Quant Letter: November 2024, Week-2
Weekly (70th Edition)
ArXiv
Finance
Estimating Shadow Rate: The research introduces a computational method to calculate the shadow riskless rate (SRR) in a risk-free market, which can help differentiate between investment asset classes. (2024-11-11, shares: 3)
AIAgent Collaboration in Investment Analysis: The study suggests a multi-agent system for financial investment research that performs better than traditional models by adapting to market conditions and optimizing performance. This shows the potential of multi-agent systems in improving financial analysis and investment decisions. (2024-11-07, shares: 3)
Credit Risk with Copulas: The study presents a new model to analyze large losses from credit portfolio defaults using the Archimedean copula family and two algorithms that improve traditional Monte Carlo methods. (2024-11-11, shares: 3)
ajdmom: Python AJD Moment Package: The article presents ajdmom, a Python package that automatically generates moment formulas for affine jump diffusion processes, useful for sensitivity analysis and can be downloaded from GitHub or the Python package index. (2024-11-10, shares: 3)
Improving FX Volatility Imputation: The study explores enhancing the prediction of missing implied volatilities in FX options using modified variational autoencoders (VAEs), which better manage data uncertainty. (2024-11-08, shares: 3)
Optimal Liquidation in HFT: The research introduces a high-frequency trading model using a hidden Markov process to examine optimal liquidation strategies under limited information, offering a practical algorithm to simulate the original liquidation issue. (2024-11-07, shares: 3)
Equilibrium Relationships in Models: The paper establishes that a two-cycle equilibrium in a model with infinitely-lived agents can also exist in an overlapping generations (OLG) model, indicating that both models can experience equilibrium indeterminacy and rational asset price bubbles. (2024-11-12, shares: 2)
Historical Trending
Random Coefficients in Volatility: The article suggests an arbitrage-free framework for randomizing parameters from the parametric implied volatility formula, improving existing parametrizations and expanding the range of acceptable implied volatilities shapes, proving especially effective in modeling the implied volatility curves of short expiry options before an earnings announcement. (2024-11-06, shares: 5)
Fast Bass Volatility: The paper introduces a new method for the Bass Local Volatility Model that merges local quadratic estimation and lognormal mixture tails for creating state price densities, showing that trapezoidal rule based schemes for numerical convolutions perform better than commonly used Gauss-Hermite quadrature. (2024-11-06, shares: 3)
Testosterone Peaks in Startup Founders: A study of 107 male Y Combinator founders shows a link between testosterone levels and company stage, with testosterone rising by 55.7% from pre-seed to seed funding, peaking at Series B, then dropping by 42.2% after Series B, indicating early startup success boosts confidence, while later-stage pressures increase stress. (2024-11-05, shares: 42)
SSRN
Recently Published
SafeHaven Assets in COVID-19: The COVID-19 pandemic showed that gold, Treasury Bonds, and the Euro were effective safe haven assets in reducing risk during the equities bear market. (2024-11-08, shares: 6.0)
Market Beta Benchmarking: The article introduces a new method for comparing market beta estimates to unobserved true betas, applicable to any beta estimate and requiring minimal assumptions about the true asset pricing model. (2024-11-07, shares: 5.0)
LongRun Disaster Risk Models: The article reviews literature on long-term risk and rare disaster risk models in asset pricing, introducing new methods to address criticisms and explain the influence of climate change on asset prices. (2024-11-07, shares: 3.0)
Recently Updated
Statistical Arbitrage with Spectral Clustering: The paper proposes a method for identifying similar assets using high-frequency intraday data and tests various trading rules based on these asset clusters. (2024-09-30, shares: 7.0)
Trend-Following Beta Replication: The article discusses the challenges of trendfollowing investment strategies and suggests mitigating them by replicating a broad index of such funds, despite potential tracking errors. (2024-10-15, shares: 205.0)
