# Artikel Terkait Data Analysis

Pusat Berita HTX menyediakan artikel terbaru dan analisis mendalam mengenai "Data Analysis", mencakup tren pasar, pembaruan proyek, perkembangan teknologi, dan kebijakan regulasi di industri kripto.

What Kind of People Is Anthropic Actually Hiring? Insights from 1,680 Resumes Provide the Answer

Contrary to the common perception of Anthropic as a PhD-heavy AI research lab, an analysis of 1,680 engineer profiles reveals its core talent is experienced "builders." The engineering team, largely assembled in the past 18 months (median tenure 10 months), hires senior engineers with a median of 12.2 years of prior experience. They primarily come from major engineering-focused companies like Google, Meta, Amazon, Microsoft, Stripe, Databricks, Snowflake, and Palantir, not primarily from other AI labs. Their backgrounds are infrastructure-heavy (40%), focusing on backend, distributed systems, databases, and security, with skills like Python, Java, C++, and AWS being common. Only 13.7% hold PhDs. The dominant job title is "Member of Technical Staff." For early-career hires (<6 years experience), entry is not based on standard work history but on prestigious signals: top internships (e.g., at FAANG or quant trading firms), competition rankings, research publications, or AI safety fellowship experience. The key insight for applicants is to present as infrastructure builders, highlighting experience with large-scale systems, not as pure researchers, reflecting Anthropic's identity as a highly engineered infrastructure company.

marsbit2 hari yang lalu 03:28

What Kind of People Is Anthropic Actually Hiring? Insights from 1,680 Resumes Provide the Answer

marsbit2 hari yang lalu 03:28

Someone Predicts South Korean Stock Market with Hyperliquid, Achieving 74% Accuracy?

A study analyzed whether weekend price movements of four Korean stock perpetual futures contracts (Samsung Electronics, SK Hynix, Hyundai Motor, and EWY) on Hyperliquid could predict their Monday opening directions on their respective primary exchanges (KRX, NYSE). Over 62 weekend observations across the four assets, Hyperliquid correctly predicted the Monday opening direction 45 times (73.8% accuracy). However, performance varied significantly. Samsung Electronics showed the strongest and statistically significant signal, with Hyperliquid's weekend close correctly predicting its KRX Monday open in 15 out of 16 cases (94% accuracy, p-value < 0.001). This signal remained strong (75% accuracy) even when using Saturday's close instead of Sunday's, suggesting genuine price discovery beyond last-minute convergence. Hyundai Motor also showed high accuracy (81%, 13/16 correct), but this was not statistically significant after accounting for a baseline downward bias in its Monday opens. SK Hynix performed marginally better than a coin flip (63%, 10/16). EWY performed the worst (54%, 7/13), underperforming a simple strategy of always predicting a Monday rise. The stark difference between Samsung and EWY is largely attributed to market timing. KRX opens shortly after Hyperliquid's Sunday close, while NYSE opens ~14 hours later, allowing new information to flow in. The results suggest that for assets like Samsung Electronics, where weekend trading on Hyperliquid precedes the primary market open by only minutes, the platform can provide a valuable predictive signal worth monitoring before the Monday auction, despite the currently small weekend trading volumes.

Foresight News06/10 10:08

Someone Predicts South Korean Stock Market with Hyperliquid, Achieving 74% Accuracy?

Foresight News06/10 10:08

Tokens Not Selling? 90% of Crypto Projects Overlook Investor Relations

The article argues that effective Investor Relations (IR) is a critical yet often neglected function for crypto projects, with 90% failing at it and struggling to sell their tokens. Good IR acts as a bridge between a project and the market, broadening the buyer base and improving holder quality. The core of a successful IR strategy is distribution: maximizing the number of target investors who know about the token and converting them into buyers. The two primary buyer types are active crypto funds (requiring clear narratives and data for value reassessment) and large strategic institutions (requiring a long B2B sales cycle). The author emphasizes the necessity of proactively controlling the project's narrative with honesty and context, rather than remaining silent. A major tactical error is poor planning for token unlocks; teams should start 30-50 weeks in advance to manage supply and demand. Data is presented as the best ally for building a compelling story, providing context and comparisons for investors. The author contends that crypto IR should not be a dry, compliance-driven task but an engaging, interactive process similar to modern marketing. To lower the barrier to entry, projects must provide ample public data and research, making it easier for funds to conduct due diligence. Furthermore, the article highlights the power of on-chain data for deep investor analysis and argues that greater transparency, not less, actually expands the market by reducing uncertainty. Success should be measured by improvements in investor base quality and breadth—such as growth in target investors and holder diversification—rather than just token price. The future of IR is envisioned as dynamic, multimedia-rich, and proactive, leveraging the inherent transparency of crypto to build a larger, more engaged investor community.

marsbit03/17 13:39

Tokens Not Selling? 90% of Crypto Projects Overlook Investor Relations

marsbit03/17 13:39

How to Systematically Track High-Win-Rate Addresses on Polymarket?

This article explores methods to systematically identify and track high-success-rate addresses on Polymarket, a blockchain-based prediction market where all transactions are publicly recorded on-chain. It highlights that while data is transparent, the key challenge lies in extracting meaningful signals from vast datasets to detect addresses with potential informational advantages. The piece outlines common characteristics of such addresses: new wallets making large, concentrated bets; specialization in specific verticals; abnormal changes in position size; and exceptionally precise timing, repeatedly entering positions hours before major news breaks. A three-step systematic approach is recommended: First, filter addresses based on sustained profitability (e.g., 30-day positive returns, >55% win rate) using leaderboards like Polymarket Analytics. Second, analyze their holdings in specific event markets, focusing on addresses that are consistently among the top holders before full market pricing. Third, scrutinize their on-chain behavior: entry timing relative to news, position-building patterns (e.g., rapid, concentrated entries), holding periods, and trading focus. Advanced strategies include monitoring exit behavior (e.g., large, unexplained sell-offs), conducting wallet clustering analysis to find linked addresses, tracking unusual volume spikes in low-liquidity markets, and cross-referencing on-chain activity with external real-world data for validation. The goal is to move beyond luck and identify addresses exhibiting repeatable, information-driven advantages.

marsbit03/02 11:35

How to Systematically Track High-Win-Rate Addresses on Polymarket?

marsbit03/02 11:35

Dissecting 290,000 Data Points: We Uncovered 6 Secrets of Polymarket's Liquidity

Based on an analysis of 295,000 markets on Polymarket, this investigation uncovers six key truths about its liquidity. A significant finding is that 67.7% of markets have a lifespan of less than 7 days, with 63.16% of current short-term markets having zero trading volume, resembling the high failure rate of meme coins. These short-term markets, dominated by sports and crypto predictions, suffer from extremely low liquidity, often under $100. In contrast, long-term markets (over 30 days), though fewer in number, attract substantial capital, with an average liquidity of $450,000. U.S. politics is the most capitalized category. The platform exhibits a stark divide: sports markets are either ultra-short-term with high volume or long-term "season bets," with mid-term interest lacking. New, complex markets like U.S. real estate face a "cold start" problem due to high expertise requirements and low volatility, deterring participation. The market is highly polarized; a tiny fraction of high-value contracts (1,000+ with over $10M volume) capture 47% of all trading volume, while the vast majority of markets are illiquid. Finally, the "Geopolitics" category is the fastest-growing, indicating rising user interest. The core insight is that liquidity in prediction markets is not evenly distributed but concentrates around events that offer either instant gratification (sports/crypto) or deep macro bets (politics), transforming Polymarket into a specialized financial tool rather than a universal prediction platform.

比推01/08 08:17