Quant Letter: October 2024, Week-2

Weekly (66th Edition)

Finance

AI Economy Score: Predicting Economic Indicators from Corporate Calls: The article discusses the use of Generative AI in analyzing over 120,000 corporate conference call transcripts to predict future economic indicators like GDP growth, production, and employment. This technology provides valuable insights for macroeconomic and microeconomic decision-making. (2024-10-04, shares: 2)

Fourier Cosine Method: The research proves that the Fourier cosine formula for the inverse Fourier transform can be applied to discrete probability distributions, indicating its usefulness in computational statistics and quantitative finance. (2024-10-06, shares: 3)

Historical Trending

Auction Designs for Electricity Markets: The research shows that strategic bidding in electricity markets can boost producer profits and minimize lost opportunity costs, but may also raise consumer costs. (2022-12-20, shares: 52)

Neural Term Structure for Option Pricing: The research introduces the neural term structure model for option pricing, providing benefits in creating the risk-neutral measure, pricing formula, and fitting the implied volatility surface. (2024-08-03, shares: 6)

SSRN

Recently Published

Quantitative

Geoffrey Hinton's Neural Networks Contributions: Geoffrey Hinton's contributions in neural networks, deep learning, and Capsule Networks have significantly influenced modern AI systems used in various fields like image and speech recognition, and natural language processing. (2024-10-08, shares: 224.0)

DataDriven Inventory Management with Financial Hedging: The study presents a data-driven approach for inventory and financial hedging for new products, using return factors to predict demand and future returns, leading to a new decision-making framework. (2024-10-04, shares: 4.0)

Spatio-Temporal Machine Learning for Mortgage Credit Risk: A new machine learning model for credit risk combines tree-boosting with a latent spatiotemporal Gaussian process model, offering more accurate predictions of default probabilities and loan portfolio losses. (2024-10-03, shares: 2.0)

Organizational Form and Liquidity Management: The research compares the liquidity and performance of open and closed-end funds, revealing that closed-end funds hold more illiquid municipal bonds, charge higher fees, but perform worse than open-end funds. (2024-10-03, shares: 2.0)

Drivers of Credited Interest Rate on UL Insurance: The research explores the factors influencing the credited interest rate on universal life insurance in China, showing a positive long-term effect of insurance company operations and market interest rate trends on credited interest rates. (2024-10-07, shares: 2.0)

Financial

Index Investing: The study suggests that index stocks have higher prices, more volatility, stronger negative price autocorrelation, and higher trading volume due to sentiment spillover from index investors. (2024-10-07, shares: 4.0)

Capital Requirements: The research shows that bank Credit Default Swaps (CDS) are influenced by regulatory capital ratios, with markets reacting more to changes in capital requirements if implemented via Pillar 1 risk weights. (2024-10-08, shares: 3.0)

Bank Credit Lines: The research contradicts previous theories, stating that non-retail funding, particularly wholesale funding, significantly drives banks' contingent commitments, not traditional retail deposits. (2024-10-07, shares: 2.0)

Momentum Metrics: The study finds that in the Indian equity market, volatility-adjusted and information discreteness momentum strategies deliver higher risk-adjusted returns. (2024-10-07, shares: 4.0)

Option Pricing Formula: The article presents a pricing formula for a ComEx option, which allows the exchange of two options, and compares its price evolution to the Margrabe option, which permits the exchange of two underlying assets. (2024-10-03, shares: 4.0)

Carbon Offset Markets: The research explores the relationship between prices in Australia's renewable energy certificate market, carbon offsets market, and electricity market from May 2018 to June 2023, using a portfolio approach for backtesting. (2024-10-03, shares: 3.0)

Recently Updated

Quantitative

Statistical Arbitrage with Mixed Frequency Data: The article discusses the use of high-frequency data to identify similar assets for statistical arbitrage strategies, testing various algorithms and trading rules on different asset classes. (2024-09-30, shares: 7.0)

Black-Litterman Model for Risk Factor Views: The paper presents an extended Black-Litterman model that allows fund managers to apply their subjective views to risk factors, simplifying the portfolio allocation process. (2024-10-01, shares: 3.0)

Pricing Factors: A study proposes a new method for choosing factors in portfolio management, highlighting announcement return and earnings predictability as key, outperforming other methods. (2023-02-28, shares: 2.0)

AI in Accounting Systems: A new course has been developed to introduce artificial intelligence into the accounting information systems curriculum, using various teaching methods to aid understanding. (2024-07-01, shares: 2.0)

