Quant LetterNo. 53
142 items across 10 sections, as sent to readers on 12 June 2024. Paper titles open their ML-Quant page; ↗ goes to the source.
Quantitative-finance and ML-for-finance preprints from arXiv.
5 items
The article discusses the use of an ARCH model to understand the pricing of a financial derivative product, the Take-or-Pay swing option, and introduces a method to understand the value function's monotonicity.
3 sharesSource ↗
The paper introduces an algebraic framework, using physics and stochastic processes, for modeling limit order books to simulate market scenarios and understand the impact of trader behavior on market dynamics.
3 shares1 citation todaySource ↗
The article outlines the winning approach of the SIAG/FME Code Quest 2023 competition to implement an Automated Market Maker in Python, focusing on minimizing the Conditional Value at Risk to reduce computational complexity and improve solution accuracy.
2 sharesSource ↗
A machine-learning study predicts that China will match the US, UK, and EU in scientific leadership by 2027-2028. However, it is not expected to reach parity in terms of per-collaborator until after 2087.
5 shares2 citations todaySource ↗
The article discusses the gas fee in Ethereum's blockchain, considering the resources needed for transactions and the effect of gas price volatility. It suggests that pricing and trading forwards could help manage cost fluctuations.
6 shares16 citations todaySource ↗
Working papers in finance and economics from SSRN.
31 items
A Turkish study found that raising personal credit card limits significantly changes household spending, especially among those with already high limits.
20 sharesSource ↗
Google's Federated Learning method allows insurance companies to predict claim frequency without sharing sensitive data, maintaining privacy.
4 sharesSource ↗
Financialisation has increased the volatility transmission from equity markets to commodity futures markets, according to a study.
2 sharesSource ↗
A proposed system uses machine learning to detect phishing websites via their digital certificates, improving upon slower traditional methods.
4 sharesSource ↗
A new method for extracting key information during the Traveling Salesman Problem local search process has been proposed, using a combination of factors to evaluate edge quality.
3 sharesSource ↗
A Deep Q-Network (DQN) machine learning model has been created to predict cyberattacks in software-defined networking with greater accuracy and less training time.
2 sharesSource ↗
The MagicGrid methodology has been introduced to build a cyber-physical system and use RAAML for reliable risk management in digital supply chain transformation.
2 sharesSource ↗
Canadian pension funds are investing less in domestic assets due to a lack of strategic assets in Canada and increased availability of such assets abroad.
4 sharesSource ↗
Speech Attribution with CNNs: EnCoNet, an ensemble-based synthetic speech attribution algorithm, has been proposed to distinguish between human and AI-generated speech using lightweight neural networks.
2 sharesSource ↗
A divisible project portfolio selection model has been developed to help businesses select the right project portfolio and achieve strategic goals, considering risk preference and maximum net present value.
2 sharesSource ↗
The article explores the use of machine learning, particularly the Kmeans method, in the insurance industry.
3 sharesSource ↗
The study exposes financial risks in Malawian holdings due to unregulated corporate bonds and flawed accounting practices.
2 sharesSource ↗
The research introduces a new equation for unsupervised data separation to improve clustering models in machine learning.
2 sharesSource ↗
The article evaluates the effectiveness of passive and active management strategies in long-term savings using historical data from nine asset classes.
2 sharesSource ↗
The paper proposes a solution to handle complex manufacturing data in process control through a simple aggregation technique.
2 sharesSource ↗
The study shows that professional training boosts the positive effect of Enterprise Risk Management on employee performance in the UAE Ministry of Interior.
2 sharesSource ↗
The research examines the use of Quantpedia's Composite Seasonal Calendar Strategy in options trading to balance risk and returns through the Volatility Risk Premium.
2 sharesSource ↗
The article studies the volatility of stock markets in China, Brazil, India, Hong Kong, and Japan during the COVID-19 pandemic, indirectly considering the Black Swan theory's implications.
2 sharesSource ↗
The study creates a logistic regression model to evaluate the impact of the COVID-19 pandemic on nonfinancial firms' default probability, emphasizing the need for robust predictive models.
2 sharesSource ↗
The research examines the Panamax market behavior, focusing on how expectations and time lags affect freight market decisions and Panamax vessels' earnings.
2 sharesSource ↗
The SEC's expected approval of Ethereum spot ETFs could democratize access to the asset class, attract more investors, and increase liquidity.
5 sharesSource ↗
The Markowitz model is a useful tool for selecting investment portfolios that outperform market indices, offering larger returns with lower risk.
6 sharesSource ↗
A study using high-frequency trading data provides insights into asset pricing, transaction costs, investor liquidity asymmetry, and seasonality effects.
