Quant Letter: October 2023, Week 3

Weekly quantitative finance newsletter

Oct 18, 2023

If you are relying on influencers or academics with a marketing budget to show you the ‘latest’ techniques in quantitative finance, you are missing out on 99% of all quantitative finance research.

To get information directly from thousands of academic-practitioners at blog.ml-quant.com we are plugged in directly to the source, e.g., ArXiv, GitHub, and SSRN.

You will most likely be able to expense this newsletter depending on your industry and role. Thanks to reader support from DE Shaw, Arrowstreet, AQR, MAN, and Blackrock, I will soon be adding a modern react website.

Its amazing how little the following papers are publicly discussed, I will just add them here for those who want to click quickly.

Few-Shot Learning Patterns in Financial Time-Series for Trend-Following Strategies
Leveraging Large Language Model for Automatic Evolving of Industrial Data-Centric R&D Cycle
Deep Structural Model: A Fusion of Machine Learning and Theory
Neural Tangent Kernel in Implied Volatility Forecasting: A Nonlinear Functional Autoregression Approach
Can GPT models be Financial Analysts? An Evaluation of ChatGPT and GPT-4 on mock CFA Exams
Observations Concerning the Estimation of Hestons’ Stochastic Volatility Model Using HF Data

As of this month, we also offer a group/corporate subscription discount of 35% if more than 2 readers join at once. (see the coupon above)

SSRN

Quantitative

Neural Tangent Kernel for Nonlinear Implied Volatility Forecasting: The study proposes a Nonlinear Functional Autoregression framework for forecasting implied volatility in financial markets, proving its effectiveness in predicting the S&P 500 Index from 2009 to 2021. (2023-08-28, shares: 2.0)

Deep Structural Model: ML and Theory Integration: The article introduces a new model that merges deep learning and structural models for better prediction of equity returns and covariances, leading to higher returns and sharpe ratios. (2023-09-21, shares: 3.0)

Interest Rate Risk Management Measurement by Financial Institutions: A new method measuring financial intermediaries' residual interest rate risk found that U.S. life insurers are more sensitive to changes in long-term interest rates than property and casualty insurers. (2023-08-31, shares: 9.0)

Liquidity Mismatch in Payment Risk and Bank Lending: Banks are at risk of liquidity issues due to unpredictable payment flows, which can hinder lending and monetary policy, particularly during financial crises. (2021-11-08, shares: 2.0)

Hedge Fund Performance and Managerial Structure: Solo-managed hedge funds perform better than team-managed ones in terms of abnormal returns and market volatility skills, but they also have higher idiosyncratic and tail risk, and are less likely to be liquidated. (2023-10-06, shares: 2.0)

Sentiment Analysis of Amazon Products using Hybrid Model: Sentiment Analysis uses social media comments to understand people's opinions on various subjects, assisting in decision-making processes like product purchases or investments. (2023-05-27, shares: 3.0)

Financial

ESG Fund Greenwashing Uncovered: The study reveals that funds focusing on climate change or decarbonisation strategies do not significantly reduce their portfolio's carbon intensity, suggesting potential greenwashing in responsible investing. (2022-04-11, shares: 550.0)

FOMC Announcements and Volume Dynamics: The research shows that stock market volume decreases before FOMC announcements and increases afterwards due to discretionary liquidity trading and information asymmetry, especially for stocks with higher market betas. (2021-01-07, shares: 75.0)

Managerial Reporting Quality and Investment Efficiency: The research examines the effect of enhanced managerial reporting quality on investment decisions, concluding that it improves investment efficiency mainly by reducing internal information asymmetry. (2022-12-14, shares: 2.0)

Uncertainty Diversification's Misallocation Effects: The research highlights a paradox in corporate risk management, where firms' diversification efforts to hedge against risk result in reduced growth, equity value, and inefficient resource allocation. (2023-09-20, shares: 2.0)

Expected Returns from Information-Driven Volatility: The paper presents an asset pricing model that links stock market volatility to information, indicating a negative relationship between past volatility and future expected returns. (2021-11-14, shares: 2.0)