Machine Learning in Finance: The paper reviews machine learning methods in portfolio management, discussing their limitations, potential future developments, and applications in systematic trading strategies. (2024-10-15, shares: 8.0)
Optimal Hedge Fund Allocation: The research justifies significant allocations to hedge funds due to their diversification benefits, particularly equity and event-driven hedge fund strategies, even without alpha generation. (2024-10-14, shares: 7.0)
Strategic Trading in Bank Portfolios: The study reveals that US banks are reluctant to sell underwater bonds at a discount, especially those not recognizing unrealized losses in regulatory capital and banks with low stock market valuations. (2024-10-18, shares: 6.0)
LLMs for Time Series Forecasting: The article evaluates the performance of Large Language Models (LLMs) for stock market forecasting, indicating varying effectiveness in different data scenarios and suggesting the need for further model refinement. (2024-10-15, shares: 21.0)
Fund Flows: Prospectus vs. Ratings: Retail and institutional fund flows are primarily driven by sustainability statements in fund prospectuses, not external sustainability ratings, based on an analysis of over 23,000 equity mutual funds and ETFs. (2024-06-04, shares: 22.0)
ArXiv ML
Recently Published
Building Large Knowledge Bases: A large general-domain knowledge base (GPTKB) can be built entirely from a large language model, containing 105 million triples for over 2.9 million entities. (2024-11-07, shares: 45)
API LLMs Info Leak: Researchers have discovered a method to extract hidden information from large language models like OpenAI's gpt-3.5-turbo using API queries, exploiting a weakness known as the softmax bottleneck. (2024-03-14, shares: 615)
Qwen.5-Coder Report: The Qwen2.5-Coder series, an upgrade from CodeQwen1.5, shows remarkable code generation abilities and achieves top performance in multiple code-related tasks, potentially advancing code intelligence research. (2024-09-18, shares: 121)
RePec
Finance
Clustering for Portfolio Optimization: The article suggests a new investment strategy using clustering to select fewer assets, potentially outperforming traditional equal weight strategies. (2024-11-13, shares: 21.0)
Peak Price Tracking for Portfolio Selection: The paper investigates the use of Online Gradient Update and Online Newton Update meta-algorithms in online portfolio selection, proving their effectiveness in various financial settings. (2024-11-13, shares: 19.0)
Sustainable Investment Portfolio Optimization: The research presents a portfolio optimization approach based on a multi-index model that includes environmental, social responsibility, and corporate governance factors, providing a flexible alternative to large-scale covariance matrix estimation. (2024-11-13, shares: 18.0)
Statistical
DeepVol: Volatility Forecasting with Dilated Causal Convolutions: The study introduces DeepVol, a model using Dilated Causal Convolutions, which effectively uses high-frequency data to predict next-day market volatility. (2024-11-13, shares: 27.0)
Index Tracking with Shapley Explanations: The paper suggests using a one-dimensional Pointwise Convolutional Autoencoder and Shapley Additive Explanations for index tracking, outperforming other stock selection strategies in various financial markets. (2024-11-13, shares: 17.0)
Online Investor Sentiment & Stock Market Risk Premium: The study proposes using machine learning to predict stock market risk premium based on online investor sentiment, improving portfolio performance. (2024-11-13, shares: 22.0)
Volatility Forecasting: Linear vs. Nonlinear: Machine learning models were found to be effective in forecasting global stock market volatility, with simpler models performing better for volatility-timing portfolios. (2024-11-13, shares: 22.0)
Asset Pricing Uncertainty: A machine learning-constructed economic uncertainty index effectively predicted stock market returns, especially during high uncertainty and sentiment periods. (2024-11-13, shares: 18.0)
GitHub
Finance
Algorithmic Trading: The course provides a comprehensive understanding of algorithmic trading, targeting individuals with basic Python programming and financial market knowledge. (2023-07-31, shares: 127.0)
Shapley ML Interactions: The article explores the use of Shapley values in machine learning in the piece Shapley Interactions for Machine Learning. (2023-10-17, shares: 213.0)
