Quantide Loop: Step 5 MoE, SHFE Options, and AQR Factor Insights
Trending Topics
StepFun Releases Step 5 Preview: 70 Days Free Access
Promotional Offer: Log in to receive 15 days; make your first API call to earn another 15 days; invite one friend to register for an additional 15 days. Maximum cumulative bonus: 45 days, totaling 75 days of free access.
The latest model employs a sparse Mixture-of-Experts (MoE) architecture with 600 billion total parameters, activating 27 billion parameters per token, and supporting a 1 million token context window. Internal benchmarks indicate that key performance metrics meet or exceed those of glm-5.3 and kimi-k3. 1
SHFE Lists Three New Option Contracts, Totaling 23 Products
On September 10, 2026, the Shanghai Futures Exchange (SHFE) listed options for hot-rolled coil, stainless steel, and low-sulfur fuel oil, simultaneously opening access to qualified foreign investors. Notably, low-sulfur fuel oil options were directly introduced to overseas traders. In the first half of 2026, SHFE options averaged 2.2168 million contracts in daily trading volume (up 63.74% year-over-year) and 1.6284 million contracts in daily open interest (up 13.98% year-over-year). This listing addresses gaps in tools for the finished steel segment of the black metal industry chain, enabling industrial enterprises to build closed-loop strategies covering production margins, inventory management, and sales. 2
August Social Financing Surges; M2 Grows 7.5% YoY
The People's Bank of China (PBOC) released financial statistics for August 2026, showing a social financing increment of 4.648 trillion yuan. Specifically, RMB loans to the real economy totaled 2.7863 trillion yuan (up 5% YoY), while foreign currency loans reached 123 billion yuan (up 3.1% YoY). Broad money (M2) stood at 35.681 trillion yuan (up 7.5% YoY), and narrow money (M1) was 11.577 trillion yuan (up 4.1% YoY). 3
Academic Research
Why Classic Factor Strategies Remain Resilient
Challenging the notion that strategies fail once widely known, AQR research explores the mechanisms sustaining the long-term effectiveness of classic factor strategies such as value, momentum, arbitrage, and quality. The study validates the robustness of simple indicator-based factor strategies through out-of-sample tests across time, geography, and asset classes, employing conservative assumptions to assess tail risks. Although specific quantitative results are not disclosed, the authors indicate that these strategies maintain a statistical win rate slightly above failure rates and will likely remain effective, albeit with potentially lower returns and altered risk profiles. The key implication is that these strategies offer low correlation with mainstream portfolios, serving as alternative sources of risk premium. However, investors must acquire them at reasonable fees and monitor for behavioral shifts. 4
Covered Calls Uncovered
AQR research highlights that covered calls not only capture equity risk premiums but also earn volatility risk premiums by selling options. Furthermore, option convexity embeds an active equity exposure akin to mean reversion strategies. The paper proposes a novel performance attribution framework that decomposes returns into passive/active equity components and short volatility exposure.
The study recommends constructing risk-managed covered calls by hedging uncompensated active equity exposures. It claims this approach maintains similar expected returns while reducing risk, downside beta, and improving the Sharpe ratio. 5
Industry News
AQR Establishes Engineering Center in Bangalore, Plans to Hire 150 Quant Engineers
The new office will collaborate with the U.S. engineering team to develop cloud-based trading systems, data visualization tools, portfolio analytics, and microservices to support front-office operations. The team is led by Chrys Kattirisetti, former Director of D.E. Shaw India, who brings 20 years of front-end development experience and will focus on team expansion. This move aims to leverage local tech talent to strengthen AQR’s quantitative investing infrastructure, potentially intensifying global competition for quant talent. 6
Vals Raises $40 Million Series A Led by a16z to Set AI Benchmarking Standards
Founded in 2024, Vals was co-founded by 25-year-old Rayan Krishnan, who previously worked at Microsoft and the Stanford AI Lab.
Seed investors included 8VC and Bloomberg Beta. To prevent models from overfitting to specific test questions, Vals does not publicly release its test items. Instead, it evaluates models based on their actual output across tasks in law, finance, and programming. Revenue has grown eightfold year-over-year, and the team has expanded from 8 to 25 members. The company has also launched evaluation programs for federal agencies. As AI companies continue to go public, Vals’ evaluation results are expected to become central documents in public filings. 7
Quant Life
BlackRock and Citi Veteran Joins CMU to Guide Students into Quant Roles
Shelli Faber, a quant veteran with over 30 years at BlackRock and Citi, previously built and led global quant analysis, risk, and technology teams, while also managing recruitment, mentorship, and analyst training programs. She knows precisely what employers look for in quant hires. Now, she brings this expertise to the classroom: helping students with career strategy, networking, interview preparation, and offer evaluation, while expanding connections with quant employers and participating in alumni mentorship and Quant-in-Residence programs.
For those entering or transitioning into quantitative finance, remember this: hiring decisions prioritize not just modeling skills, but project experience and communication abilities validated by industry veterans. Clearly articulating every technical decision on your resume is more effective than grinding through ten extra practice problems. The fact that a 30-year quant veteran is mentoring students underscores the significant talent gap and career ceiling in this field. 8
Word count: 1558 / 2400