Quant Letter: August 2024, Week-1

Weekly (57th Edition)

After 3 great years, I'm leaving NYU this month to launch a data-software company. We're helping traditional asset managers (<100mn AUM) experiment with machine learning.

While we're in the early stages, our initial research suggests significant potential. If you're curious to learn more, reply "interested" to this email or message me on linkedin.

ArXiv

Finance

Hedge Fund Portfolio Construction: The article explores the use of PolyModel theory and deep learning in creating hedge fund portfolios for high returns and low risks. (2024-08-06, shares: 8)

Correlation in Limit Order Books: The study demonstrates the emergence of an Epps effect in two coupled diffusive limit order books using random walks. (2024-08-06, shares: 7)

PDEs for Interest Rate Derivatives Pricing: The paper discusses the use of partial differential equations in pricing interest rate derivatives under the generalized Forward Market Model. (2024-08-05, shares: 7)

NeuralFactors for Factor Learning: The research presents NeuralFactors, a machine-learning method for factor analysis that improves performance and efficiency in stock embedding. (2024-08-02, shares: 6)

Data Time Travel for Market Making: The article discusses the use of consistent data time travel in offline reinforcement learning for market making in limit order books. (2024-08-05, shares: 6)

Risk Sharing with Lambda VaR: The research investigates risk distribution among multiple parties using Lambda value at risk, offering formulas for optimal allocations under differing beliefs. (2024-08-06, shares: 5)

Quadratic and Linear Mean-Variance Equilibria: The article reexamines quadratic and linear mean-variance equilibria, providing conditions for the existence and uniqueness of these equilibria in both discrete and continuous time. (2024-08-06, shares: 5)

NeuralBeta: Estimating Beta: A new method, NeuralBeta, uses neural networks to estimate beta in finance, capable of handling both single and multiple variable scenarios and tracking beta's dynamic behavior. (2024-08-02, shares: 5)

Existence, Uniqueness, and Positivity of Volatility Solutions: The study confirms the existence and uniqueness of a solution to a path-dependent volatility model used to predict the price of an equity index and its spot volatility. (2024-08-05, shares: 5)

Machine Learning Relative Valuation of Municipal Bonds: A data-driven model using the CatBoost algorithm is introduced to create a supervised similarity framework for the muni bond market, outperforming both rule-based and heuristic-based methods. (2024-08-05, shares: 5)

Efficient Asymmetric Causality Testing: A study discusses the importance of significant differences between positive and negative components in asymmetric causality tests, applying this theory to the interaction between the world's two largest financial markets. (2024-08-06, shares: 2)

Random Forest Proximity for Regression: The article introduces a new method for calculating quantile regressions from random forests, showing improved performance and efficiency in predicting the average daily volume of corporate bonds. (2024-08-05, shares: 3)

Historical Trending

Deep Hedging with Implied Volatility: A new hedging strategy for S&P 500 options is introduced, using a unique reinforcement learning algorithm and hybrid neural network, which performs better than traditional benchmarks in tests and simulations. (2024-07-30, shares: 7)

Deep Learning for Trading: A new machine learning algorithm for options trading strategies is presented, which uses market data to create optimal trading signals, showing notable performance improvements over current strategies, particularly when using turnover regularization. (2024-07-31, shares: 6)

SSRN

Recently Published

Quantitative

Mutual Funds and Asset Pricing Anomalies: The anomalies in stock returns, specifically low-risk and momentum, are due to demand pressure from mutual funds, especially those with high-beta assets. (2024-08-06, shares: 3.0)

Unobservable Assets Comparison: CIOs find it challenging to compare liquid public assets and illiquid private assets due to real-world factors not reflected in reported returns. (2024-08-01, shares: 5.0)

Machine Learning and Stock Market Index: A novel method combining machine learning and Monte Carlo simulation significantly improves returns in Chinese A-share markets, surpassing existing benchmarks. (2024-08-02, shares: 3.0)

Score-Driven Model for Stock Indices: The BetatQVAR model, a volatility model for the t-distribution, outperforms other models in statistical and density forecasting when used on 15 international stock indices. (2024-08-05, shares: 3.0)

Analyzing Marginal Sharpe Ratio: The Marginal Sharpe Ratio (MSR) of an investment strategy considers the new strategy's impact on the portfolio's expected returns and the expected change in the portfolio risk profile due to diversification. (2024-08-05, shares: 3.0)

Validity of Post hoc Explanations: The research explores the effectiveness of post hoc explainers, SHAP and LIME, in determining the significance of variables in machine learning models, questioning their accuracy in revealing the real marginal effects of these variables. (2024-08-03, shares: 2.0)

Management Guidance: The research investigates the impact of management earnings guidance on market responses to earnings announcements in China, revealing that guidance increases trading volume but also raises bid-ask spreads and return volatility, especially for smaller, less visible firms. (2024-08-05, shares: 4.0)

