Quant Letter: November 2023, Week 1

Research of the Week (23rd Edition)

ArXiv

Finance

Estimating Realized Correlation in High-Frequency Financial Data: A new method for analyzing high-frequency financial data shows that intraday market changes are mainly driven by intraday correlation changes. (2023-10-30, shares: 5)

Agent-based Model for Deep Hedging: The Chiarella-Heston model, an advanced agent-based model, enhances deep hedging strategies by incorporating different types of traders, and performs better in creating realistic financial time series than three other models. (2023-10-28, shares: 9)

Estimating Systemic Risk in Networks: The article proposes a two-step nonparametric estimation method for measuring financial systemic risk, showing that only the second step's estimation error affects the results. (2023-10-28, shares: 4)

Optimal Fees in Hedge Funds with Compensation: The research suggests alternative fee schemes for hedge funds, arguing that traditional management and performance fees are suboptimal and that the recommended schemes reduce the fund's volatility. (2023-10-29, shares: 3)

Visibility Graph Analysis of Oil Futures Markets: A study using visibility graph methodology examines the effects of the Russia-Ukraine conflict and COVID-19 on crude oil futures markets, uncovering distinct market reactions to global disturbances. (2023-10-29, shares: 5)

Investing Characteristics' Impact on Financial Performance: A research study identifies 13 key investment characteristics that contribute to success in the equity market, offering a deeper understanding of the necessary traits for success in these markets. (2023-11-01, shares: 5)

Corruption's Impact on Performance: The study investigates the effect of managerial corruption on company performance, emphasizing the need for ethical corporate governance and careful manager selection. (2023-10-30, shares: 4)

Characterizing Law-Invariant Measures: The paper introduces new characterizations for law-invariant star-shaped functionals, demonstrating their wide use in finance, insurance, and probability scenarios. (2023-10-30, shares: 2)

Historical Trending

Optimal Execution with Machine Learning: A study introduces a numerical algorithm using dynamic programming and deep learning for optimal order execution, highlighting the convenience of using neural-network substitutes in stochastic control issues. (2022-04-18, shares: 52)

Analysis of Nonlinear Pricing: A paper proposes a method to calculate the best price schedule considering consumer diversity in continuous-choice situations, demonstrating that optimal price discrimination can boost a firm's profit by at least 5.5% compared to linear pricing. (2023-02-22, shares: 43)

Risk Evaluation and Robust Optimization with Model Aggregation: The model aggregation (MA) approach is a new method for risk evaluation that provides a robust value and distributional model, refining Value-at-Risk and Expected Shortfall characterizations. (2022-01-17, shares: 33)

Deep Reinforcement Learning for Portfolio Management Enhancement: A reinforcement learning framework for portfolio management is introduced, allowing for continuous asset weights, short selling, and decision-making, with three reinforcement learning algorithms compared for effectiveness. (2019-11-26, shares: 33)

Option Valuation on a Credit Index using Levy-driven Ornstein-Uhlenbeck Process: A Levy-driven Ornstein-Uhlenbeck process is proposed to model the risk-free rate and default intensities for evaluating option contracts on a credit index, with derived formulas and numerical experiments conducted. (2023-01-12, shares: 27)

TabR: Tabular DL Meets Nearest Neighbors: TabR, a new deep learning model for tabular data, outperforms existing models by using a k-Nearest-Neighbors-like component for better predictions. (2023-07-26, shares: 68)

Online Estimation & Community Detection of Network Point Processes: The research introduces a fast online variational inference algorithm for estimating latent structure in dynamic event arrivals on a network, offering comparable performance to non-online variants with computational benefits. (2020-09-03, shares: 19)

Crypto & Blockchain

NFT Market Fluctuations: Statistical Properties: The study shows that the Non-fungible token (NFT) market, although new and unique in its trading methods, has many statistical similarities with traditional financial markets, with some variations in certain quantitative measures. (2023-10-30, shares: 7)

High Frequency Analysis of Bitcoin Volume-Volatility: Research shows that unexpected trading volume is the key factor in spot volatility in Bitcoin futures and spot markets, while Bitcoin futures volumes have a calming effect on systemic volatility. (2023-10-27, shares: 2.0)

