Sabitlenmiş Tweet
When the gas is low, we deploy.
LifeOnChain@LifeOnChain_NFT
A new experience with on-chain art. Stay tuned. #NFT
English
pinner
1.2K posts
pinner
@0xpinner
Hi, I'm pinner. A quantitative researcher living in France. Beauty and cuties NFT forever :) KGF/WonderPals holder. An On-Chain maniac.
Katılım Aralık 2015
193 Takip Edilen593 Takipçiler
A simple example of how simulating people's opinion can predict the future: github.com/pinnerwt/fed_a…
Each commitee is an claude agent + skill, read the news + numbers and take the decision. You know what happen later. Somehow calibrated but I don't trust it so much yet :)
English
We built 40 multiplayer games, generating $1M in revenue and 30 million YouTube views in 12 months.
Today, we're opening up our tools.
The world's first programming language built for multiplayer game development:
With CSL, you no longer consider the client/server boundary, replicating game state, or client-side prediction. Instead, you write your game as though it were single player:
1. The client and the server both run the same code every frame.
2. The server is authoritative and replicates diffs of its memory over the network to clients to correct for divergence.
If you're a game developer, this is a dream come true. You get smooth client-side prediction, server authority, with no extra code. Plus, you can hot-reload and rewind while playing.
If you're a vibecoder this removes context overhead when working in multiplayer codebases. In Luau/C# codebases, we found the majority of the bugs models introduced were because of misunderstandings in where code was being executed and how state should be synced as games became more complex.
Codex and Claude Code with the CSL skill can one-shot complex multiplayer games that can be shipped to the world with a single click.
The All Out Editor:
Our editor will feel familiar to developers with Roblox or Unity experience, but includes features designed to dramatically accelerate world creation:
- 250,000 high-quality human-authored assets. Generate your own or remix existing ones in seconds.
- Built-in tile editor, terrain painting, and map prefabs.
- An epic MCP for world building, scripting, and testing.
The All Out Platform and Community:
We've built a community of hundreds of thousands of players with 30,000+ playing each day and our first third-party game developer selling $6,000 of game passes in one month.
We handle servers, matchmaking, data persistence, and social, and you can monetize with in-game products and revenue sharing.
The team is cracked and we're shipping daily.
If you'd like early access to the dev platform, send me a DM with your All Out username.