FactorLab Library: FactorLab is a Python tool designed to assist in identifying and analyzing alpha and risk factors in investment algorithm creation. (2022-10-05, shares: 13.0)
Price Data Downloader: The project enables users to access historical price tick data for various financial instruments such as Crypto Stocks, ETFs, CFDs, and Forex through CLI and Node. (2019-06-08, shares: 362.0)
Pairs Trading Project: The project employs statistics and financial theory to demonstrate the Pairs Trading strategy often used in equity markets. (2019-09-04, shares: 308.0)
Trending
DearPyGui: Dear PyGui is an article about a fast and reliable GUI toolkit for Python with few dependencies. (2020-05-28, shares: 13270.0)
Sharebookkrpykrx: KRX stock information scraping discusses the process of extracting stock data from the Korea Exchange. (2018-12-12, shares: 696.0)
Pytrade.org: Pytrade is an article about a trading platform that is based on Python. (2024-02-16, shares: 55.0)
Aphroditeengine: Largescale LLM inference engine discusses a large-scale inference engine designed for logical latent models. (2023-06-23, shares: 1122.0)
AwesomeFastAPI: A curated list of awesome things related to FastAPI is a collection of useful resources and tools for FastAPI. (2020-05-03, shares: 8658.0)
LinkedIn
Trending
Open Investment Data: Sov.ai plans to launch an open investment data initiative, offering over 100 investment datasets for free for research purposes by 2026. (2024-11-09, shares: 4.0)
Geometric Overfitting: The article introduces a geometric definition of overfitting in generative models, which could be applied to risk modelling. (2024-11-11, shares: 14.0)
Metalabeling in Quantitative Trading: IllumiaResearch has improved their quantitative trading models by incorporating Meta-labeling, making them more adaptable to market changes. (2024-11-09, shares: 4.0)
Rough Transformers for Sequence Modeling: Fernando Moreno-Pino's paper on Rough Transformers for continuous-time sequence modelling has been accepted for NeurIPS 2024. (2024-11-07, shares: 6.0)
Symmetry Principles in Portfolios: CFM researchers have developed a new portfolio strategy, Eigenrisk Parity, which ensures equal risk across all principal components of the covariance matrix. (2024-11-07, shares: 16.0)
Survey on LLMs in Financial Management: A paper on 'LLMs for Financial and Investment Management', co-authored by Yaxuan Kong, has been published in the Journal of Portfolio Management. (2024-11-12, shares: 5.0)
Informative
PyARV Package Speed Comparison: The PyARV package is more efficient than SciPy in performing the inverse transform method for the Gaussian distribution. (2024-11-09, shares: 5.0)
Enhancing Portfolio Theory: John H Cohrance criticizes the complexity of Modern Portfolio Theory, suggesting a simpler, procedural, top-down allocation scheme for investors. (2024-11-12, shares: 3.0)
Twitter
Quantitative
Research Recap: Fixed Income, Stock Returns, ML Models, Arbitrage, and More: The article summarizes recent research on various financial topics including fixed-income investing, stock return predictability, economic constraints in machine learning models, and statistical arbitrage. It also mentions resources like blogs, repositories, and podcasts. (2024-11-12, shares: 0)
Competitive Advantages & Market Performance: Firms with significant competitive edges usually outperform those without, but not always the overall market. (2024-11-12, shares: 0)
Stocks Added to S&P 500: The article investigates the performance of stocks when they are included in the S&P 500, a crucial measure of the U.S. economy's health. (2024-11-07, shares: 0)
Reddit
Quantitative
Paper with Code
Trending
WebRL: Training LLM Web Agents: WebRL employs a self-evolving curriculum, a robust outcome-supervised reward model, and adaptive reinforcement learning strategies for consistent progress. (2024-11-08, shares: 146.0)
HTML for Modeling Retrieved Knowledge: The article introduces HtmlRAG, a tool that uses HTML in place of plain text for RAG, with the code accessible on GitHub. (2024-11-08, shares: 118.0)
Large Multimodal Model with Tabular Data Integration: The article presents TableGPT2, a model trained with a vast amount of table-related data, with the code available on GitHub. (2024-11-08, shares: 113.0)
LLMtimesMapReduce: LongSequence Processing: The LLMtimesMapReduce framework splits documents into sections for LLMs to handle and merges the results for the final product. (2024-11-13, shares: 104.0)