SSRN

Recently Published

Quantitative

VolGAN: Realistic Volatility Surfaces: VolGAN, a new model that can generate realistic scenarios for the joint dynamics of implied volatility surfaces and underlying assets, is introduced. (2023-10-30, shares: 173.0)

Machine Learning for Earnings Forecasts: Using machine learning models and comprehensive Compustat financial statement data for earnings forecasting can yield predictions that are up to 13% more accurate than traditional linear approaches. (2023-10-31, shares: 7.0)

Smart Beta ETFs & Increased Flow Sensitivity to Multi-Factor Alphas: Smart beta ETFs trading activity significantly impacts mutual fund flow sensitivity, especially in funds with high nonmarket risk factor exposure. (2023-11-01, shares: 2.0)

Google Trends Credit Interest Analysis: The paper discusses the use of Google Trends for analyzing credit interest in Armenia, eliminating the need for traditional surveys by gathering online search data. (2023-10-31, shares: 2.0)

Projected Fuzzy C-Means Algorithm: The article proposes a new algorithm for high-dimensional data clustering in machine learning, aiming to improve performance and manage anomalous instances. (2023-10-31, shares: 2.0)

KMeans Initialization: The article highlights the role of clustering in data mining and machine learning, focusing on the Kmeans algorithm and the challenge of selecting optimal cluster centroids. (2023-10-28, shares: 2.0)

Financial

Efficient Heston Model for Climate Contracts: A proposal suggests using Bitcoin-denominated derivatives contracts on carbon bonds to help governments hedge against climate change and influence carbon bond and cryptocurrency prices. (2023-10-31, shares: 3.0)

Private Equity Investment & Liquidity Shocks: Private equity investment outcomes can be influenced by investor composition, with funds from property and casualty insurers investing less during natural disasters, resulting in lower returns. (2023-10-27, shares: 3.0)

Green Derivatives & Climate Risk: The EU Green Deal aims to make Europe carbon-neutral by 2050, requiring 1 trillion euro in sustainable investments, with derivatives markets and 'green derivatives' crucial for managing climate risk. (2023-10-27, shares: 4.0)

Time & Frequency Analysis of Oil Futures Market: A study of the oil futures market from 1986 to 2020 reveals patterns and relationships between inventory, basis, hedging pressure, and futures risk premium, emphasizing the importance of the data measurement period. (2023-10-30, shares: 4.0)

Art as an Alternative Asset for Diversification: Art and collectibles can act as alternative assets for portfolio diversification, with art performing well compared to standard investments and showing a unique seasonal pattern in returns. (2023-10-30, shares: 4.0)

Indian Mutual Funds Performance Analysis: The study examines the performance and risk characteristics of Indian mutual funds across market capitalization groups, offering insights for investors and financial professionals. (2023-10-27, shares: 2.0)

Recently Updated

Quantitative

The Common Factor in Volatility Risk Premia: Firm-level volatility risk premium has a strong factor structure, with stocks with the weakest exposures to the common bad volatility risk premium factor earning higher average returns, and the common factor in total bad volatility risk premium predicting stock market returns. (2023-10-31, shares: 3.0)

Batch-Stochastic Sub-Gradient Method for Non-Smooth Loss Functions: The new machine learning method, Batchstochastic Subgradient, offers stable loss value estimates and is more memory efficient, as demonstrated using SQL. (2023-10-21, shares: 2.0)

US. Treasuries: Liquidity Premiums and Results: A new model of U.S. Treasuries suggests that liquidity factors are more significant than others, Federal Reserve asset purchases impact expected rates and term premiums, and inflation expectations are less stable than previously thought. (2022-05-06, shares: 2.0)

Global Uncertainties and Emerging Market Sectors: A study finds a significant link between global financial uncertainties and emerging market sectoral indices, based on data from 2008 to 2021. (2023-07-17, shares: 2.0)

Financial

Algorithmic Trading's Influence on Human-Only Markets: The potential existence of algorithmic trading can impact human price predictions, trading activities, and price dynamics in human-only asset markets, even if no actual algorithmic trading is present. (2023-07-17, shares: 3.0)

