好運狗 (@goodluckdogg) profile picture

热爱纯粹 任性的去爱

Joined June 2022
Photo by 好運狗 (@goodluckdogg)Photo by 好運狗 (@goodluckdogg)Photo by 好運狗 (@goodluckdogg)Photo by 好運狗 (@goodluckdogg)Photo by 好運狗 (@goodluckdogg)Photo by 好運狗 (@goodluckdogg)Photo by 好運狗 (@goodluckdogg)Photo by 好運狗 (@goodluckdogg)Photo by 好運狗 (@goodluckdogg)Photo by 好運狗 (@goodluckdogg)
已轉錄分析,可以私信我找原始csv,接下來會發更詳細的交易紀錄復盤,重點鎖定在魏神的掛單撤單邏輯,掛買掛賣的倉位管理(總倉位追蹤還在還原,目前只有full trading record:2020-05-01 to 2025-11-23)
我就是只看K线,手动交易的“原始交易者”, 但不是崇拜者,只是论道,跟爱憎也没关系。 因为我认为条条道路通罗马,不同方法体系各有优劣,以及不同的方法,对于不同的人和市场来说,适合度也不同。不认同粗糙地褒扬一种方法的同时,把另一种方法过度贬低至无用。 1. 多数“逐笔/精细量化分析交易者”, 平均表现上, 能超越“只看粗糙K线的原始交易者” 这个结论问题不大, 但 “A>B” 和 “B行不通、没有用” 显然是两码事, 这样偷换概念,会更容易引起更多拉低质量的片面讨论。 2. 我的“只看粗糙K线的原始交易” 有近4年多实时全透明公开数据Bitmex实盘,在币coin, 近一周/4个月,也开了HyperLiquid、币安两个实盘, 在最近一个月的推文里都可以看到,平台上公开都可查, 长期、中期都跑赢BTC。 并且前两个实盘,都是每笔成交明细、盈亏实时展示, 甚至HyperLiquid平台上,连挂单状态都可以看到。 向来无任何带单、不付费加群、不搞交易返佣、不接广。 “不为结果负责”这一点倒是说对了, 因为没有上述这些,纯凭兴趣、交流热情, 不跟人有资金往来,自然不用为任何人的结果负责。 说某类方法有没有用, 最简单的论证方式就是,拿可验证的成绩辅助论证。 只要这方法,有人能长期做出还可以的表现,便是有用, 这是再简单不过的逻辑。 除非说,我这不是“原始交易”,其实是在偷偷跑量化? 那么再严谨点, 我直接把上述4年多实盘近2年多交易量99%+都是BTC的 只读API丢这里,公开几天供检阅: Id: ADU5yR0QPA6622twEtmPxVsW Key: arodVpSl9kBFWSLkxK5gCZIFI6BV-GZ8PQWUbGS4ncI5GBhQ 欢迎“录一下数据,写程序做逐笔分析” 看看有没有“非原始交易”的量化特征? 就像竞技运动员,不看比赛成绩, 光靠嘴讨论方法有没有用,始终是片面低效的。
125
67
4
74
19,122
好運狗 retweeted
Replying to @pika_labs
This looks a lot like the Black Mirror episode “Hotel Reverie” (Season 7, Episode 3). An actress’s consciousness is dropped into an AI-generated virtual film set, where she performs live with AI versions of the original characters. Pika Enact feels like that concept brought into reality.
