Quant LetterNo. 64
145 items across 10 sections, as sent to readers on 5 September 2024. Paper titles open their ML-Quant page; ↗ goes to the source.
Quantitative-finance and ML-for-finance preprints from arXiv.
7 items
The article talks about how lapse-supported premiums can raise the costs of adverse selection in life insurance, especially when high-risk individuals keep their policies longer. It also proposes three ways to handle surplus from policy lapses.
2 shares1 citation todaySource ↗
The article introduces a new technique for estimating the linear relationship between multiple variables with measurement error. This method treats all variables equally, extends the geometric mean functional relationship to multiple dimensions, and provides a scale-invariant and simple solution for data without a specific distribution.
3 shares1 citation todaySource ↗
Stock Volatility Prediction: ECCAnalyzer, a new tool, uses advanced language models to extract data from earnings conference calls, enhancing stock volatility predictions and surpassing traditional analysis methods.
11 shares20 citations todaySource ↗
The Greeks Delta and Gamma in plain vanilla options are inaccurately approximated by three Fourier methods, indicating finite differences may provide a better approximation.
9 shares2 citations todaySource ↗
A Special Issue of the International Journal of Theoretical and Applied Finance provides an overview of Thomas Robert Hurd's scientific career and a list of his scientific publications.
7 sharesSource ↗
A new method for estimating sparse second-order stochastic spanning suggests no advantage in expanding a sparse opportunity set beyond 45 assets, with the best sparse portfolio investing in 10 sectors.
6 shares15 citations todaySource ↗
A new framework for optimizing and learning Nash equilibrium uses neural networks to estimate players' cost functions, with two methods proposed based on data availability, and tested in numerical experiments.
3 shares1 citation todaySource ↗
Working papers in finance and economics from SSRN.
31 items
The paper explores the role of AI and machine learning in enhancing wealth management by optimizing investment returns and risk management.
4 sharesSource ↗
The article examines the failures of AI in investment strategies, proposing network data analysis as a solution to build stronger AI systems.
8 sharesSource ↗
The study presents two new diversification indices, MDQ and MDQM, and compares them with existing indices using real data sets.
2 sharesSource ↗
The research uses hybrid machine learning models and balanced data to increase the accuracy of predicting financial distress, with the hybrid model outperforming the single model.
3 sharesSource ↗
The paper introduces the Citizen Prosperity Index (CPI), a new method of measuring prosperity beyond economic indicators, and discusses its use in east London.
9 sharesSource ↗
A new algorithm using Empirical Bayes approach enhances machine learning results in remote sensing image analysis.
3 sharesSource ↗
Technology venture acquisitions boost venture capital investment, especially in areas with fewer ventures and high acquisition activity.
2 sharesSource ↗
The paper examines the structure of Transformers, large language models, and their potential and challenges in patent research.
3 sharesSource ↗
The paper explores the applications and challenges of Big Data, including various types of data analytics, with a focus on India.
2 sharesSource ↗
An evaluation index for anomaly detection in power quality data proves effective and sensitive to big data attributes and anomaly types.
2 sharesSource ↗
The article proposes a new approach to estimating the zerobeta rate in asset pricing, arguing that current methods are flawed and result in high, volatile rates.
3 sharesSource ↗
The piece highlights how AI has transformed software engineering, improving efficiency, accuracy, and adaptability in areas like defect detection and software testing.
5 sharesSource ↗
The article explores the challenges and solutions in scaling machine learning systems, emphasizing on high throughput and reliability with large data sets.
2 sharesSource ↗
The article notes that US Venture Capital activity is increasingly using digital data and machine learning to guide investment decisions, leading to more investments outside major hubs.
2 sharesSource ↗
The study applies predictive analytics and machine learning to enhance energy efficiency in Nigeria's power sector.
3 sharesSource ↗
The research explores the link between beta wave activity in the motor cortex, keyboard force, and its impact on typing speed and accuracy.
3 sharesSource ↗
The paper introduces a robust nonparametric estimator using high-frequency financial data to study integrated betas' dynamic structures.
2 sharesSource ↗
The paper examines the use of machine learning and mathematical algorithms in real estate for advanced marketing strategies.
2 sharesSource ↗
The article highlights the difficulties in detecting anomalies in the financial service industry, especially in fraud detection and cybersecurity.
2 sharesSource ↗
Contrary to popular belief, informed short sellers are found to supply liquidity and provide valuable information about future returns, especially on news days.