Bond Funds and Liquidity Provision: Changes in regulations have moved profits from liquidity provision in the corporate bond market to mutual funds, increasing volatility and vulnerability to market disruptions like the COVID-19 pandemic. (2023-10-23, shares: 23.0)

ETFs and Market Efficiency: Capital constraints on intermediaries can affect the pricing efficiency of assets they manage, as seen in ETFs and their lead market makers during the COVID-19 debt market disruptions. (2022-03-30, shares: 369.0)

ETF Closures: Inaction for Investors?: Research indicates smaller ExchangeTraded Funds (ETFs) often yield higher daily returns and typically close after positive returns. Investors usually fare better by not reacting to closure announcements. (2023-01-23, shares: 60.0)

Investor Returns: Market-Based Statistics: The study presents three market-based approximations of actual return from market trades, which deviate from traditional evaluations based on time series analysis of investors' returns. (2023-04-11, shares: 25.0)

Cost of Capital: Cross-Sectional Analysis: Research spanning 20 years across multiple countries shows that most variations in perceived capital cost are not supported by subsequent returns, questioning the production-based asset pricing model. (2020-12-11, shares: 2.0)

Equity and Credit Index Options: Risk & Return Analysis: A new credit risk model accurately prices equity and credit index options, contradicting previous claims of inconsistent pricing, and highlights the need to balance three systematic risk sources. (2021-07-14, shares: 2.0)

Levered ETF Rebalancing: Market Volatility Impact: The study reveals that the interaction between investor behavior, ETFs fund flows, and index return autocorrelation can either temper or intensify market volatility, as observed during the COVID-19 pandemic onset. (2022-04-08, shares: 2.0)

RePec

Machine Learning

Explainable AI Reveals Bond Excess Return Determinants: The SHapley Additive exPlanations technique is used in a paper to identify key factors influencing bond excess return predictions made by machine learning models. (2023-11-02, shares: 21.0)

Forecasting volatility with machine learning: Panel data perspective: The study uses machine learning to predict volatility in high-frequency data, with panel-data-based methods proving most effective. (2023-11-02, shares: 45.0)

Cross-market info & stock market volatility prediction: A study reveals that cross-market information greatly impacts the volatility of the Chinese stock market, especially in medium and long-term forecasts. (2023-11-02, shares: 18.0)

Finance

Intraday profitability and trading behavior in algorithmic trading: Profitability and behavior in algorithmic trading.: The study examines the intraday profitability and interactions among traders, revealing that algorithmic traders profit while non-algorithmic traders lose, with market volatility causing contrasting trading behaviors. (2023-11-02, shares: 29.0)

Dynamic bond portfolio optimization with a stochastic interest rate model: Bond portfolio optimization with stochastic interest rate model.: The paper introduces a new framework for multi-period dynamic bond portfolio optimization, showing that multi-period optimization outperforms single-period optimization, particularly over extended investment and utilization periods. (2023-11-02, shares: 26.0)

Multiperiod portfolio allocation with volatility clustering and non-normalities: Portfolio allocation with volatility clustering and non-normalities.: The research investigates the dynamic multiperiod portfolio choices of a U.S. stock market investor, discovering that considering volatility clustering decreases hedging demands and non-normalities slightly affect allocations. (2023-11-02, shares: 23.0)

Performance of U.S. ESG ETFs: A study finds that ESG equity ETFs in the U.S. generally outperform the S&P 500 Index, challenging the notion that ESG investing compromises financial returns. (2023-11-02, shares: 15.0)

High-Dimensional Portfolio Optimization with Tree-Structured Factors: A new portfolio optimization method using a tree-structured portfolio sorting technique predicts stock returns and risk exposures, outperforming benchmark strategies in the Chinese A-share market. (2023-11-02, shares: 15.0)

Volatility Smile in Emerging Markets: Dynamic Approach: A study shows the Dynamic Nelson-Siegel model is more effective than static models for predicting volatility in options markets. (2023-11-02, shares: 23.0)