Factor Models for Chinese A-Shares: Performance Comparison: The study evaluates asset pricing models in the Chinese Ashares market, concluding that a modified FamaFrench sixfactor model and a fourfactor model perform best, especially when transaction costs are factored in. (2021-09-10, shares: 2.0)

Access to Finance Bureaucracy and Capital Allocation Efficiency: The research explores capital allocation efficiency in developing countries, revealing that factors like limited finance access, bureaucracy, information asymmetry, and gender inequality obstruct efficient capital allocation. (2021-03-17, shares: 2.0)

Determinants of Portfolio ESG Performance: Attribution Framework: The study introduces a framework for assessing the ESG performance of a portfolio, indicating that U.S. public pensions' positive ESG performance is largely due to their underlying holdings. (2023-06-07, shares: 2.0)

Recently Published

Quantitative

Tensor Mutual Information for Clustering: Researchers suggest new methods for mutual information of tensors, applying it to k-means algorithm in 2D and 3D datasets. (2023-10-13, shares: 3.0)

Detecting and Predicting Structural Changes: The article highlights the role of data science in identifying patterns for decision-making, and how changes like economic and political shocks can alter these patterns. (2023-10-16, shares: 2.0)

Size Reduction in ML Models with Pyramid Training: A study presents a method to reduce the size of machine learning models, making research more accessible for those with limited hardware. (2023-10-17, shares: 2.0)

Privacy-Preserving K-Means Clustering Algorithm: The paper introduces a privacy-preserving k-means clustering algorithm for data mining, ensuring data privacy while using cloud servers. (2023-10-16, shares: 2.0)

Machine Learning for Big Data Analysis: A survey reviews different machine learning techniques and their uses, emphasizing the need for appropriate technique selection for data analysis. (2023-10-13, shares: 2.0)

Financial

Estimating Heston's Volatility Model with HF Data: A study on the Heston stochastic volatility model shows challenges in producing accurate parameter estimates and limitations in capturing tail behavior with high-frequency financial data. (2023-10-13, shares: 2.0)

Multivariate Score-Driven Filter: New: A new asset pricing model, the generalized ACB, is introduced, enhancing the autoregressive conditional beta model by driving dynamic interaction effects among beta coefficients. (2023-10-13, shares: 5.0)

ESG Risk Premia Measured with Contingent Claims: The study suggests a method to calculate ESG risk premiums using market data, revealing a significant link between ESG scores and asset process volatility and drift terms, indicating ESG factors impact company value. (2023-10-12, shares: 16.0)

Corporate Sector in Preferred-Habitat Model: The research introduces a model where the same investor prices government and corporate bonds, causing credit spread variations beyond credit quality changes, driven by credit quality variation, risk-neutral correlation of risk factors, and portfolio rebalancing. (2023-10-16, shares: 5.0)

Lessons from Silent Sustainable Investing: Talk vs. Walk: The study finds that investors favor internal ESG labels over external sustainability ratings, significantly influencing fund flows, but high sustainability rating funds don't necessarily rebrand as ESG funds due to regulatory pressure and limited benefits. (2023-10-14, shares: 15.0)

Pricing Perpetual Futures: The paper offers explicit formulas for the no-arbitrage price of various perpetual contracts, demonstrating that the futures price is determined by the risk-neutral expectation of the spot sampled at a random time reflecting the intensity of the price anchoring. (2023-10-16, shares: 4.0)

Treasury Market Liquidity Evolution in 2023: The article discloses that liquidity conditions in the U.S. Treasury securities market deteriorated suddenly in March 2023 following two banks' failures, but quickly recovered, with liquidity continuing to closely follow the level expected by the path of interest rate volatility. (2023-10-17, shares: 6.0)

Unique Asset Class: Wine: Research indicates that adding a unique South African wine index to a portfolio can enhance risk reduction and diversification. (2023-10-13, shares: 2.0)

Finance's Upcoming Revolution: The article suggests that improved assessment of required risk premia and fundamental values can significantly enhance the efficiency of financial markets. (2023-10-13, shares: 2.0)

