Bitcoin Headed for First Monthly Loss in 6 Months

CoinDesk发布于2023-05-31更新于2023-05-31

文章摘要

Bitcoin (BTC) has regained some poise since last Thursday, but the cryptocurrency still appears on track for its first monthly loss since December.

Bitcoin (BTC) has regained some poise since last Thursday, but the cryptocurrency still appears on track for its first monthly loss since December.

The leading cryptocurrency by market value traded near $27,800 at press time, a 7.5% rise from lows under $25,900 registered last week. However, prices were still down about 5% for the month, the first monthly decline of the year (assuming, this loss is held through Wednesday's UTC close). Bitcoin has put in a positive performance in January, March and April and ended February on a flat note.

Against ether (ETH), bitcoin looked set for a monthly decline of nearly 7%, CoinDesk data show.

Bitcoin's dour monthly performance comes as bond traders have reinstated bets that the Federal Reserve (Fed) will keep interest rates elevated for longer in response to sticky inflation and a resilient labor market. Earlier, interest rate traders expected the Fed funds rate, the benchmark borrowing cost, to fall to 4.5% or lower by the end of 2023 from the current 5%. However, the market no longer foresees the Fed implementing rate cuts this year.

The renewed hawkish Fed bets have given a boost to the U.S. dollar this month, lifting the greenback by 2.7% against a basket of fiat currencies, including the euro. Bitcoin tends to move in the opposite direction of the dollar.

Capital has been leaving the crypto market since early last year. The trend has persevered this month, with the stablecoin market capitalization shrinking to a 20-month low of $130 billion. Stablecoins are digital assets with values pegged to an external reference like the U.S. dollar and have been widely used to fund purchases of other cryptocurrencies over the past three years.

"We can assume that the liquidity wave of lower inflation has now run its course and the market needs a new driver and theme to lift prices higher," Markus Thielen, head of research and strategy at crypto services provider Matrixport, said. "The tech sector tends to be correlated with BTC, but the former has found new life with the AI and Chat GPT revolution, which is not benefiting BTC yet."

Bitcoin has decoupled from Wall Street's technology-heavy index Nasdaq, which has risen nearly 8% this month.

Griffin Ardern, a volatility trader from crypto asset management firm Blofin, said the continued high-interest rate environment would keep the odds against bitcoin bulls.

"In a high-interest rate environment, high risk-free returns such as money market funds are more attractive to investors, which means the lack of liquidity in the crypto market continues," Ardern said.

Dick Lo, the founder and CEO of quant-driven crypto trading firm TDX, said, bitcoin's 4% rise on Sunday was a relief rally triggered by U.S. leaders announcing a provision deal to lift the $31.4 trillion debt limit hit in January and further gains may be hard to come by.

"The rebound we saw on Sunday night / Monday morning was very much a relief rally on the back of the U.S. debt ceiling package. The market will likely return its focus to the possibility of another 25 basis points interest rate hike at the June FOMC meeting and the potential liquidity drain as the Treasury needs to sell at least $500 billion in bills in the short-term to refill its cash position, which will weigh on risk assets," Lo said.

We see strong resistance on BTC at $28,500 with initial support seen at $27,350, followed by a potential retest of $26,200," Lo added.

你可能也喜欢

黄仁勋:Prompt正在过时,Loop才是新范式

黄仁勋等AI领域领军人物提出,传统的“Prompt Engineering”(提示词工程)正在被“Loop Engineering”(循环工程)取代。新范式的核心是从手动编写具体指令,转变为设计和管理能够自动运行、自我验证并持续优化的循环系统。 **Loop是什么?** Loop(循环)指一种系统设计:用户设定一个目标,AI Agent(智能体)会自动执行任务,完成后由独立的验证机制(如另一个模型)进行验收;若未通过,则带着反馈重新执行,直至成功或达到预设限制。人的角色从“逐句指挥”转变为“规则设计者”。这与单纯使用Agent的关键区别在于,Loop为其提供了自主持续工作的管理框架。 **当前落地情况** 目前,Claude Code和OpenAI Codex是Loop理念的主要实践者。两者都实现了将复杂任务拆分,由多个Agent并行处理并汇总结果。其核心设计原则是“执行与验证分离”,以避免自我评分时的宽松问题。例如,“Claude Code之父”Boris Cherny描述其工作流:数百个Agent在隔离环境中自动处理问题,仅疑难事项才需人工干预。 **如何构建有效的Loop?** 1. **前提测试**:任务需具备重复性、可自动化验收、成本可控且Agent拥有必要工具。 2. **从最小可行Loop开始**:包含触发器、技能定义、状态记录文件和自动化验证门禁。务必先手动跑通流程再自动化。 3. **执行与验收分离**:必须由独立的Agent或模型进行验收,确保评估客观。 4. **避开常见陷阱**:设置硬性停止条件(如token上限)、持久化状态、不让Loop处理需要主观判断的任务、坚持阅读代码变更以维持理解。 5. **核心衡量指标**:关注“每个被采纳改动的平均成本”,接受率低于50%意味着Loop可能效率低下。 **范式演进路径** Loop的兴起是AI应用控制粒度不断上移的结果:从**Prompt Engineering**(优化单次指令)到**Context Engineering**(组织背景信息),再到**Harness Engineering**(构建执行环境),最终演进至**Loop Engineering**(设计自主循环系统)。这一过程逐步将人类从具体操作中解放出来。 **渊源与反思** Loop的理念在学术界早有雏形,例如姚顺雨等人提出的ReAct框架,将“推理-行动-观察”构建为循环。业界火爆的背后也需冷思考,专家提醒需警惕token成本,并指出“AI可以外包思考,但无法外包人的理解”,强调在自动化中保持对问题的根本认知至关重要。