1
1
6
847
any investor who studies Nvidia's actual business structure/history wont be convinced it will remain the same company without Jensen. Both Jensen and Morris have said Nvidia and TSMC did tens, perhaps hundreds, of billions of dollars of business over 30 years without a detailed contract. think about what that means in semis. Nvidia's asks often times can be brutally aggressive: fix yield while the design is still moving, reserve staggering capacity before the cycle is visible, then hit a ship date the entire stack depends on. only strong personal bonding and accumulated trust does that level of heavy lifting, it's also how you can do big scale business in East Asia as a high trust high context society. that trust is how Blackwell went from a moving design into a 24/7 build across 45 factories. it is also a huge part of how Nvidia compressed Hopper, Blackwell and Rubin into a system-level product cadence in merely 4 years the rest of the compute industry can barely catchup with the last 4 years were were a war campaign commanded by one founder who could spend the morning in California arguing kernels with architects, fly to Taipei the next night for fab meeting, geonbae with Korean HBM suppliers the day after, then somehow get everyone to commit staggering capacity into a demand curve no one could yet see. no one else in global tech could fuse America's best architecture + software, Taiwan's wafers + advanced packaging, Korea's HBM, and the brutally unglamorous systems work into one iron army moving at fierce speed, i dont think even Elon can pull this off. this is why I dont see Jensen dependence as a normal keyman discount. the founder premium is embedded in the supply chain itself. It's the true founder mode for 30 yrs, unfortunately sacrificed by some hand weavy "spy" allegation as the victim of a bipastian political scheme Also now the world sees that 30 years of trust built by an American founder can be turned into political liability overnight, every global supplier will think twice before making any commitment on the next American company. that is a brutal tax on American entrepreneurship going forward
25
78
7
585
33,398
好運狗 retweeted
看完李沧东的《可能的爱》 韩国人在工人运动上是亚洲有史以来,最成功最彻底最暴力,战斗性最强也是最团结的。 韩国的工会有多成功呢? 海力士在工会的压力下不得不拿10%分红,连门卫大爷都能拿到320万的年终分红。2025年,韩国在工会的支持下,各行各业均涨薪50%以上。 20世纪60年代末,韩国经济在朴正熙出口导向政策的推动下有了汉江奇迹,在这时,国际上对韩国工人的评价和大部分东亚工人差不多,是顺从听话的,大家都沉浸在经济高速发展的幻觉中,当时朴正熙用宏大叙事洗脑pua民众忍耐、拼搏、为国奉献,默许工厂给工人吃兴奋剂。 尤其是战后韩国上下都非常反共,工人运动被视为共产主义的一部分,于是工人运动在学界和政治界都无人敢支持,不然就会被认为和北韩串通一气,从而一开始会有对工人运动的恐怖镇压,取消全国所有的独立工会,有工人反抗就暴力血腥镇压。 直到一个叫全泰壹(전태일)的年轻人在首尔东大门平和市场附近自焚抗议,手里拿着《劳动基准法》,并呼喊“我们不是机器”,自此韩国的知识分子,教会,政界才开始看到韩国的工人处境。 韩国人第一次开始摆脱顺从,听话,经济发展为先的思想观念,开始罢工,组织工会,逐渐发展成覆盖全社会的工人运动。 2008年全球金融危机后,电影的背景,韩国双龙工厂77天罢工,比起之前更加分裂,正如电影里拍的,有人下岗,有人在工厂继续工作,大家不团结,于是在警察的暴力镇压下输了,工人没能阻止裁员。 但是双龙工厂的工人硬生生打了10年官司,持续争取复职、追究警方清场责任,并通过法律诉讼、公共倡议和社会声援,在2018年迎来了他们的胜利,让失去工作的工人返回岗位,直到2020年,最后一批35名失业工人重返岗位。 李沧东的《可能的爱》里,失业职工仍聚在一起分享生活,创伤,接受纪录片导演的采访,把创伤一遍遍翻开求得大众的同情,最终获得斗争的胜利。 一场罢工即使没有当场胜利,也改变了社会如何理解失业、工会、警方和企业责任。 我看到很多中国网友其实是不理解为什么美国人,欧洲人反ai,其实很多人可能已经处于失业之中,或者在恐惧失业之中,但绝大多数的人其实对即将发生的事情是毫无感知的。 在上个世纪,亚洲经济高速发展时期,很多国家都爆发了工人运动,在工业化时期,国家、资本和工厂还需要依赖劳动力,工人可以以罢工来和企业和资本博弈,即使工人运动是血腥暴力绝望的,至少还有一丝转机。 但是当AI和机器替代人类劳动这件事真的发生,机器和资本不再需要大多数人的劳动,人还能凭什么获得收入、尊严和影响力? 时至今日,可能未来几年我们大多数人都会失去工作,请不要忘记韩国人为人权和尊严的抗争。
8
9
3
72
8,683
What a strange week for @JensenHuang and @nvidia. It all feels orchestrated and targeted, and I hope we get to the bottom of it. I’ve been at NVIDIA for 12.5 years, started as an intern and made it to Research Director thanks to incredible colleagues and especially Research management and Jensen. The journey I’ve witnessed has been fascinating. 🧵
90
361
15
3,537
156,619
好運狗 retweeted
It's a mistake for Nvidia to let Jensen resign.
5,344
6,724
1,272
93,178
3,681,866
一個人一輩子的努力跟貢獻真的可以這樣暴力瞬間被搶走的嗎!