3 shares1 citation todaySource ↗
Clearing and settlement infrastructures are crucial in securities and derivatives markets, with US and EU regulations focusing on cross-border coordination and market linkages.
5 sharesSource ↗
A study on the effects of price staleness on volatility functionals estimation in financial markets proposes a consistent global estimator that corrects bias.
2 sharesSource ↗
Antitrust laws fail to account for acquisitions of firms with high intangible capital, leading to unreported deals that increase equity values and potentially foster anti-competitive behavior, especially in the tech and pharmaceutical sectors.
7 sharesSource ↗
Market disruptions in the euro area futures market result in reduced trading activity, liquidity shortages, and price deviations, with dealers pulling out and clients trading amongst themselves, leading to significant mispricing.
3 shares2 citations todaySource ↗
Rhetorical engineering, which uses identity, emotion, and intention, can significantly alter a client's perception of a company's strengths and weaknesses, thereby influencing their choice of companies for portfolio inclusion.
3 sharesSource ↗
Smart beta indices can be tailored to address sustainability biases while maintaining the Sharpe ratio, with a stronger ESG tilt correlating with improved sustainability, in contrast to the high emissions seen in Value and Quality factors.
3 sharesSource ↗
The mean squared error (MSE) is a useful statistical loss function in portfolio selection, even though it is often used in decision problems with different loss functions.
2 sharesSource ↗
The study explores how investors' negative views on pollution-related disasters can result in increased equity premiums, even when the rate of substitution over time is marginally less than 1.
2 sharesSource ↗
Economics working papers from RePEc's NEP field reports.
21 items
The research finds China's copper futures market on the Shanghai Futures Exchange to be efficient with no significant arbitrage opportunities.
31 sharesSource ↗
The study suggests that the rise in algorithmic trading with the NYSE Hybrid Market introduction has enhanced market efficiency by using both market and limit order flows information.
29 sharesSource ↗
The research introduces a periodic regime-switching model to better predict stock market volatility, taking into account information from trading and non-trading periods.
27 sharesSource ↗
The article introduces a new, faster test procedure for alpha in linear factor pricing models, which is valid even with a larger number of securities and can handle some pricing errors.
23 sharesSource ↗
Low beta anomaly yields higher returns for hedge funds than mutual funds, particularly for less skilled managers.
22 sharesSource ↗
Machine learning can predict exchange traded funds returns, with some models surpassing the buy & hold strategy.
22 sharesSource ↗
No Free Lunch: Diversification can lower both risk and returns if expected returns are predictable, challenging the idea that diversification is always beneficial.
17 sharesSource ↗
Gold effectively reduces investor risk in the Pakistani stock market, particularly during bullish markets and after the COVID-19 pandemic.
16 sharesSource ↗
The article introduces a new machine learning method for predicting one-day-ahead scenarios for portfolio optimization, resulting in more accurate forecasts and less risky portfolios.
27 sharesSource ↗
The study combines traditional forecasting and machine learning to predict inflation in an emerging economy, with machine learning proving more effective, especially when foreign exchange reserves are included.
26 sharesSource ↗
The research compares linear regression and machine learning for modeling credit spread changes, with machine learning showing superior performance due to its ability to handle complex non-linearities.
19 sharesSource ↗
The study explores the use of the Large Language Model, ChatGPT, in portfolio management, revealing that it selects a more diverse range of assets that outperform random selections.
18 sharesSource ↗
The research applies machine and deep learning models to predict metal futures in commodity markets, finding that the efficiency of different models varies significantly based on the metal, sample period, and inputs.
17 sharesSource ↗
Machine learning, particularly regression-tree methods, can improve the accuracy of cryptocurrency market options pricing compared to traditional models.
45 sharesSource ↗
Machine learning methods offer better performance and insights in modeling corporate bond recovery rates than traditional methods.
39 sharesSource ↗
Machine learning models, specifically random forests and bagging, are superior in predicting S&P 500 returns using volatility indices.
26 sharesSource ↗
Machine learning can predict return in cryptocurrency markets, primarily based on simple factors like market price and momentum.
25 sharesSource ↗
The Random Forest algorithm is the best machine learning model for predicting dividend payouts in the Vietnamese stock market, enhancing market transparency.
23 sharesSource ↗
Scientists have created a deep learning model that enhances the accuracy of Bitcoin's volatility predictions, offering investors a dependable risk warning system and trading strategy.
21 sharesSource ↗
A new deep learning financial strategy has been developed to manage inventory risks due to fluctuating commodity prices, offering significant economic advantages for businesses.
16 sharesSource ↗
The study finds that companies with more gender diversity on their boards are perceived as having less future risk, indicating that diverse boards can greatly affect a firm's risk management and financial performance.