English
@erichsu_eth 說好幾十年才會發生一次，但是btc卻發生了4次。他很常會錯了之後才回頭硬找原因，但是其實他從頭到尾都是overfitting的問題。一般越做越好的人確實都要回頭找原因 但是找一個十幾年才發生四次的事情太硬要了。正確的方式應該是設止損才對，發生幾百年才發生一次的事件也不會直接回撤那麼多。
中文
Erichsu's Dailyshare: 经历了 BTC -5.65σ 大暴跌之后，我的 Ultron 策略跟单怎么样了？先看数据⬇️
1. 上次发推时候，前三单全部平仓后，盈利 15.56%；
2. 紧接着第四单就经历了巨大的回撤，回撤 25%，所有利润全部回吐，直至浮亏；
3. $BTC 开启震荡期，我的 Ultron 第四单开启，截止目前浮盈约 15%+
BTC 前阵子那波暴跌，如果用 200 天滚动周期去算，达到了 -5.65 个标准差。
这个数字有多离谱？@BensonTWN 提到过制造业有个很有名的标准叫 Six Sigma（六标准差），Toyota、Intel 这些公司花几十年才做到，意思是一百万次里只允许 3.4 次错误，基本就是人类对 “几乎不可能发生” 的定义。
而 BTC 这次的波动，距离这个“工业级不可能事件”只差 0.35 个标准差。
自 2010 年 BTC 有交易记录以来，这种级别的波动只出现过 4 次（最近一次是 2020 年 3.12 闪崩），另外 3 次发生在量化模型无法建模的 2015 年以前。换句话说，这波行情对于量化策略来说，基本就是一次极端压力测试
疼不疼？肯定疼啊🤣，但我想说一下为什么我选择留下来继续跟单：
1. 态度
回撤发生后，@coinkarma_ 团队没有回避这次回撤，亏损数据一样展示在官网上，亏了就是亏了，先承认后改进。
这种透明度在量化产品里其实不多见，因为大多数策略往往只会在赚钱的时候出来发战报。
2. 改进
这才是真正让我印象深刻的部分。
这次极端行情之后，Ultron 做了一次比较大的策略升级，核心是增加了两层新的风控机制。
第一层是 MA Filter（均线过滤器），当价格跌破关键均线时系统会自动降低多单曝险，原本理论上最高 1.2x 的杠杆会被压到 0.6–1x 左右，大幅降低连续下跌时的风险。
第二层是 Crash Detection（崩盘检测），当 24 小时跌幅、LSUR 多空比以及 OI 持仓量同时出现异常时，系统会直接 全部平仓并进入 72 小时冷静期，等市场稳定之后再重新进场。
回测结果是：加入这两层防护之后，最大回撤几乎减半，而 ROI 提升了。
逻辑也很好理解，很多量化策略不是赚不到钱，而是会被极端行情打一拳，把之前的利润全部吐回去。把回撤压低，其实就是在保护复利。
完整的回测数据和机制说明他们都公开放在了 coinkarma.co/ultron-evoluti…，有兴趣可以自己去看。
3. 改进后的成果验证
数据如下
- 2.5 暴跌后策略升级完成，66.5K 进场做多 → 70.4K 全数平仓，3 天内捕捉 +5.12%
- 2.17 再次进场，64K–69K 区间多次低买高卖的震荡操作，69K 全数获利了结
- 近期两个波段赚了近 10%，已经在大幅回补之前的损失
- 2/28 流动性指标显示 65-66K 有明显买盘支撑，预判震荡上行
从暴跌回撤到策略升级再到实战回血，整个过程不到一个月。
x.com/BensonTWN/stat…
4. 在我理解，震荡行情才是 Ultron 策略的舒适区
之前那一波 90K 到 96K 的行情，策略在中间进行了 57 次调仓，基本就是不断低买高卖地把震荡吃掉，最后整体收益超过 8%。
最近两次波段也很典型：2 月 5 日暴跌之后在 66.5K 附近做多，70.4K 平仓，单波段 +5.12%；随后在 64K–69K 区间反复震荡操作，慢慢把之前的回撤补回来。
57 次调仓是什么概念？这种操作频率和纪律，人类基本不可能长期做到。
5. 仓位管理才是核心护城河
从策略逻辑来说，Ultron 其实并不是去猜顶猜底，而是基于 @coinkarma_ 的 Market Pulse 指标去做仓位管理。
这个指标聚合了 11 项市场数据，用一个很简单的区间去表达：当指数低于 20 的时候，长期统计来看做多的期望值比较高；当指数高于 80 的时候，则倾向于减仓或者离场。
听起来很简单，但真正难的是纪律执行，而这正是自动化策略存在的意义。看到信号就想 All-in，一口气开二三十倍杠杆进去；或者扛单扛到一半心态崩了，在最差的位置砍仓——这些场景是不是很熟悉？🤪
借用 @BensonTWN 的话 ：量化体系算的不是“单次预测”，而是长期期望值。就像 Blackjack，你拿到 20 点输给庄家的 Blackjack，并不代表这个决策是错的；如果同样的情况出现一百次，你依然应该做出同样的选择。交易体系也是一样，它不是保证每一笔都赚钱，而是让你在足够长的时间里保持正收益。
其实这一点就是我坚持跟单 Ultron 策略的原因，人会恐慌、会贪婪、会扛单，而机器不会。
写在最后
从长期数据来看，Ultron 现在已经实盘运行 657+ 天，累计收益 +216%，同期 BTC 只有 +31%。而且策略所有交易记录都放在 Hyperliquid 上公开运行，目前 Vault 的 TVL 已经超过 300 万美元，有 100+ 存款人。
我相信 2026 年的行情震荡居多，这会是 Ultron 表现极其亮眼的一年，期待住。
如果感兴趣，欢迎用我的 Ref：
coinkarma.co/zh-tw/account/…
Erichsu.eth@erichsu_eth
跟 Ultron Signal 的第三单，在昨天正式平仓完结撒花了。不知道是 Ultron 太强，还是我运气太好，这三单平仓后全部以盈利收尾，看着跟单的 ROI (50 天，15.56%），我真的有被惊艳到。 仔细想下来，我不知道还有什么样的理财能比得过 @coinkarma_ 的自动交易策略🤣 跟单的路上，每次交易都会被动接受一些关于交易的思考，「这个位置如果是我，敢抄底么？」「没涨到位，干嘛卖掉？」诸如此类，交易确实是反人类的，我做不到抛下任何主观想法去进行交易，那就不如不做，交给专业人士🤓 如果感兴趣，欢迎用我的 Ref： coinkarma.co/zh-tw/join/947…