217
82
7
2,832
164,447
好運狗 retweeted
說句實在話:如果黃仁勳最後被政治風暴拉下台,那不是輝達的損失,而是美國國運到頭的訊號。 一個移民之子,從丹尼餐廳的一張桌子起家,三十年把一家顯卡公司做成整個 AI 時代的地基。 這種人是美國夢最好的廣告,也是美國在科技競賽裡最難被複製的資產。美國的對手不必研發出更好的晶片,只要看著美國自己把最會做晶片的人推下台就夠了。 一個國家強盛的時候,會把人才捧在手上;開始衰落的時候,才會把人才當成政治祭品。
423
727
93
9,553
922,239
An unbelievable arbitrage happened in 44 trading venues swapping 24 tokens. It involved @CurveFinance, @MakerDAO, @balancerlabs, @Uniswap 's V2, V3 and V4, @BitcoinCom's Verse, @pancakefi, and @Fluid_Protocol. Making $15.8K as revenue, the bot 🤖 tipped @beaverbuild 5.6 ETH and pocketed $1.5K as income. bit.ly/4ijw5OQ
17
69
39
540
158,519
好運狗 retweeted
向大家介绍 POKI —— 一个能够真正接管并操控你 iPhone 的 iOS Agent 它可以运行在任何地方,支持通过多种方式随时调用: • 信息 (iMessage) • Telegram • 终端 (Terminal) • 原生 iOS App 同时,POKI 彻底抛开以往繁琐的使用门槛 : • 不需要任何硬件-不需要任何模拟鼠标的设备 • 不需要越狱-完全模拟原生触摸,我运行下来从未遇到风控 • 不需要开发者账号-自带签名 这两天没有发新作品,就是在全力履行我的 Phone Use 开源承诺。希望这次能给大家带来一个真正开箱即用的 iOS Agent。 今日开源,敬请期待
大的真的来了!!!用用 ESP32 控制手机,我只说了一句“问问老婆今天吃什么”,它就自己打开微信问老婆,收到回复后再打开美团外卖,选餐、下单,一路推进到付款密码页,停下来等我接手! (最后的付款密码它可以帮我输入,但是我还是留给自己输吧,安全边界还是要有的。 幼年版贾维斯现在进化为少年版!@bot @elonmusk
192
173
30
1,652
361,198
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math. rikiyat.github.io/blog/weight-sp… github.com/RikiyaT/lm_und… huggingface.co/Qwen/Qwen3.5-0… huggingface.co/RikiyaT/Qwen3.…
2,368
7,257
5,755
55,917
10,086,802
好運狗 retweeted
here's my talk at Cursor Compile some thoughts on how we build in the age of AI and what doesn't change
108
152
63
2,492
275,211
好運狗 retweeted
我觉得苹果新机发布亮点,不是 iPhone Duo, 而是2个跟 AI 有关的“黑镜”感小特性: 1. Apple Watch “倒带现实”功能 双击表冠,看过去 15 秒的对话; 聊天结束后自动生成总结。 2. iPhone 18 Pro 相机照片真实性签名 在 AIGC 泛滥时代, 像素级签名证明“这是真的薄肌照片”。 youtube.com/watch?v=39BalPDu…
61
3
1
47
78,498
我有个文科生朋友,曾经用AI vibe coding做出了一个自己的小工具。 他特别兴奋,终于体验到了自己当程序员的aha moment,于是分享到X上。结果评论区一群程序员开始挑毛病:什么生产环境不能用、代码质量太差、有bug、不规范…… 让他觉得很受挫也很困扰 对这个群体性的反应我一点都不意外。 其实程序员这个职业有一个很大的分水岭。 一类人长期负责接需求、写代码,把产品经理拆好的需求实现出来。 另一类人已经做到Tech Lead,负责技术管理和架构设计,需要理解业务、权衡成本、制定全局架构方案,对最终结果负责。 Claude Code刚出来的时候,这两类人的评价就完全相反。 做过技术管理的人,几乎都能立刻意识到它的价值,因为它极大提升了代码生产效率。 而长期做执行开发的人,更容易盯着它写出来的代码:这里不优雅,那里有bug,需要Review,最后得出一个结论:"还是人写靠谱。" 问题在于,他们关注的是代码本身,而不是代码创造的价值。 事实上,这部分程序员里有相当大的比例,沉迷于欣赏所谓的"技术"。 他们会不断挑AI写代码的问题:架构不好、屎山、稳定性差、生产环境不能用…… 可他们往往讲不清楚,自己写的代码到底创造了什么价值。 因为只要没人给他们提需求,他们就无事可做。 写代码,到底是目的,还是满足市场需求、创造商业价值的手段? 对于真正有判断力的人来说,知道该做什么远比知道怎么写代码重要。 AI出现以后,补足的正是代码生产力。 于是,当需求方开始直接使用Code Agent满足自己的需求,而不是再把需求层层传递给程序员时,一部分人的第一反应不是学习,而是否认。 之前有朋友问我,他一直不理解为什么那么多程序员喜欢跟AI较劲。 明明AI就是个工具,不好用改一改、优化一下不就好了? 我回答了一句话: 因为他们自己在工作里,本来就像一个Agent。 他们不会把AI当作自己的杠杆,而会把AI当作自己的竞争对手。 因为AI正在接手的,恰恰就是他们一直负责的那部分工作。 在没有AI之前,一个成熟的软件团队本来就是这样运作的。 Tech Lead负责理解需求、权衡人力成本、开发成本、维护成本、业务优先级,甚至参与商业决策,再设计整体架构,保证长期可扩展,同时避免过度设计。 模块拆分完成之后,再分配给执行层开发,由Tech Lead审核方案和代码。 听起来是不是很熟悉?执行层做的事和Agent工作几乎一样。 所以后来我安慰那个文科生朋友。 我告诉他,不用和那些人争论,你应该理解他们为什么会有这样的情绪。 因为当你开始学着用AI自己vibe coding的时候,你其实在做空他们的未来。
53
11
3
137
21,184
好運狗 retweeted
Introducing GLM-5.2: Frontier Intelligence, Open Weights - Significant improvements in coding and agentic tasks - Strong long-horizon capabilities with a 1M context window - Two levels of reasoning effort: GLM-5.2 (max) pushes the limits, while GLM-5.2 (high) strikes a strong balance between performance and token efficiency - MIT-licensed open weights - Same API pricing as GLM-5.1 Tech Blog: z.ai/blog/glm-5.2 Weights: huggingface.co/zai-org/GLM-5… API: docs.z.ai/guides/llm/glm-5.2 Coding Plan: z.ai/subscribe Chat: chat.z.ai
719
1,726
957
13,059
7,684,224
好運狗 retweeted