6 sharesSource ↗
The general machine-learning papers the letter carried in 2023-25.
17 items
Circuit breakers is a new method proposed to stop AI systems from causing harm or producing damaging outputs, even under unseen powerful attacks.
486 shares345 citations todaySource ↗
The performance of diffusion models in language modeling has been enhanced by using an effective training recipe and a simplified objective, setting a new standard among diffusion models.
180 shares850 citations todaySource ↗
3D Scene Reconstruction: Flash3D, a new method for scene reconstruction and novel view synthesis from a single image, has been introduced, achieving top results on multiple datasets due to its efficiency and generalizability.
113 shares139 citations todaySource ↗
Weight Quantization: A new weight quantization method has been developed to reduce the size of diffusion-based image generation models by 7.9 times, while also improving the quality of the generated images.
86 shares47 citations todaySource ↗
The verbalized machine learning (VML) framework uses large language models parameterized by text prompts to solve traditional machine learning problems, providing easy encoding of inductive bias, automatic model class selection, and interpretable learner updates.
58 shares20 citations todaySource ↗
A new method for creating high-quality videos in a single step using adversarial training and pre-trained video diffusion models is proposed, reducing computational costs.
41 shares42 citations todaySource ↗
GLACE, a new technique for scene coordinate regression in large-scale scenes, uses feature diffusion and a position decoder for improved output positions.
29 shares71 citations todaySource ↗
DiffRIR is a new framework that uses a sparse set of recordings and a planar reconstruction to recreate the acoustic characteristics of a space.
16 shares18 citations todaySource ↗
PGSR, a new method for high-quality surface reconstruction and rendering, uses an unbiased depth rendering method and a camera exposure compensation model to handle scenes with large illumination variations.
14 shares415 citations todaySource ↗
A new neural network framework is proposed for solving forward and inverse problems in ordinary differential equation systems, demonstrated through a case study on mosquito population dynamics.
14 shares18 citations todaySource ↗
ArtPrompt, a new ASCII art-based attack, exploits vulnerabilities in large language models, bypassing safety measures and causing unwanted behaviors.
9,360 shares294 citations todaySource ↗
The Self-Improving Robust Preference Optimization (SRPO) framework aims to improve AI alignment robustness, using self-improvement to optimize policy independently of the training task.
381 shares25 citations todaySource ↗
Retrieval Augmented FineTuning (RAFT) is a new training method that enhances large language models' ability to answer domain-specific questions by training them to ignore irrelevant documents and cite relevant ones.
260 shares401 citations todaySource ↗
The article discusses Visual AutoRegressive modeling (VAR), a new image learning method that outperforms diffusion transformers in terms of speed, image quality, and scalability.
257 shares1,169 citations todaySource ↗
The paper introduces SafeDecoding, a strategy to protect large language models from jailbreak attacks, maintaining the quality of responses to user queries while reducing attack success rate.
103 shares300 citations todaySource ↗
The survey reviews the use of large language models in tabular data modeling, summarizing key techniques, metrics, datasets, models, and suggesting areas for future research.
61 shares272 citations todaySource ↗
The paper suggests a chronological approach to defining learning tasks using the PAC learning framework, aiming to standardize definitions for various learning paradigms.
60 shares22 citations todaySource ↗
Papers that shipped their code, from the Papers with Code feed (2023-25).
11 items
Mobile Navigation: The article discusses the complexity of mobile device operation due to task progress navigation and focus content navigation challenges, caused by the single-agent architecture.
2,047 shares
Less Scheduled Road: The article highlights a new approach that surpasses existing learning rate schedules by avoiding them entirely, demonstrating superior performance in various problems, including large-scale deep learning.
1,421 shares
Jailbreak Prompts Evaluation: The study aims to support the research community and LLM vendors in ensuring safer and regulated LLMs.
1,267 shares
MoE and Dense SpeedAccuracy: The article reevaluates settings using step time as a measure of model complexity and determining the total compute budget under Chinchilla compute-optimal settings.
1,144 shares
Gsplat Library Supplement: The report explains the mathematical aspects of the gsplat library, a toolbox for efficient differentiable Gaussian splatting as proposed by Kerbl et al.
1,123 shares
The study explores the relationship between autoencoder size and sparsity, using specific methods to establish clear scaling laws.
405 shares
The research shows that the suggested MatMul-free models perform as well as advanced Transformers, but require less memory for up to 2.7B parameters.
223 shares
The article introduces SimulS2ST, a system that allows real-time translation of speech, providing target speech outputs while receiving continuous speech inputs.
200 shares
The article explores the difficulties in producing high-quality scientific figures and offers a supporting code.
