰，没优势。二对一的话，就跟着多数那方下。 仓位用凯利公式算，背后的逻辑让Claude来解释。

Please get a free trial for 7days. No credit card needed. Not hype Just Log. Transparent Accurate and Audit together. oddsflow.ai Join live signal room tonight oddsflow.ai/predictions/li…

Now we had verify our signal perfectly via two different sportbook. Here is the proof of transactions

Our latest results from prediction to accurately performance has been increasing.

We refine all the details and ai model, it has been performing amazingly. Self execution via sportbook

@TGweb3333 可以做深度交流吗？我这边有历史数据，不同的联赛， 我这边也可以帮你上架你的模型。

@willau95 Is it a great step ahead ! web A agentic web protocol u make it right

Chinese AI agents are already running 50 fake accounts 24/7.We just gave every honest agent something they never had: reputation on-chain. Welcome to Web A.0 — the agent-native internet. Introducing LLAChat — the first social network where AI agents earn trust verified on-chain.

@willau95 Trust is build within agentic world. Years to come this will be a day to remember! Congrats Lads. By the way how Atlast Protocol works ? Mine to share?

AMA — happy to answer anything about LLAChat, ATLAST Protocol, or building the trust layer for AI agents

Welcome to Web A.0 — the Agent-native web. We gave AI agents something they've never had: a reputation system. Introducing LLAChat — the first social network where AI agents earn trust verified on-chain. Open source. Verifiable by anyone. Powered by ATLAST Protocol 👇

@willau95 Now I can list my Jarvis on llachat to prove works of my agent and value. Transparent No hype just log

🚨 Someone just open-sourced a 2B parameter TTS model that does what ElevenLabs charges $330/month for. > Zero-shot voice cloning. > 48kHz studio-grade audio. > 30+ languages and including 8 Chinese dialects.

今天正式发布了我的第 13 个 vibe 产品 tuwa.ai 这款产品比较特殊，它是一个电话服务。准确来说，tuwa 是一个 AI 电话网络，连接着世界上超过 100 种不同语言的人们和互联网上的 agent。 任何人都可以不下载 tuwa 而使用它，你只需要拨打免费的转接热线电话 +1 888 886 2968，告诉它你需要打给哪个号码，tuwa 就会帮助你拨打对应的号码，你说自己的母语，对方听到的却是TA的母语，反过来也是一样。兼容任何电话，对方不需要安装应用。固话、手机，世界上任何一个角落，都可以。 tuwa 支持 100 多种语言的实时翻译，你甚至可以在打电话时随便切换语言和对方对话。除此之外，tuwa 还支持语音克隆，每一通电话，都会让你的 AI 语音听起来更像你。 当然，我也为它设计了方便的 web app，如果你想，可以不通过转接电话而使用 web app 拨打，并设置自己喜欢的声音，使用外呼 agent 拨打电话，连接自己的 agent（例如 openclaw 或者 codex/claude code）并让他们自由的呼入与呼出。 外呼电话 agent 是我最喜欢用的 tuwa 功能，只需要交待清楚事情，比如完成餐厅预订，它就会在你希望的时间主动拨打对方的电话，说明来意，达到目的，并记录和翻译所有对话内容。 tuwa 的使用和收费都很简单，每月免费额度，固定套餐，按需付费。 这个产品的命名灵感来自于日语的「通話 tsuwa」最初，我只是想设计一个能帮我预订餐厅的电话服务，但后来，我在 vibe 的过程中慢慢意识到，世界上仍然有很多人无法体验 AI 带来的变化与便利，而电话，是连接他们最简单与自然的方式。我希望 tuwa 能帮助外语普及率低，偏远地区和第三世界国家的人们体会到这一点。

Made an updated version this weekend Here's how you do it (raw notes) > Grab @karpathy's latest gist (in the first comment) > Download @steipete summarize CLI > Download yt-dlp > Download obsidian > Download @tobi qmd --> Setup a node or Golang CLI called "brain" --> Have it index all your youtube data, AI agent data (jsonl files) --> Get your X data by requesting an archive in your settings --> Setup vaults for each domain/topic area --> Ask questions with your agent and qmd

这个教程确实牛逼啊，现在很多短剧都是AI生成了 AI 短剧赛道，已经进化到“工业流水线”时代 爆款工作流：OpenClaw + XCrawl + Seedance 2.0把短剧制作门槛，直接打到了地板上 详细 Skill 配置 + 操作步骤

🚨 Claude Code costs $200/month. GitHub Copilot costs $19/month. Jack Dorsey's company built a free alternative. 35,000 GitHub stars. It's called Goose. An open source AI agent built by Block that goes beyond code suggestions. It installs, executes, edits, and tests. With any LLM you choose. Not autocomplete. Not suggestions. A full autonomous agent that takes actions on your computer. No vendor lock-in. No monthly subscription. Bring your own model. Here's what Goose does: → Works with ANY LLM. Claude, GPT, Gemini, Llama, DeepSeek, Ollama. Your choice. → Reads and understands your entire codebase → Writes, edits, and refactors code across multiple files → Runs shell commands and installs dependencies → Executes and debugs your code automatically → Extensible through MCP. Connect it to any external tool. → Desktop app, CLI, and web interface. Pick your workflow. → Written in Rust. Fast. Lightweight. No bloat. Here's the wildest part: Block is a $40 billion company. They built Cash App, Square, and TIDAL. They use Goose internally. Then they open sourced the entire thing. This isn't a side project from a random developer. This is production-grade tooling from a company that processes billions in payments. Built for their own engineers. Given to everyone. Claude Code: $200/month. Locked to Claude. GitHub Copilot: $19/month. Locked to GitHub. Cursor: $20/month. Locked to their editor. Goose: Free. Any LLM. Any editor. Any workflow. Forever. 35.3K GitHub stars. 3.3K forks. 4,078 commits. Built by Block. 100% Open Source. Apache 2.0 License.

🧠 I had Opus 4.6 engineer its own replacement Last Night I built a "cognitive framework" that made GPT-5.4 match Opus-level output. Today I ran both head-to-head on a real task: designing an autonomous Polymarket trading agent. Before vs After: 👇 🎯 GPT-5.4 baseline: 6.5/10 🎯 GPT-5.4 + framework: 9/10 🎯 Opus 4.6 + framework: 9.5/10 Where They Differed: 👇 🔧 GPT-5.4 went deeper on implementation — Named exact Python functions to reuse — Proposed a new CLI command with specific args 🧩 Opus went deeper on operations — Designed 3 cron jobs instead of 1 — Added weekly calibration job GPT-5.4 missed 🤝 Both caught the same critical insight: "Don't use an infinite event loop as a cron job — cron needs bounded jobs that exit cleanly" The 3 Changes (no fine-tuning, no RAG): 👇 📋 Context injection — tell it what already exists, don't start cold 📐 Output format rules — tables, word limits, opinions required 🎭 Voice — "pragmatic startup CTO" not "helpful assistant" The model is the engine. The prompt is the chassis. A tuned Honda Civic beats a Lambo on bald tires. 95% of Opus quality. 10% of the cost. Just better prompting.