30 sharesSource ↗
A new method reveals that individual mutual funds strategically trade ESG stocks, particularly around the extreme ESG rating categories' cutoffs.
3 sharesSource ↗
The expected return rate of an asset does impact the option pricing model, contrary to continuous-time finance literature, necessitating a linear function between risk-free interest rate and the asset's expected return rate and return variance.
3 sharesSource ↗
Retail market making in the German equity market is highly profitable, with market makers willing to sacrifice about 60% of their revenues for access to retail order flow.
5 sharesSource ↗
Banks in vulnerable euro area countries hold a significant amount of high-yielding sovereign debt, with variations in bank funding costs explaining the differing risk appetites for riskier government bonds over bonds issued by core countries or EU supranational entities.
4 sharesSource ↗
Self-indexed ETFs often charge higher fees than public-index ETFs due to potential conflicts of interest, especially when the issuers also provide wealth management services.
2 sharesSource ↗
A study using smartphone geolocation data shows that frequent SEC visits to larger firms and those with past enforcement actions significantly affect stock prices and insider behavior.
4 sharesSource ↗
Investment regret has a strong positive correlation with future returns, particularly for small stocks and in emerging markets, though this effect diminishes over time.
4 sharesSource ↗
The stock market's reaction to monetary policy surprises is primarily due to changes in the default-free term structure of yields, not equity premium changes, as per a new method using dividend futures prices.
2 sharesSource ↗
Mutual fund managers' tunnel vision towards forecasts from connected analysts can negatively affect their trading performance due to exposure to these analysts' biases.
2 sharesSource ↗
The article introduces a new method for evaluating customer value using a continuous-time option pricing framework, suggesting marketing costs are irreversible investments dependent on uncertain profit margins, a departure from the traditional discounted cash flow approach.
4 sharesSource ↗
The article introduces a new equilibrium model with a unique habit sensitivity function and nonconvex adjustment costs, explaining aggregate and individual asset prices and investment rates through significant volatility in marginal utility, and provides computational benefits for expected returns considering aggregate risks.
2 sharesSource ↗
Economics working papers from RePEc's NEP field reports.
28 items
Fintech lenders are using advanced data and models to provide more loans to high-risk areas, potentially aiding small businesses.
22 sharesSource ↗
A study shows the potential of using compositional volatility modeling, with data from German politics and US income shares, for applications beyond budgeting.
22 sharesSource ↗
A new hedge strategy has been created to lessen carbon risk in diverse portfolios, enabling investors to lower their carbon risk without major losses in returns.
21 sharesSource ↗
The article discusses the use of PCA and LSTM machine learning methods for predicting Poland's yield curve, with LSTM providing the most accurate predictions.
16 sharesSource ↗
The study analyzes the effect of limit-to-arbitrage factors on the distress risk puzzle in Vietnam's stock market, noting a negative correlation between distress risk and corporate profitability that disappears after bankruptcy regulations are implemented.
15 sharesSource ↗
The research examines calendar anomalies and their impact on returns and volatility in the cryptocurrency market, highlighting significant shifts in volatility and day-of-the-week effects on returns during the COVID-19 pandemic.
15 sharesSource ↗
The paper presents arbitrage-free conditions for a parametric yield curve in the P-world, first imposing a family of shapes on the yield curve and then deriving market prices of risk in a self-consistent manner.
14 sharesSource ↗
Research indicates that traditional statistical methods are more effective than basic machine learning in predicting bank failures, but the use of Principal Component Analysis (PCA) can enhance machine learning's performance, particularly the K-Nearest Neighbors (KNN) model.
20 sharesSource ↗
A study using machine learning reveals that personality traits, financial literacy, and behavioral factors influence participation in Private Pension Systems, with extraversion and advanced pension literacy being key factors; it recommends customized pension literacy programs to boost participation.
15 sharesSource ↗
A study reveals that machine learning models using non-traditional data are more effective in predicting credit losses and defaults, particularly during economic shocks.
30 sharesSource ↗
A hybrid model combining LSTM and DE algorithms significantly enhances the accuracy of stock market predictions, as shown in the Bombay Stock Exchange.
26 sharesSource ↗
The improved binary crayfish optimization algorithm (IBCOA) boosts feature selection in data mining and machine learning by reducing dimensionality and improving classification accuracy.
19 sharesSource ↗
Text mining of newspaper articles reveals that negative narratives about the future significantly impacted the spread of Greek bonds during the 2009-2015 debt crisis.