Financial

GIS Analysis: The study reveals that France's history with growth-contingent bonds shows the equity premium isn't just compensation for GDP risk, as it persists even when hedged against GDP changes. (2024-09-01, shares: 2.0)

Profitable Day Trading Myth: The research indicates that less than 1% of day traders consistently make profits, with most suffering major losses due to poor risk management, emotional control, and unrealistic expectations fueled by aggressive marketing. (2024-08-19, shares: 2.0)

Competition vs. Collusion in Trading: The study proposes that collusion in trading may not always be advantageous for informed investors, particularly when there's high uncertainty about other traders' actions, as individual profits in competition can surpass total profits in collusion. (2024-10-01, shares: 2.0)

ArXiv

ArXiv ML

Recently Published

Regression CP Bias: A study on Conformal Prediction intervals shows that asymmetric adjustments are unaffected by bias and maintain the same validity as if the bias never occurred, unlike symmetric adjustments. (2024-10-07, shares: 10)

SIEVE: Data Filtering System with GPT-4o Accuracy: The article introduces SIEVE, a cost-effective method for filtering web-scale data that matches the accuracy of GPT-4o and is efficient in curating large datasets for language model training. (2024-10-03, shares: 8)

RePec

Finance

Chinese Futures Market Evolution: Research shows high-frequency and algorithmic trading in China enhances market liquidity and lowers slippage costs for investors, despite the country's unique market structure. (2024-10-09, shares: 17.0)

Hedge Funds Geopolitical Risk Hedging: Hedge funds with larger minimum investments and management fees are more effective at hedging geopolitical risks, while global macro hedge funds excel at timing these risks, offering crucial insights for private investors during high geopolitical risk periods. (2024-10-09, shares: 16.0)

Dynamic Portfolio Selection with Factor Models: A new factor model system, capturing both return and risk dynamics, has been introduced, with U.S. market data indicating it offers a better out-of-sample Sharpe ratio than benchmark policies. (2024-10-09, shares: 15.0)

Statistical

Predicting Cryptocurrency Volatility: The SHARV-MGJR model, which includes volatility leverage effects and current return data, is suggested to enhance the precision of cryptocurrency market volatility forecasts, surpassing GARCH-type models. (2024-10-09, shares: 22.0)

Explainable AI Framework for Risk Management: The article highlights the difficulties of using machine learning models in practical risk management in banking due to their opacity and lack of explainability, and introduces a framework for leading eXplainable AI methods. (2024-10-09, shares: 20.0)

Herding Patterns in AI and Big Data Token Markets: Research shows that investors in AI and big data token markets tend to follow the crowd during stable, low-volume market conditions, but act independently during volatile, high-volume conditions, which has implications for market regulation and stability. (2024-10-09, shares: 17.0)

Machine Learning

Bottom-up Inflation Forecast: Machine learning techniques like gradient boosting or regularised regression can accurately forecast CPI inflation, as shown with Russian data. (2024-10-09, shares: 27.0)

Detecting Collusion in Public Procurement: A new algorithm has been developed to detect collusion in auctions using public procurement data, identifying a significant number of contracts with high collusion likelihood. (2024-10-09, shares: 18.0)

Stylized Facts of DeFi: The chapter discusses Decentralized Finance (DeFi), its potential impact on centralized finance, and its market efficiency, volatility, leverage effects, and return volume relationship. (2024-10-09, shares: 18.0)

GitHub

Finance

Data Stack Guide: This article offers a detailed guide on constructing a straightforward and efficient data stack. (2024-02-17, shares: 30.0)

IB Fundamental Data: The piece describes how Interactive Brokers delivers essential data in a way that is easy for users to understand and use. (2024-05-27, shares: 28.0)

Time Series Models: This article introduces TimeMoE, a base model designed to manage billion-scale time series data using a combination of experts. (2024-09-22, shares: 151.0)

Awesome Web Apps: The piece presents a collection of open-source, frequently updated web apps specifically designed for LLM applications. (2023-09-24, shares: 326.0)

LinkedIn

Trending

MIT Professor's Warning on AI Overinvestment: MIT professor Daron Acemoglu believes that the billions invested in AI are wasted, as he predicts AI will only impact 5% of the workforce. (2024-10-04, shares: 6.0)