ArXiv

Finance

Stat Arb Portfolio Construction: The article suggests a new method for portfolio construction using preference relation graphs, which can improve the performance of statistical arbitrage methods by reconciling contradictory trading signals. (2023-10-12, shares: 6)

Robust Trading in Lattice Market: The article presents a new robust trading paradigm, multi-double linear policies, within a generalized lattice market model, demonstrating its effectiveness using data from the top 30 S&P 500 companies. (2023-10-17, shares: 6)

Uncovering Market Disorder & Liquidity Trends Detection: The article proposes a new method to detect significant changes in liquidity in an order-driven market using a market liquidity model and Marked Hawkes processes, validated with real market data. (2023-10-13, shares: 3)

Early Warning Signals for Bank Crisis: The research introduces a method using high-frequency data to detect early signs of potential bank crises, proving that certain indicators can predict periods of high volatility in banking. (2023-10-16, shares: 3)

Few-Shot Learning for Time-Series Forecasting: The article introduces X-Trend, a new time-series trend-following forecaster that adapts quickly to market changes, showing improved performance and faster recovery from the COVID-19 downturn than other models. (2023-10-16, shares: 3)

Mean-field Libor Market Model & Valuation of LT Guarantees: The article introduces a numerical asset-liability management model based on the multi-dimensional mean-field Libor market model, which can calculate future discretionary benefits in line with Solvency II regulation, using public life insurance data. (2023-10-13, shares: 5)

Generalizing the Rough Heston Approximation: The research expands the rational approximation of the rough Heston fractional ODE solution to the Mittag-Leffler kernel case, demonstrating numerical proof of the solution's convergence. (2023-10-13, shares: 2)

Importance of Resolution in Energy System Models: The study emphasizes the significance of structural aspects in energy system models over input uncertainties, suggesting that temporal resolution greatly impacts results and should be prioritized over spatial resolution. (2023-10-16, shares: 2)

Miscellaneous

GPT Models for Financial Analysis: The study assesses the financial understanding of Large Language Models (LLMs) like ChatGPT and GPT-4 using CFA Program mock exam questions to improve their use in finance. (2023-10-12, shares: 8)

Large Language Models for R&D Evolution: The paper investigates the role of Large Language Models (LLMs) in speeding up data-centric R&D evolution, using quantitative investment research as a case study, showing promising results on the open-source research platform, Qlib. (2023-10-17, shares: 4)

Reducing Hallucination in Financial Report Extraction: The paper showcases the use of Large Language Models (LLMs) for precise and efficient extraction of information from the Q&A section of company financial reports, proving their method's superiority through various metrics. (2023-10-16, shares: 2)

EfficientOCR: Open-Source Text Digitization: EffOCR, a new open-source OCR package, is presented as an efficient, cost-effective solution for digitizing large amounts of public domain documents, successfully applied to historical U.S. newspapers and Japanese documents. (2023-10-16, shares: 5)

Economics

AI Enhances Decision-Making: A study reveals that humans enhance their Go game skills after learning from an AI-powered program, with younger players and those from AI-exposed countries showing more improvement. (2023-10-12, shares: 4)

Trade Sanctions & Global Commerce Shifts: A study examines how sanctions affect a country's global economic standing, suggesting sanctioned countries may redirect trade to neutral or friendly nations, forming an alternative 'core'. (2023-10-12, shares: 4)

Biased Beliefs & Belief Formation: A proposed model indicates that agents are biased towards theories with strong supporting evidence and weak opposing evidence, causing skewed beliefs among less sophisticated agents. (2023-10-12, shares: 5)

Survival of MaaS Platforms: A proposed pricing model for Mobility-as-a-Service platforms could benefit all parties involved - the platform, travelers, and transit operators. (2023-10-12, shares: 4)

Crypto & Blockchain

Blockchain Protocol Risk Management: The article introduces a risk management framework for blockchain protocol risks, created with financial institutions, blockchain developers, and regulators, to guide DLT adoption in finance. (2023-10-16, shares: 7)