人的能量是有边界的。要远离“扶贫式社交”。 那种关系不是互相滋养,而是你单方面输出时间、情绪、资源和耐心,对方却很少回流。久了以后,人会变钝,机会会变少,连财运都会被拖慢。 被别人的情绪、因果、欲望搅进来,把本来属于你的主线打散。低质量关系会消耗你,过度社交会稀释你。
2
5
50
4,086
Had a 4hr+ dinner with a SpaceX tech lead fren who joined in 2015. Gotta write down the hot take because it is very different from the usual “ @elonmusk is a tech visionary” story The less glorified employee perspective is actually more interesting: many core engineers did not stay locked in for 10+ yrs purely because of “the vision.” As frontline operators, they often did not even know what concrete vision to believe in. Going to Mars 10 yrs ago sounded too delusional to be a real operating plan. My fren joked that it was more like “believing in the power of believing” lol What they did believe in was Elon’s ability to keep winning the next war chest: capital, debt, federal contracts, customer deposits, attention, narrative, most importantly market patience. That matters a lot more than outsiders realize, and ofc thanks to the big mom who keep all the shits together @Gwynne_Shotwell In frontier engineering, morale does not only come from mission. It comes from believing the company will survive long enough for impossible work to compound. Elon’s real genius may be that he turns capital formation into an engineering input. Put together marginally enough resources > buy engineering time > ship impossible progress > earn more credibility > unlock more resources > make the next impossible bet fundable This is why his companies keep forming silo network effects then Elon threaded them into a massive macro network effect - x com first - Tesla on earth physical traffic + charging grid + solar grid - Tesla FSD data network & edge compute network, which was a very underrated E2E track for AV (and most wouldnt bet on that track, ie Waymo spent massive money relying on lidar and HD map) - Starlink for connectivity first and now compute - Twitter/X for on earth digital traffic - xAi and Colossus compute network ... He also mentioned how X Money fills in the puzzle, which is not random, as the macro network now is looking for its own banking layer, all of which are high-stakes financial gambits that constantly need capital, debt capacity, payment rails, and balance-sheet engineering load. He has apparently hinted more than once that maybe spacex should just build a bank. The X Money card is still beta, but the perks are already wild: 6% APY and 3% cashback. Employees are extremely doped about it, not just for the perks, but because it makes them feel like first-class citizens of “Elon Nation.” and this might ignite another network effect that will connect them all and eventually evolve into a full bank. Tesla built the energy/data network. Starlink built the connectivity/orbital compute network. X built the attention/distribution network. xAI built the reasoning and inference network. X Money tries to close the financial network. Ya ofc Elon is one of the biggest "technology visionary" in our life time, but I increasingly think the he's the best network effect engineer on earth, which will get us to brute force out of earth itself. And the biggest network of all like @balajis has shared, is the network state. Welcome to Elon Nation
37
70
14
1,055
121,963
好運狗 retweeted
Reminder that you can use the Codex App, CLI and SDK with any open source model, not just with OpenAI models. developers.openai.com/codex/…
477
707
435
6,898
2,175,671