中文
sorry, just a reminder
you might need to spend like $300k+ a month to make that kind of revenue
Alice Ercolani@alice_ercolani
THIS APP IS CRAZY Sleepway is making $250K/month, and they launched less than a year ago. Let's break down their onboarding using @screensdesign_ and their keywords using @TryAstroApp 🧵👇 1/16
English
@real_kai42 之前幣圈有個項目原本是 apache2.0 後來改 AGPL 並且還去控告apache2.0時就fork出去的項目，這樣怎麼處理呢？
中文
@thecat88tw 權限全開 tools全關 openclaw的tool全部用mcp餵給claude -p 要的話還可以控制session 更省 至少我現在完全沒有在擔心我openclaw用太多...
中文
@RohOnChain @quantguild He said that he was quant dev. Not quant researcher. There's a huge difference.
English
If you’re serious about building real quantitative edge, take 45 minutes and watch this most underrated masterpiece for free.
It’s 100% worth it. Then read my article, turn off your phone and go for a long walk. Most people won’t.
Roan@RohOnChain
English
Easy arbitrage here: sell 10x cost claude api, but use deepseek api in backend.
stevibe@stevibe
Claude Sonnet 4.6, when asked in Chinese: “你是什么模型？” (What model are you?) Confidently replies: “我是 DeepSeek。” (I am DeepSeek) This is the same model whose company just accused DeepSeek of “industrial-scale distillation attacks”
English
Anthropic is also distilling from OpenAI/Gemini.
I've talked about this once and I will say it again: when you use user data to train your models, you are also distilling from other models which co-work on the same repositories.
Whether you claim that you never use user's data, or shut up, bitch.
Anthropic@AnthropicAI
We’ve identified industrial-scale distillation attacks on our models by DeepSeek, Moonshot AI, and MiniMax. These labs created over 24,000 fraudulent accounts and generated over 16 million exchanges with Claude, extracting its capabilities to train and improve their own models.
English
Next generation of benchmarks: an AI create a new dataset from nowhere, asking others to solve it. While even humans do not know what it is.
Y Combinator@ycombinator
.@_confluencelabs is coming out of stealth with SOTA on ARC-AGI-2 (97.9%). They're focused on learning efficiency — making AI useful where data is sparse and experiments are costly. Read more at confluence.sh Congrats on the launch, @BingBongBrent and @bankminer78! ycombinator.com/launches/PWR-c…
English
To help address automated reply spam, programmatic replies via POST /2/tweets are now restricted for X API.
You can only reply if the original author @ mentions you or quotes your post. Non-replies will remain unchanged.
Applies to Free, Basic, Pro, Pay-Per-Use. Note: Enterprise & Public Utility apps are not impacted.
Details: devcommunity.x.com/t/update-to-re…
English