DeFi Market Invariants and Arbitrage: The study establishes that every DeFi market has an invariant linked to arbitrage and completeness, offering a condition for no-arbitrage and completeness in nonlinear DeFi theory. (2023-10-15, shares: 5)

Bitcoin Economic Analysis: The research conducts an economic analysis of Bitcoin, examining its impact, transaction volume, market capitalization, mining activities, and effects on traditional finance. (2023-10-15, shares: 3)

Crypto-Asset Regulation NLP: The paper explores the use of Natural Language Processing in analyzing crypto-asset white papers for regulatory compliance under the EU's crypto-asset regulation, discussing the potential and challenges of this integration. (2023-10-16, shares: 3)

Historical Trending

Latent Socioeconomic Health Index: The study develops a LAtent Causal Socioeconomic Health index at a national level, using spatial and statistical causal modeling to understand the impact of an observational variable on a spatially correlated latent trait. (2020-09-24, shares: 43)

Big Winners' Impact on Investment Strategies: The research examines the effect of big winner stocks on investment strategies, suggesting that passive investing is more likely to achieve long-term financial goals due to the risk of active managers missing substantial returns. (2022-10-17, shares: 25)

Supercharged RQMC vs. QMC: The research compares two randomizing methods in Quasi Monte Carlo applications, finding the Randomized QMC method to be superior in performance with better convergence rates and error bounds. (2022-10-29, shares: 67)

Fast Option Pricing: The study introduces faster and more energy-efficient algorithms for option pricing under various models, outperforming existing methods. (2023-03-04, shares: 23)

Liquidity in Bond Markets: The article suggests a new method for valuing illiquid securities using the concept of micro-price and Fair Transfer Price, applicable in illiquid or one-sided markets. (2023-09-08, shares: 21)

Investment in Financial Markets: The paper presents a solution for the multi-asset Merton investment problem under drift uncertainty, exploring the influence of drift distribution and ambiguity preferences on investment strategies. (2023-03-15, shares: 20)

Nested Simulation for Estimating Functionals: The paper presents a new method using kernel ridge regression for nested simulation, which improves the convergence rate in high-dimensional problems, given a sufficiently smooth conditional expectation. (2022-01-09, shares: 16)

RePec

Finance

Algorithmic Trading: Profitability and Behavior: The research finds that algorithmic traders profit while non-algorithmic traders lose, with market volatility influencing their trading behaviors differently. (2023-10-18, shares: 29.0)

Virtual Digital Asset Volatility Analysis: Digital assets offer higher average returns than equity and commodities, but are more susceptible to short-term external shocks, according to a study. (2022-09-16, shares: 28.0)

Dynamic Bond Portfolio Optimization: The paper introduces a new framework for dynamic bond portfolio optimization over multiple periods, proving it to be more effective than single-period optimization. (2023-10-18, shares: 26.0)

Multiperiod Portfolio Allocation: The research finds that considering volatility clustering reduces hedging demands in dynamic multiperiod portfolio choices, while non-normalities have minor effects on allocations. (2023-10-18, shares: 23.0)

fBetas and Portfolio Optimization: The article introduces a new method for portfolio optimization using f-divergence induced coherent risk measures, showcasing its effectiveness through numerical experiments with selected stocks against the S&P 500 market index. (2023-10-18, shares: 18.0)

HighFrequency Alternative Data for Japan's GDP Nowcasts: The study reveals that credit card data can enhance the accuracy of real-time nowcasting of Japan's GDP, particularly in the early stages of forecasting when traditional data is unavailable. (2023-10-18, shares: 15.0)

HighDimensional Portfolio Optimization with Tree-Structured Factor Model: The paper proposes a new portfolio optimization method that uses multiple characteristic information to predict stock returns and risk exposures, demonstrating its effectiveness in achieving higher Sharpe ratios, smaller standard deviations, and lower turnover. (2023-10-18, shares: 15.0)