ECBs Impact: The research uses a GVAR model to study the effects of the European Central Bank's unconventional monetary policies on six Central and Eastern European countries, showing that these policies reduce liquidity spread and raise yield spread, suggesting increased economic activity and a preference for bonds among investors. (2024-08-01, shares: 4.0)

Hedge Accounting: The research shows that only derivatives designated for hedge accounting assist firms in overcoming underinvestment issues, implying that the Financial Accounting Standards Board has developed an effective signaling tool about the success of firms' hedging programs, but firms using complex strategies often cannot designate some of their successful derivatives due to strict criteria. (2024-08-03, shares: 2.0)

Recently Updated

Quantitative

Detecting Criminal Firms with ML: A machine learning algorithm has been created to identify private firms linked to organized crime using financial accounting data, with a 91.4% accuracy rate. (2024-07-08, shares: 2.0)

Transmissions Among Assets: The study examines volatility and return spillovers in a network of variables, emphasizing the portfolio diversification benefits of commodities, fiat currencies, and crypto coins. (2024-06-07, shares: 2.0)

Dutch Book Argument for Banks: The paper outlines seven mathematical rules to prevent bank arbitrage, pointing out that existing models like Black-Scholes and the Heston model violate these rules. (2024-07-30, shares: 2.0)

Financial

AI Trading: AI can predict NFT prices accurately, but struggles with emotional dividends, potentially leading to financial losses over time. (2024-07-15, shares: 5.0)

Chinese Mutual Funds: A study on Chinese mutual funds reveals a significant positive risk premium, with lottery preferences accounting for nearly 40% of this premium, impacting investor decisions and risk regulation. (2024-06-30, shares: 2.0)

RePec

Statistical

Predicting Marine Accident Severity: The research suggests a new method for predicting marine accident severity using a two-stage feature selection and six machine learning models, with the Light Gradient Boosting Machine performing best. (2024-08-07, shares: 14.0)

Google Search Volume Index and Investor Attention: The study finds that Google Search Volume Index can be used to predict stock market movements and volatility, improving forecasting models. (2024-08-07, shares: 12.0)

Machine Learning

Cryptocurrency Portfolios Diversification: A stock-bond portfolio can gain significant diversification benefits by incorporating size- and momentum-based cryptocurrency factors, which can be further enhanced using machine-learning strategies. (2024-08-07, shares: 23.0)

Improved Crayfish Optimization for Feature Selection: The newly developed Improved Binary Crayfish Optimization Algorithm (IBCOA) enhances feature selection in data mining and machine learning, thus improving classification accuracy. (2024-08-07, shares: 19.0)

CoDe Model for Risk Perception: The Risk Co-De model, a machine learning-based system, can automatically classify social media posts about risk events with an accuracy of 86%. (2024-08-07, shares: 16.0)

ML Price Indices in Commercial Real Estate: The article introduces a new way of creating property price indices using machine learning, which is more accurate but less stable with small samples. (2024-08-07, shares: 15.0)

GitHub

Finance

Quantorch: Derivatives Pricing Library: The article presents a high-efficiency PyTorch library specifically created for pricing derivatives. (2022-01-06, shares: 135.0)

Automating Technical Analysis: The article explores the use of data analytics in making trading decisions based on price action. (2019-12-21, shares: 240.0)

AI for Trading Projects: The repository contains comprehensive notes and Python projects focusing on AI and Finance. (2020-08-11, shares: 283.0)

Pair Trading Environments: The article investigates the use of reinforcement learning in cryptocurrency trading using Backtrader. (2023-09-19, shares: 18.0)

Event Backtester: The article reviews a tool that tests trading strategies based on specific events. (2018-02-18, shares: 85.0)

SP 500 Components: The article offers updated and historical lists of S&P 500 components since 1996. (2019-11-16, shares: 386.0)

JSON Generator: The article introduces a reliable method for creating structured JSON from language models. (2023-04-29, shares: 4154.0)

LinkedIn

Trending

Chatbot creation with Llama 3.1 405B model: The Llama 3.1 405B model now enables the development of a chatbot for any GitHub repository using HuggingFace assistants. (2024-08-07, shares: 5.0)

Self-Organized Criticality in Economics & Finance: The article explores self-organised criticality in economics and finance, a concept explaining large fluctuations in complex systems like financial markets and global economies. (2024-08-01, shares: 5.0)

AI Predicting Lottery: Lol. The piece introduces SmileyNet, an innovative neural network that can predict coin flips with 72% accuracy using tea leaves, hinting at AI's potential to defy natural laws. (2024-08-02, shares: 6.0)

Sahm Rule and Recessions: The article examines the Sahm Rule, a recession indicator, its recent activation, and the possibility of using AI to identify macroeconomic trends from earnings transcripts. (2024-08-05, shares: 4.0)

Reconsidering Factor Models: The article reviews a paper that questions the simplicity and consistency of factor models, suggesting that more factors are relevant and their significance fluctuates over time. (2024-08-05, shares: 4.0)