17 sharesSource ↗
Deep transfer learning models like BERT can greatly enhance the analysis of large political text corpora in social sciences research by reducing the need for extensive manually annotated training data.
17 sharesSource ↗
The article advocates for the use of machine learning in international business research to drive significant advancements in the field.
16 sharesSource ↗
The study applies machine learning to predict stock prices in the Malaysian market, influencing investors' strategies and risk management.
15 sharesSource ↗
The study presents Housing GANs, a model that generates realistic housing data, addressing the issue of scarce data in housing market research.
14 sharesSource ↗
The paper examines the sentiment of private investors on online platforms and its effect on herd behavior in the Russian stock market.
14 sharesSource ↗
The paper shows the effectiveness of Random Regression Forests for optimal lag selection in data series, outperforming other methods.
14 sharesSource ↗
The research compares risk tolerance of institutional investors in China and the US, revealing that US investors are more risk-averse.
15 sharesSource ↗
The study indicates that extended exposure to high volatility leads to underestimation of volatility, which can be exploited for stock return predictability.
13 sharesSource ↗
The paper demonstrates that cross-sectional global factor model provides better long-term country equity premium forecasts than time-series prediction models in the US and other developed markets.
10 sharesSource ↗
The research finds a significant correlation between Global Consciousness Project data and the S&P 500 Volatility Index, suggesting its potential in predicting market sentiment.
10 sharesSource ↗
Research suggests that 1.23% of the world's GDP is laundered yearly, negatively affecting economic and financial indicators, except for inflation.
10 sharesSource ↗
Insider trading by individuals holding multiple directorships can provide valuable insights into their entire portfolio, with unsold stocks outperforming unbought ones.
8 sharesSource ↗
Investment strategies that incorporate momentum and valuation styles in leveraged loans yield significant returns, indicating missed profit opportunities for credit managers not using these methods.
6 sharesSource ↗
A study in Kenya reveals financial service access disparities across generations, with Generation Y having more access than Generation X, and rural residents and women having less access.
1 sharesSource ↗
In Iran, companies with higher agency costs tend to hire lower-tier auditors, but this is less likely if the board has more financial experts; the quality of the internal control system doesn't influence this trend.
1 sharesSource ↗
The general machine-learning papers the letter carried in 2023-25.
20 items
African Language Model: InkubaLM, a language model for African languages, is introduced, performing well in tasks like machine translation and sentiment analysis despite limited resources.
177 shares31 citations todaySource ↗
A new deep learning framework generates high-quality medical images and segmentation masks, improving model performance in data-limited situations and reducing the need for extensive training data.
18 shares20 citations todaySource ↗
The study uses Bayesian optimization to solve complex, two-stage stochastic programs, showing better results than standard methods and providing a more efficient solution to uncertain optimization problems.
9 shares2 citations todaySource ↗
Pisco, an email visual similarity detection method, is proposed to enhance spam detection by identifying visually similar emails and reused email kits, improving defense against evasion techniques and zero-day attacks.
8 sharesSource ↗
Video Understanding Framework: BREASE, a model for understanding long-form videos that mimics episodic memory accumulation and reinforces it with semantic knowledge, is introduced, achieving top performance in both zero-shot and fully-supervised settings.
6 shares13 citations todaySource ↗
SYNTHEVAL is a testing framework that uses large language models to generate tests for evaluating NLP models, particularly in sentiment analysis and toxic language detection.
6 shares8 citations todaySource ↗
A new approach using large language models improves multi-talker automatic speech recognition, outperforming traditional methods on real-world datasets.
6 shares19 citations todaySource ↗
SelectTTS, a new method for multi-speaker text-to-speech, uses self-supervised learning to capture speaker characteristics, reducing model complexity and training data while improving performance.
5 shares1 citation todaySource ↗
A study comparing AppStore and large language model approaches for refining app features finds both are effective, but LLMs excel in novel unseen app scopes, emphasizing the role of human analysts.
4 shares16 citations todaySource ↗
A new framework reduces bias in the estimation of graphical models related to protected attributes, effectively balancing bias mitigation and model performance.
4 shares6 citations todaySource ↗
The article discusses the influence of foundation models on the music industry, emphasizing their potential in music generation and the need for ethical research on issues like transparency and copyright.
405 shares62 citations todaySource ↗
The study introduces a framework that uses adversarial math problems to enhance the performance and robustness of Large Language Models (LLMs).
50 shares13 citations todaySource ↗
The research explores the performance of adaptive gradient optimizers without the square root, showing they maintain performance on transformers and improve on convolutional architectures.