Bond-Commodity Volatility Spillover & Global Liquidity Risk: Research reveals significant volatility spillovers between gold and bond markets, and oil and some bond markets, suggesting limited diversification benefits for investors. (2023-11-02, shares: 20.0)

fBetas & Portfolio Optimization with f-Divergence Risk Measures: A new f-Beta for portfolio optimization, which assesses portfolio performance under an optimally disturbed market probability measure, offers flexibility and interpretability. (2023-11-02, shares: 18.0)

Performance of Actively Managed ETFs: A study from 2018-2021 reveals that actively managed Exchange Traded Funds (ETFs) in the U.S. did not yield significant above-market returns, indicating managers lacked superior market timing skills. (2022-12-23, shares: 18.0)

GitHub

Trending

FinGAN for Financial Time Series -> FinGAN for Time Series: This article shares the code related to the FinGAN paper, which uses Generative Adversarial Networks for financial time series forecasting and classification. (2023-10-26, shares: 16.0)

Easy Data Loading with DLT -> Data Loading with DLT: The article presents 'data load tool dlt', a Python library that simplifies data loading. (2022-01-26, shares: 669.0)

SolidGPT: Code Collaboration: The article explores a platform that facilitates interaction with your code repository and discussion of coding needs. (2023-08-08, shares: 1369.0)

LinkedIn

Trending

VolGAN: A Generative Model for Arbitrage-Free Volatility Surfaces: The article presents VolGAN, a generative model for arbitrage-free implied volatility surfaces, and discusses its performance on SPX implied volatility time series. (2023-11-01, shares: 1.0)

The New Era of Systematic Investing and Parallels to ESG: The article analyzes Campbell Harvey's views on the role of Machine Learning/AI in investing, discussing potential benefits, risks, and its relation to ESG investing. (2023-11-01, shares: 1.0)

Quantitative Models in Chinese Stock Market: The Chinese stock market's growth and adaptability make it ideal for quantitative models, as discussed at a London forum. (2023-11-01, shares: 1.0)

Informative

Nasdaq's SEC-Approved AI Order Type: The U.S. Securities and Exchange Commission has approved Nasdaq's use of an AI-driven order type, the first of its kind, for executing orders. (2023-11-01, shares: 1.0)

New Paper on Statistical Arbitrage Portfolios: A new paper on statistical arbitrage portfolio construction based on preference relations has been published by Fredi Šarić, Stjepan Begušić, Andro Merćep and Zvonko Kostanjcar. (2023-11-01, shares: 1.0)

Machine Learning Applications: Tricky Properties and Catastrophic Forgetting: The article highlights the difficulties in implementing machine learning applications, focusing on issues like 'catastrophic forgetting' and the need for model and data input adjustments. (2023-11-01, shares: 1.0)

Challenging the Belief in More Data for ML Models: Clint Howard, in a seminar, challenged the notion that more data used in training machine learning models always results in better performance. (2023-11-01, shares: 1.0)

Stock Market Efficiency in Pricing Climate Change Risks: Man Institute researchers suggest that the stock market often underestimates the impact of climate-related news, creating opportunities for savvy investors. (2023-11-01, shares: 1.0)

Common Domain Model (CDM): Revolutionizing Finance: The Common Domain Model (CDM), an open-source framework, is transforming finance by standardizing processes and reducing operational risks and costs. (2023-11-01, shares: 1.0)

Podcasts

Quantitative

Scariest Options Strategies Revealed: The Options Insider Media Group talks about the current market situation, the forthcoming earnings season, and the five most daunting options strategies. (2023-11-01, shares: 8)

Macro Volatility and Recession Risks with Boris Vladimirov: Goldman Sachs' Boris discusses fiscal policy's impact on growth, private sector rate sensitivity changes, and recession odds in a podcast. (2023-10-27, shares: 4)

Corey Hoffstein on Bitcoin ETF and TBill Discussion: Corey Hoffstein and Meb discuss Bitcoin ETF, BlackRock's TargetDate ETFs, and the end of the 60/40 strategy on a radio show. (2023-11-01, shares: 3)

Efficient Use of Graphs with LLMs in GraphText: In a podcast, Jianan Zhao, a Computer Science student, talks about the efficient use of graphs with LLMs. (2023-10-31, shares: 2)