Informative

Arxiv Paper Comments: The author discusses a new global platform for academic paper commentary, similar to one they helped develop for Roche two decades ago. (2024-08-01, shares: 3.0)

Portfolio Returns & Non-Normal Assets: A notebook inspired by Helder Alves shows that the assumption of normally distributed portfolio returns fails when assets are non-normal and correlated. (2024-08-06, shares: 4.0)

Data Juggernaut on Wall Street: An interview with Paul Humphrey, who established a significant data operation on Wall Street, has attracted over 3,000 views. (2024-08-01, shares: 4.0)

Performance Monitoring for HFT Systems: Constant performance monitoring is essential for high-frequency trading systems as it offers real-time insights, predictive analytics, benchmarking, data-driven decision making, and system resilience. (2024-08-03, shares: 5.0)

Podcasts

Quantitative

Jesse Felder: Insider Trading & Algorithmic Investing: Jesse Felder highlights the importance of insider trading in predicting market trends, drawing parallels between today's market and the dotcom bubble, and questioning the S&P 500's forward PE ratio's alignment with current economic data. (2024-08-05, shares: 12)

Diego Parrilla: AntiBubble Investing & Market Risks: Former trader Diego Parrilla presents his anti-bubble investment strategy, discussing the influence of disruptive technologies on the market and the use of gold volatility to navigate through unstable periods. (2024-08-06, shares: 11)

Mark Yusko: AI, Data Security & Fintech: Mark Yusko, founder of Morgan Creek Capital Management, shares his transition from traditional to digital assets, his belief in Bitcoin as a superior value store, and his predictions for blockchain technology's future. (2024-08-03, shares: 9)

Álvaro Cartea: AI Trading Strategies Evolution: Professor Álvaro Cartea explores the development of AI trading strategies, the unexpected outcomes of AI market makers, and the regulatory considerations of AI in finance. (2024-08-02, shares: 8)

News

Quantitative

SEBIs options curbs impact quants, HFTs: The Securities and Exchange Board of India is proposing new rules to control the increase in index options trading, impacting high-frequency traders, quantitative funds, and brokerage firms. (2024-08-05, shares: 7)

Hedge funds accused of market chaos: The recent rise in the yen, causing hedge funds to unwind carry trade bets, has been identified as a major cause for the global stock market selloff. (2024-08-07, shares: 7)

Quant Hedge Fund CIO Culture: The article discusses a common name among hedge fund founders. (2024-08-05, shares: 3)

Jain Global Trading Update: Jain Global, a hedge fund firm founded by ex-Millennium Management Co-CIO Bobby Jain, reported a slight loss of 0.65% in its first trading month, according to Business Insider. (2024-08-05, shares: 3)

Hudson River Trading Hires Female Data Scientists: An individual is relocating internationally for work. (2024-08-01, shares: 2)

Large Hedge Funds Using AI: Major hedge funds such as AQR, Balyasny, and Man Group are using artificial intelligence, as reported by Pensions & Investments. (2024-08-06, shares: 2)

Twitter

Quantitative

TSlib: Time Series Analysis Library: TSLib is an open-source tool for deep learning-based time series analysis with multiple models. (2024-08-02, shares: 3)3)

Free Historical FX Data: A free resource for historical FX data offers detailed millisecond data for various currency pairs. (2024-08-07, shares: 2)

Portfolio Optimization Practice: A new two-stage method for portfolio optimization is suggested, designed to produce portfolios applicable to real-world investment. (2024-08-05, shares: 2)

Miscellaneous

Market Returns During Recessions: Research shows that investors who convert to cash at the beginning of recessions can dodge initial market declines and enhance their Sharpe ratio. (2024-08-07, shares: 1)

Clifford Asness Episode: The author commends a recent episode featuring Clifford Asness. (2024-08-05, shares: 0)

Weekly Recap Expansion: The author intends to publish more detailed articles frequently due to the positive feedback on their weekly summary. (2024-08-04, shares: 0)

Macro Model Forecast Errors: A new study indicates that a basic macro model can predict inaccuracies in analysts' predictions of S&P 500 earnings, which could be beneficial for investors. (2024-08-03, shares: 0)

Paper with Code

Efficient Matrix Profile: The ACAMP algorithm has been found to be quicker than the SCRIMP matrix profile algorithm for z-normalized Euclidean distance. (2024-08-01, shares: 3281.0)

RecurrentGPT Generation: AIGC is investigating the use of RecurrentGPT to create interactive stories that directly involve consumers. (2024-08-05, shares: 134.0)

OVDINO OpenVocabulary Detection: A new method called OVDINO has been introduced to handle large-scale datasets in a unified system. (2024-08-03, shares: 97.0)

Multimodal Discrete Representation: The Dual Crossmodal Information Disentanglement model uses a single codebook for detailed representation and crossmodal generalization. (2024-08-01, shares: 49.0)