marsbit40分钟前

黄仁勋:Prompt正在过时,Loop才是新范式

marsbit40分钟前

以太坊基金会临时执行董事发声:我们的使命是什么?

以太坊基金会(EF)临时联合执行董事 Aerugo 发文明确了基金会的核心使命:确保以太坊始终是真正无需许可、保障用户自主权的基础设施,能够抵御审查、开源自由、私密安全,并支持大规模的自主协调。 文章首先明确了EF的“不为”:不追求自身重要性或商业吸引力,不服务短期投机者,也不推广所有应用。其核心作用是“消除弱点”,防御以太坊在协议、访问、用户和机构各层可能出现的榨取性、控制性或监视性风险。具体工作重点包括: 1. **自身实践**:将EF薪酬与财务转向以太坊原生资产,以身作则使用其力图改进的技术栈。 2. **对抗有害MEV**:将其视为核心威胁而非次要市场问题,致力于从系统层面降低MEV提取,防止交易流程被特权供应链把控,维护可信中立执行。 3. **强化隐私**:认为缺乏严格隐私默认设置的公链实为监控平台,必须优先实现强大的无条件隐私保护。 4. **保障质押去中心化**:视质押为协议安全基石,防止其集中在少数发行者或运营商手中。 5. **维护访问层自主性**:确保用户和机构能自主、抗胁迫地访问以太坊,而非通过削弱其核心价值的妥协方式。 同时,EF也致力于“抓住机遇”,推动以太坊发展为:抗量子攻击的全球基础设施;具备完全自主可验证、最小化有害MEV的协议栈;私密、尊重尊严的普通数字现金与协调平台;用户拥有完整主权的个人AI代理钱包;以及在机构场景中凭借其可信中立性胜出,而非沦为后端工具。 文章最后提及了近期的人员变动与“衍生公司”问题,强调离职原因多样,EF尊重所有贡献者但不会公开讨论个人事宜。对于从EF分离出去的项目,EF将谨慎评估资助,标准在于其是否对实现以太坊核心使命(无需许可、自主权等)至关重要,而非出于人情或延续旧项目。EF明确表示对以太坊的发展方向并非中立,将全力支持并优先推进那些保障其核心特性的工作。

链捕手1小时前

以太坊基金会临时执行董事发声:我们的使命是什么?

链捕手1小时前

读博最后一年转方向,拿到OpenAI offer:我的面试之路充满「意外」

布朗大学博士生Yong Zheng-Xin(中文名应为“永正新”)将在下个月以Astra Fellow身份加入OpenAI,专注于AI安全研究(AI Safety Research)。他在博士最后一年从多语言大模型研究方向转向AI安全领域,并分享了求职过程中的六个“意外”发现: 1. **论文数量并非关键**:获得面试和工作机会,真正起作用的往往只有一两篇高质量论文,甚至无需论文,更看重当场解决问题的能力。 2. **面试形式高度多样**:除了常规技术面试,还可能涉及系统设计、并行编程,甚至考察使用AI智能体的能力。 3. **工作试用期渐成常态**:尤其是在AI初创公司,候选者可能需要与团队合作完成一项有时长达一周的有偿任务,这会影响同时准备其他面试。 4. **时机至关重要**:就业市场的热点、职位需求窗口期、个人研究作品的走红时机,都会极大影响求职过程和结果。 5. **研究职位鲜有“留用”机会**:与软件工程岗位不同,研究类职位(如实习、奖学金结束后)通常仍需经历完整的面试流程才能转正。 6. **面试内容常与研究方向无关**:尽管他转向了AI安全领域,但许多面试环节并未专门考察该领域的知识,而是评估其作为AI研究员的综合基础能力。 他建议,在准备面试时要广泛学习,夯实基础,并可根据情况合理要求调整面试时间。最终,他对自己能获得满意的工作机会感到庆幸。

marsbit1小时前

读博最后一年转方向,拿到OpenAI offer:我的面试之路充满「意外」

marsbit1小时前

交易

现货
合约
活动图片