45 shares27 citations todaySource ↗
3D Scene Reconstruction: The paper presents ReconX, a new 3D scene reconstruction method using pre-trained video diffusion models, proving its superior quality and generalizability over existing methods.
33 shares153 citations todaySource ↗
The article introduces LR-QAT, a memory-efficient training algorithm for Large Language Models (LLMs), demonstrating its effectiveness and memory efficiency over common post-training quantization methods.
30 shares60 citations todaySource ↗
A new quantum-classical deep learning model has been created to predict drug binding affinity, increasing accuracy by 6% and providing more stable performance than traditional models.
25 shares16 citations todaySource ↗
Client filtering, a new method to enhance model generalization and client participation in federated learning, has been proposed, leading to better learning efficiency, quicker convergence, and up to 10% increased test accuracy.
23 shares7 citations todaySource ↗
OmniRe, a new 3DGS framework, has been launched for efficient reconstruction of dynamic urban scenes from device logs, surpassing previous methods and allowing real-time simulation of reconstructed scenarios.
20 shares188 citations todaySource ↗
CWGrasp, a new method for creating 3D whole-bodies that grasp objects realistically, has been developed, offering improved performance, reduced runtime and budget, and the capability to manage both right- and left-hand grasps.
18 shares6 citations todaySource ↗
The paper explores the responsible requirements of generative AI models, highlighting five key considerations and emphasizing the significance of responsible GenAI across various fields, with the aim to shed light on practical safety-related issues.
17 shares14 citations todaySource ↗
Papers that shipped their code, from the Papers with Code feed (2023-25).
5 items
The article emphasizes the need for research on human interaction with large language models due to their widespread use in different applications.
324 shares
The second article reveals that WavTokenizer, despite having fewer tokens, provides high-quality reconstruction and superior UTMOS scores, indicating a higher level of semantic information.
312 shares
The paper introduces a new method, Attributeguided Metric Distillation (AMD), to help interpret existing ReID models.
86 shares
The article explores the use of channel prediction to decrease feedback or estimation overhead in massive multi-input multi-output systems.
55 shares
The article proposes that activation sparsity can enhance the inference speed in large language models by minimizing the computation and memory movement required for matrix multiplications.
48 shares
Repositories the letter featured.
10 items
Python for Algorithmic Trading Cookbook is a new book by Packt that provides guidance on using Python in trading algorithms.
71 shares
A new Python package has been launched, specifically tailored for algorithmic trading, with features based on ICT's smart money concepts.
313 shares
A new alpha mining framework has been introduced, providing a formula-based approach for quantitative investment.
28 shares
The ISLP package offers data and code for labs, facilitating practical learning and experimentation.
100 shares
The VolGAN paper, along with its code, presents a generative model for creating arbitrage-free implied volatility surfaces.
19 shares
The article explores a Python library with various financial technical indicators.
651 shares
The piece presents a Python package for tuning hyperparameters and selecting features in Gradient Boosting Models.
558 shares
The article showcases a MLX port of FLUX, based on the Huggingface Diffusers implementation.
363 shares
The article describes a programming framework that assists in selecting models for time-varying parameters.
150 shares
The piece introduces a tool that enables users to interact with their documents using RAG-based technology.
2,477 shares
Industry news: funds, hiring, markets and regulation.
20 items
Goldman Sachs' Petershill Partners has sold an 18% stake in LMR Partners for up to $258m to address asset valuation accuracy concerns.
7 shares
In July 2024, hedge funds almost matched the performance of global equities, with Preqin's All Hedge Fund Index increasing by 1.61%.
6 shares
Morgan Stanley is targeting hedge funds to reclaim its leading position in stock trading, which it lost to Goldman Sachs following the 2021 Archegos Capital Management collapse.
5 shares
Due to market volatility in early August, Millennium Management, Balyasny Asset Management, and BlueCrest Capital Management suspended at least six traders and their investment pods.
5 shares
The Securities and Exchange Commission has settled charges against Galois Capital Management, a former Florida-based cryptocurrency hedge fund, for not protecting client assets.
5 shares
Business-critical applications firm 3forge has received an investment from Morgan Stanley.
5 shares
Kite Lake Capital Management and Astaris, London-based hedge funds, are closing to new investors due to significant capital outflows.
5 shares
TS Imagine has launched Platform 3.0, an upgraded service for financial institutions specializing in trading portfolio and risk management.