Performance of ESG ETFs in the U.S.: The research finds that ESG equity ETFs in the U.S. generally outperform the S&P 500 Index, with approximately 16% providing positive and significant alphas. (2023-10-18, shares: 15.0)

GitHub

Finance

Quant Finance Resources: Resources for Quantitative Finance lists various tools and resources for conducting quantitative financial analysis. (2018-07-09, shares: 464.0)

abess: Fast BestSubset Selection: The piece presents a library designed for efficient selection of the best subset. (2020-12-20, shares: 320.0)

Python Data Exploration: A nextgeneration Python notebook explore data build tools deploy apps introduces a new Python notebook for data analysis, tool creation, and app development. (2023-08-14, shares: 115.0)

Data Labeling Tool: Label Studio is a multitype data labeling and annotation tool with standardized output format presents Label Studio as a tool for data labeling and annotation with a uniform output format. (2019-06-19, shares: 14463.0)

ML Engineering Guides: Machine Learning Engineering Guides and Tools provides a range of guides and tools for machine learning engineering. (2020-09-02, shares: 1156.0)

Custom Instructions for ChatGPT: 🧠💬 Supercharged Custom Instructions for ChatGPT noncoding and ChatGPT Advanced Data Analysis coding presents improved custom instructions for non-coding and advanced data analysis using ChatGPT. (2023-09-29, shares: 1558.0)

Informative

Analyzing Digital Filters for Financial Time Series: Henry Stern's article explores four digital filters used in financial time series analysis for trend-following, such as moving average and exponential smoothing. (2023-10-17, shares: 1.0)

ADGM Academy's Abu Dhabi ML Meet-up: The fourth Abu Dhabi Machine Learning Meet-up was held at the ADGM Academy Research Centre, where speakers discussed the latest developments in machine learning and its industry impact. (2023-10-17, shares: 1.0)

CPI Inflation Components Forecasting with HRNN: A Hierarchical Recurrent Neural Network (HRNN) model has been created to predict disaggregated inflation components of the Consumer Price Index (CPI), aiding policy and market makers. (2023-10-12, shares: 1.0)

VisualHFT: Enhancing the Open-Source Project: The open-source project VisualHFT has introduced new features for real-time visibility into algorithmic trading activities, easier integration with multiple exchange data sources, and tools to manage increasing trading volumes and data. (2023-10-14, shares: 2.0)

Podcasts

Quantitative

Hedging in Investments: In an Interactive Brokers podcast, Cassidy Clement and Kevin Davitt discuss hedging as a method to reduce investment risks. (2023-10-13, shares: 13)

Transportation Forecasting & Emerging Technologies: A podcast discusses the use of real-time data, GPS, and advanced algorithms, including AI and machine learning, in modern transportation forecasting. (2023-10-17, shares: 6)

Five Equity Market Trading Sessions: In a podcast with IBKR’s Jeff Praissman, Market Chameleon cofounders discuss the five trading sessions in the equity market. (2023-10-12, shares: 6)

Charlie McElligott on Positioning, Flow & Market Correction: In a MacroHive podcast, Charlie McElligott of Nomura Securities International discusses the effect of low real yields on recent market correction. (2023-10-13, shares: 3)

Buy vs. Build Debate: Matt Barrett, CEO of Adaptive Financial Consulting, discusses the pros and cons of buying versus building trading systems and the future of cloud-based trading technology in a podcast. (2023-10-13, shares: 3)

Risks in US CRE: Scott Rechler and Stijn Van Nieuwerburgh analyze the risks associated with the commercial real estate crisis, influenced by rising interest rates and reduced lending from regional banks. (2023-10-17, shares: 2)

Barclays PLC's Past Strategies: Ian Jones' thesis explores how Barclays Group Archives have helped Barclays PLC achieve its strategic goals, particularly during 2012-2015 when the bank used its history to manage and recover from scandals. (2023-10-17, shares: 1)

News

Quantitative

Bloomberg's Portfolio Manager Workspace: Bloomberg has launched Portfolio Manager Workspace, a tool that combines portfolio data with industry data, risk analytics, and liquidity insights in real time. (2023-10-12, shares: 8)