Geodesic Curve for Covariance: A new method for regularized covariance estimation uses a geodesic curve to balance top-down and bottom-up portfolio construction. (2024-10-04, shares: 17.0)

Verisimilitude's Role in Modeling: In his book Making Sense of Chaos, Doyne Farmer criticizes models with unrealistic assumptions, advocating for models that align with facts and plausible assumptions. (2024-10-07, shares: 8.0)

Dangers of Repeated Markowitz Optimization: Line Chen and Xun Yu Zhou's paper warns that overuse of Markowitz's portfolio theory can increase risk, highlighting the need for continuous time portfolio calculus. (2024-10-03, shares: 9.0)

Statistical Trading Strategy Predictions: Álvaro Cartea's paper, predicting trading algorithm behavior using Euronext exchange data, has been accepted for publication in the Journal of Financial Econometrics. (2024-10-03, shares: 3.0)

Causal Factor Themes with GPT: Alik Sokolov's paper uses AI and do-calculus to automate the process of finding cause-effect relationships in financial markets, enhancing factor orthogonality. (2024-10-07, shares: 4.0)

Interest Rate Machine Learning: Autoencoders for Yield Curve Modeling: The author announces an updated version of their paper on using machine learning for interest rates, introducing a new method for developing risk-neutral models. (2024-10-03, shares: 4.0)

Informative

Market Signal Prediction: The article discusses Predicting the Predictor, a strategy that aims to predict derived market signals instead of raw price movements for better trading strategies. (2024-10-08, shares: 4.0)

Portfolio Optimization with MD Methods: The article presents a new portfolio optimization model that uses linear programming and mixed-integer linear programming for improved risk-adjusted returns and quicker computations. (2024-10-05, shares: 3.0)

Podcasts

AIs Influence on Hedge Funds: Abraham Thomas of Quandl explores the influence of AI and LLMs on the alternative data and hedge fund sectors, predicting potential winners, losers, and future scenarios. (2024-10-05, shares: 6)

AI Investing with Brad Gerstner: Brad Gerstner, Altimeter Capital's founder, talks about his investment journey, AI's impact on technology, and his Invest America initiative, which aims to improve financial literacy and inclusion for U.S. children. (2024-10-04, shares: 3)

News

Quantitative

Bloomberg's Intraday Pricing Solution: Bloomberg has launched OpenHighLowClose Bar, a new tool that allows users to create custom intraday pricing datasets, simplifying quant workflows. (2024-10-03, shares: 5)

London Experts Discuss AI in Finance: At the Quant Strats 2024 in London, experts discussed the impact of AI and geopolitics on financial disruption. (2024-10-08, shares: 4)

Columbia Threadneedle Partners DBS: Columbia Threadneedle is collaborating with DBS Bank to offer its tech-focused hedge fund strategy, managed by Seligman Investments, to affluent clients in Asia, as per CityWire. (2024-10-07, shares: 3)

Young Quants Depend on LLMs for Portfolio Growth: The article reflects on the era when Google was the dominant platform for online searches. (2024-10-08, shares: 2)

Twitter

Quantitative

Recent Research on Quant Investing: New research in quantitative investing explores predicting option returns, sentiment and equity predictability, mutual fund strategies, and the application of large language models in trading. (2024-10-08, shares: 4)

Generative AI for Timeseries Prediction: Generative AI solutions for time series prediction are available, with the smallest pre-trained model, Chronos T5 Tiny, trained on a vast amount of public and synthetic data. (2024-10-06, shares: 3)

Momentum Study to Reversal: A momentum investing study indicates a shift to reversal after high past 1-month volatility, showing strong results in the US, China, Brazil, and India, but cautions about high turnover. (2024-10-04, shares: 2)

Trend Following in High Interest Rates: A white paper from Quantica suggests that trend following investments are more appealing during periods of high interest rates. (2024-10-07, shares: 1)

ARP Episode: The article delves into a significant episode on alternative risk premia by Choffstein. (2024-10-07, shares: 1)

Small Models Outperform: The article evaluates a new study proposing that smaller, domain-specific language models excel in trading, particularly for small-cap stocks. (2024-10-03, shares: 1)

Paper with Code

Fast Transformers: The article explores the slow data processing of large autoregressive models like Transformers, which need K serial runs to decode K tokens. (2024-10-03, shares: 3578.0)

EU Speech Data Training: Regulatory efforts addressing the risks of foundation models are increasing their popularity and driving interest in open-source versions. (2024-10-08, shares: 116.0)