4 shares
Kirin Holdings' majority stake purchase in Fancl Corp has been delayed due to share acquisitions by hedge fund MY.Alpha Management HK Advisors.
4 shares
BlueCrest Capital Management has entered the foreign exchange market with the launch of a new currency trading platform, BlueX.
3 shares
Appital has upgraded its peer-to-peer price discovery and liquidity sourcing technology to improve liquidity and pricing dynamics for large ADV orders among its buyside clients.
3 shares
CoinShares' weekly report reveals a $305m outflow from digital asset investment products last week, suggesting a negative sentiment across various providers and regions.
3 shares
Hedge funds are helping banks manage risks associated with deal contingent trades, a complex derivative used to hedge against market fluctuations during major mergers and acquisitions, according to an IFR report.
3 shares
The University of Houston System's endowment management committee has approved a $20m investment in the Broad Reach Master Fund, a global macro hedge fund targeting emerging markets, as per Pensions & Investments.
2 shares
The article explores the viewpoints and ideas of Giuseppe Paleologo.
2 shares
Cheah Cheng Hye is stepping down as CoChairman of Value Partners Group Ltd due to increased control from GF Securities Co.
2 shares
The article claims that certain exits are unrelated to each other.
2 shares
A top technologist from Goldman has left to join a hedge fund.
2 shares
Elliott Investment Management has bought 10% of Southwest Airlines' common stock, allowing them to convene a special meeting.
2 shares
Episodes on markets, quant methods and economics.
9 items
Dan Russo talks about the significance of volatility as a trading signal, market breadth, sector performance in predicting market downturns, and the role of technical analysis and systematic investing in maintaining portfolio stability.
14 shares
Former Wells Fargo employee, Agus Sudjianto, shares insights on his work with machine learning models, risk management model validation, and fostering a productive work culture in a discussion with Dimitri Bianco.
11 shares
Terisa Roberts and Sarah Murphy highlight the importance of real-time customer decision-making in financial services, stressing on the need for quick, reliable, and automated decisions in the credit value chain.
5 shares
A new podcast named The Best Investment Writing has been introduced, showcasing top-notch investment research narrated by esteemed money managers and investment researchers.
4 shares
Dr. Manuela Moschella's book delves into the evolution of central banks since the 1970s, focusing on their unique monetary policies and political aspects.
2 shares
In a podcast, J. Doyne Farmer, a leader in complexity economics, talks about chaos theory, economic modeling, and his professional journey.
2 shares
Stefano Grasso offers practical tips for investors on handling and profiting from corporate actions in a conversation with Guillaume Roux-Chabert.
1 shares
Charlie Marsh discusses the features of uv, Astral's new major release aimed at unifying Python packaging, in a podcast episode.
0 shares
The recent decline in AI investments is scrutinized, pondering if it marks the end of the AI boom or is merely a temporary downturn.
0 shares
Posts from quant and economics blogs and newsletters.
8 items
The article offers a guide on a trading strategy centered on the EURUSD price.
2 shares
The article explores the increasing use of technical indicators and analysis tools by traders.
1 shares
The article reports on the WazirX hack in July, resulting in a loss of $230 million in crypto.
0 shares
Dr. Tejaswi Nadahalli investigates the July WazirX hack, where $230 million in crypto was stolen.
0 shares
Dr. Tejaswi Nadahalli analyzes the $230 million loss from the July WazirX crypto exchange hack.
0 shares
The July security breach at WazirX, resulting in a $230 million theft, is explored by Dr. Tejaswi Nadahalli.
0 shares
Dr. Tejaswi Nadahalli discusses the July WazirX hack that led to a $230 million loss.
0 shares
Dr. Tejaswi Nadahalli studies the July WazirX hack, where a security loophole led to a $230 million theft.
0 shares
Posts from quant researchers on X.
7 items
Sina Seyfi's recent study emphasizes Momentum and Volatility as key factors in determining long-term winners and losers in the US stock market.
3 shares
The newest research summary on quant investing discusses various topics such as bond ETFs predictability, cryptocurrency, equities' technical trading rules, macro, option pricing, etc.
2 shares
A recent article investigates potential alpha sources using analyst forecasts, providing an in-depth analysis of the latest research.
2 shares
The article reviews two studies on Sharpe ratios, examining the inaccuracies in their estimation and issues with time aggregation.
2 shares
The writer advocates for high-volatility investment options in his recent article.
0 shares
The article offers a novel viewpoint on the betting-against-beta factor.
0 shares