ExMan Group Launches Hedge Fund: Priya Kodeeswaran, formerly of Man Group and Millennium Management, is set to establish and lead his own hedge fund, Katamaran Capital, as Founder and Chief Investment Officer. (2023-10-16, shares: 5)

SEC tightens hedge fund reporting: The US SEC has introduced new regulations requiring hedge funds to disclose more details about their short-selling activities. (2023-10-16, shares: 5)

Dynamic Funds introduces credit opportunities fund: Dynamic Funds has launched the Dynamic Credit Opportunities Fund, aiming to generate returns through long and short positions in primarily corporate debt securities. (2023-10-12, shares: 4)

Credit Suisse Quant MD survives UBS cull: UBS has hired the MD responsible for trading products in London. (2023-10-17, shares: 3)

Miscellaneous

Acadian joins BNY Mellon's Middle Office platform: Acadian Asset Management, a Boston-based firm with $100bn in assets, has begun using BNY Mellon’s Middle Office services platform. (2023-10-16, shares: 3)

Bank of America trading fees rise 8% after market push: Bank of America's trading revenue has risen following investment, however, dealmaking fees are still underperforming. (2023-10-17, shares: 3)

Trumid hires ex-Millennium PM as Global Sales Head: Fintech firm Trumid has hired Jason Quinn, ex-portfolio manager at Millennium Management, as its Global Head of Sales. (2023-10-17, shares: 3)

FN100 unveils influential women in finance 2023: FN has published its yearly ranking of the most influential women in European finance. (2023-10-15, shares: 2)

Twitter

Quantitative

SP Volatility Analysis: Article 1: A new method breaks down SP500 volatility into long, medium, and short-term elements, uncovering hidden predictive factors. (2023-10-17, shares: 2)

Llama Language Models for Math: Article 2: Llemma's open language models, using Llama 7Band and 34B parameters, show advanced mathematical abilities suitable for various tasks in math and financial market modeling. (2023-10-17, shares: 1)

LLMTime: Time Series Forecasting: Article 3: LLMTime, a large language model, translates numbers into text and predicts possible extrapolations for time series forecasting. (2023-10-15, shares: 1)

Equity Return Sign Predictability: Article 4: A recent study proposes a long-short strategy based on future positive return probabilities, resulting in higher Sharpe ratios and fewer crashes than the conventional momentum strategy. (2023-10-14, shares: 1)

Miscellaneous

Bond Strategies: Costs and Market Impact: The article explores the costs and market effects of strategies that involve only buying and holding corporate bonds. (2023-10-16, shares: 0)

Factor Investing: Can it be Scientific?: Marcos M. López de Prado's article discusses the possibility of making factor investing a scientific process. (2023-10-15, shares: 0)

Multimodal GPT Models: Transfer Learning Unlocked: Microsoft's paper suggests that multimodal GPT models could enhance immersive experiences by enabling transfer learning from visuals to text. (2023-10-14, shares: 0)

New TS Forecasting Paper: Introducing 2 Model Architectures: A new research paper presents a new benchmark for timeseries forecasting and two new model architectures, and discusses the limitations of popular models like DeepAR and LSTM. (2023-10-12, shares: 0)

Wall Street Bets: Impact of Social Media: The article investigates the influence of Wall Street Bets and social media on the financial market. (2023-10-12, shares: 0)

Videos

Quantitative

Bayesian Inference on Probabilistic Surveys: The University of Chicago Booth School of Business held a discussion on probabilistic surveys and inflation risk market, with speakers from different universities and financial institutions. (2023-10-16, shares: 1.0)

Overcoming Mental Blocks in Quantitative Finance: The article offers methods to tackle complex quantitative finance issues, such as taking breaks, online research, peer consultation, and textbook study. (2023-10-15, shares: 33.0)

Rising

Cointegration Definitions: Avellaneda and Lee's paper presents a unique interpretation of cointegration in price time series. (2023-10-12, shares: 7.0)