In-depth Conversation with Traderade Cofounder on MH Ep.: Kevin and Traderade Cofounder Horselover Fat discuss trading setups, Traderade's origins, and experiences in the trading industry. (2023-10-31, shares: 1)

Insights on FOMC Meeting: The Financial Conditions Dummy: Neil Azous from Rareview Capital predicts no further policy tightening ahead of the November FOMC meeting. (2023-10-31, shares: 1)

Twitter

Quantitative

Total Return vs. Derivative Income in Covered Call Strategies: Israelov and Ndong's paper discusses the inverse relationship between expected total return and derivative income in covered call strategies. (2023-10-29, shares: 3)

Decoding the Volatility Puzzle: Swedroe's article investigates the idiosyncratic volatility puzzle by studying the fundamental aspects. (2023-10-28, shares: 2)

SciPhi ΨΦ: Custom Data Generation with LLMs: The article introduces SciPhi ΨΦ, a system for creating synthetic data to meet specific requirements using LLM-based OpenAI Anthropic Llama. (2023-10-28, shares: 2)

MS Report on Wealth Management and Generative AI Tipping Point: The report by OliverWynan and MS explores the convergence of wealth and asset management and the critical point of generative AI. (2023-10-27, shares: 2)

Langchain Extensions for Coordinated Computation: The article presents Permchain and Langchain extensions, tools that enable multiple agents to coordinate over several computation steps using LangChain Expression Language and Pregel. (2023-10-30, shares: 0)

Miscellaneous

Large Language Model Inferences on Stock Factors: A new study has been released discussing the implications of Large Language Model on different stock factors. (2023-10-27, shares: 0)

China LLM with Advanced Question Answering Abilities: Article 2: DISCFinLLM is a novel Chinese financial LLM that features multiturn question answering, text processing, mathematical computation, and enhanced retrieval generation. (2023-10-30, shares: 0)

Abductive Reasoning in Financial Language Model Building: Article 3: A new financial LLM that uses abductive reasoning surpasses standard financial LLMs, setting new high scores in financial analysis and interpretation tasks. (2023-10-30, shares: 0)

Python and R Time Series Library: Pytimetk is a high-performance timeseries library, compatible with Python and R, that utilizes Polaris dataframes for simplicity. (2023-10-30, shares: 0)

Videos

Quantitative

Discovering Supply Chain Edges with Graph Neural Networks: Achintya Gopal from Bloomberg uses graph neural networks to predict unknown suppliers and customers, improving supply chain risk analysis. (2023-11-01, shares: 9.0)

Where Did All the Quants Go?: A LinkedIn comment criticizes quant programs for lacking intuition and rigor, stressing the need for continuous learning and understanding of financial market logic and mathematics. (2023-10-29, shares: 52.0)

Reddit

Quantitative

Two Sigma Hedge Fund Scandal: The article explores the differences in pay at proprietary trading firms, with some requiring negotiations on a per-portfolio manager basis. (2023-10-29, shares: 230.0)

Famous Quants in History: The author is asking for suggestions of well-known quants, besides Pat Haber and Martin Artajo, whom they already know. (2023-10-28, shares: 117.0)

Quant Trader in HK or SG: The author is looking for guidance on a potential Quant Trader role in the Asian branches of a London hedge fund, particularly in Singapore and Hong Kong. (2023-10-30, shares: 17.0)

Million Market Experiment Loss: The author is exploring strategies for a hypothetical scenario where all money is lost through market investments as part of an experiment. (2023-10-30, shares: 114.0)

Paper with Code

Rising

Natural Language Graphs: ChatGPT, a large-scale pretrained language model, has significantly advanced various fields of artificial intelligence research. (2023-11-01, shares: 117.0)

Tuning Graph Instructions for Language Models: GraphGPT uses a graph instruction tuning paradigm to align large language models with graph structural knowledge. (2023-10-30, shares: 92.0)

Graph-based Tools for Language Model Augmentation: ControlLLM is a new framework that enables large language models to use multimodal tools to tackle complex real-world tasks. (2023-10-31, shares: 45.0)