Renaissance and Medallion Fund Janitors: The article investigates the potential of entering a profession via janitorial work. (2023-10-12, shares: 222.0)

Trinomial Tree for Collecting Greeks Data: The author gathers options greeks data using a trinomial tree and threading, but faces a possible 15-minute delay due to Yahoo Finance's API. (2023-10-13, shares: 42.0)

Paper with Code

ArXiv ML

Historical Trending

Quasi-Arithmetic Mixtures and Minimization: The research analyzes the 'centroid' property using Bregman divergences under a monotonic embedding of the density function, linking common divergences with intermediate densities and highlighting the relationship between parametric families, quasi-arithmetic means, and divergence functions. (2022-09-15, shares: 92)

Feature Learning in Infinite-Depth Neural Networks: The research explores depthwise parametrizations in deep residual networks, identifying Depth-$\mu$P as the best parametrization for feature learning and diversity, but notes limitations in deeper blocks. (2023-10-03, shares: 53)

SelfCheckGPT: Hallucination Detection for Language Models: The study introduces SelfCheckGPT, a method for fact-checking black-box model responses without an external database, showing its ability to detect and rank factual and non-factual sentences, and outperforming grey-box methods in hallucination detection and factuality assessment. (2023-03-15, shares: 99)

StoryBench: Continuous Story Visualization Benchmark: The article presents StoryBench, a new benchmark for evaluating text-to-video models, including three video generation tasks and guidelines for human evaluation of video stories, emphasizing the need for improved automatic metrics for video generation. (2023-08-22, shares: 62)

Refer and Ground Anything Anywhere at Any Granularity: Ferret, a new Multimodal Large Language Model, is introduced, excelling in understanding spatial references in images and grounding descriptions, surpassing existing models in multimodal chatting. (2023-10-11, shares: 52)

Recently Published

Llemma: Open Math Language Model: The article discusses Llemma, a superior language model for mathematics that can prove theorems without additional fine-tuning. (2023-10-16, shares: 434)

OpenAgents: Language Agent Platform: OpenAgents, a platform for utilizing and developing language agents in daily life, is introduced, providing a user-friendly interface and a basis for future research. (2023-10-16, shares: 62)

MemGPT: LLMs as OS: The article suggests MemGPT, a virtual context management method that enhances the context window of large language models, improving document analysis and multi-session chat performance. (2023-10-12, shares: 33)

In-Context Pretraining: Modeling Beyond Documents: The paper introduces In-Context Pretraining, a novel method that improves language models' performance by pretraining them on related documents, enhancing their contextual reasoning skills. (2023-10-16, shares: 19)

Real-time Scene Representation and Rendering: The 4DGS model is introduced, capable of reconstructing dynamic 3D scenes from 2D images and generating diverse views over time, providing real-time rendering efficiency. (2023-10-16, shares: 16)

Task Planning with Language Models: A new framework is proposed that uses language models for interactive task planning, allowing for easy adaptation to different tasks and precise replanning based on new user requests. (2023-10-16, shares: 11)

Transformers as Decision Makers in Reinforcement Learning: A theoretical framework is provided that examines supervised pretraining for in-context reinforcement learning, offering insights into training transformers for reinforcement learning algorithms and the impact of distribution mismatch in offline training data. (2023-10-12, shares: 10)

Machine Learning & Quant Finance

Discussion about this post

Machine Learning & Quant Finance

Quant Letter: October 2023, Week 3

Weekly quantitative finance newsletter

SSRN

Recently Updated

Quantitative

Financial

Recently Published

Quantitative

Financial

ArXiv

Finance

Miscellaneous

Economics

Crypto & Blockchain

Historical Trending

RePec

Finance

GitHub

Finance

LinkedIn

Trending

Informative

Podcasts

Quantitative

Related

News

Quantitative

Miscellaneous

Twitter

Quantitative

Miscellaneous

Videos

Quantitative

Reddit

Rising

Paper with Code

Trending

ArXiv ML

Historical Trending

Recently Published

Discussion about this post