159 shares
The second article presents a new technique called Buffer of Thoughts (BoT) to enhance the performance of large language models, along with a code.
118 shares
The third article emphasizes the benefits of using polygon mesh for 3D data representation, such as flexibility and storage efficiency, and includes a related code.
81 shares
Repositories the letter featured.
10 items
Portfolio Backtesting Engine: An Open Source Portfolio Backtesting Engine for Everyone is a tool for testing investment strategies with historical data.
886 shares
Trade Implementation: Our codebase trials showcases a new pair trading method using hierarchical reinforcement learning with real-world stock data.
99 shares
Data Preprocessing Library: NVTabular is a library for processing large tabular data, useful for training deep learning-based recommender systems.
1,018 shares
Technical Analysis Indicators for polars provides a guide for using data analysis indicators on the Polars platform.
55 shares
Implementation of PIN Probability of Informed trading on AShare daily public data uses a new estimation method to calculate informed trading probability with AShare's public data.
28 shares
Multiagent Applications Framework: Langflow is a Python-based open-source framework designed for developing multiagent and RAG applications.
18,728 shares
Python Causal Impact is a tool developed using TensorFlow Probability, inspired by Google's R Package.
584 shares
Graphic notes provide a summary of Gilbert Strang's book, Linear Algebra for Everyone.
15,210 shares
Packt Publishing offers a guide on how to build low latency applications using CPP.
313 shares
This repository provides a comprehensive list of links to courses and resources related to Artificial Intelligence (AI).
4,649 shares
Industry news: funds, hiring, markets and regulation.
20 items
The HFRI Fund Weighted Composite Index increased by 1.3% in May, driven by equity hedge and event-driven strategies, with relative value arbitrage strategies also contributing.
8 shares
Bloomberg’s Data Licence Plus is now accessible via Arcesium's platform, providing mutual customers with improved data management capabilities.
7 shares
BitMEX, a crypto derivatives exchange, will boost the leverage of its ETHUSD perpetual swap to 200x, following US approval of ether spot ETFs.
6 shares
GoldenSource and Snowflake have introduced GoldenSource Omni, an app aimed at simplifying financial reference data integration for investment managers.
6 shares
Allison DiClemente has been named Director of Capital Introduction at Marex, a firm that offers liquidity market access and infrastructure services to hedge funds and other clients.
5 shares
Bloomberg reports that hedge funds are increasingly shorting bitcoin, reaching an all-time high in short positions.
4 shares
Rokos Capital Management and Andurand Capital Management have profited from betting on rising copper prices, according to Bloomberg.
4 shares
Omega Systems has expanded its services by acquiring Amnet Technology Solutions and Cloudpath, enhancing its IT, cybersecurity, and regulatory compliance offerings.
3 shares
A Citadel Securities software engineer is spending her free time building a ladder.
3 shares
Yao Hua Ooi, Principal and Global Co-Head of macro strategies at AQR Capital Management, has left the company, as reported by Pensions & Investments.
3 shares
Dr Brendan Campbell is appointed as the new CEO of AlphaGrep Global Capital, effective immediately.
3 shares
Premialab hires Jens Janke, former Head of Risk, to lead its Data & Risk Solutions development.
3 shares
Gregoire Tournant, ex-Allianz SE hedge fund manager, pleads guilty to fraud charges linked to billions in investor losses.
3 shares
Michelle Moran is appointed as a partner in Faegre Drinker's investment management group in London, overseeing all aspects of fund management.
3 shares
The article suggests that moving to the buyside does not automatically imply a transition to a hedge fund.
2 shares
Hedge funds are speculating on a decline in Eurozone government bonds, anticipating the European Central Bank's difficulty in further reducing interest rates this year.
2 shares
The US Securities and Exchange Commission's update to Regulation SP is causing financial institutions to reevaluate their customer data management practices.
2 shares
CoinShares reports that digital asset investment products saw a $2bn inflow last week, bringing the five-week total to $4.3bn.
2 shares
ET Now Hangouts discusses the role of technology in financial strategies, specifically highlighting quant investing with Sonam Srivastavas.
2 shares
EEX Group's global power markets experienced a 47% growth in May, primarily driven by the expansion of French, Spanish, and Japanese power derivatives markets.
2 shares
Episodes on markets, quant methods and economics.
10 items
Macro Endgame: Macro investor Alfonso Peccatiello discusses the instability of the current financial system and strategies to handle increasing macro volatility in a ReSolve team episode.
14 shares
The Quantum Computing in Quant Finance Conference discusses predicting recessions using quantum machine learning and probability distribution classification.
11 shares
Bull Market: Data reveals that 70.4% of Russell 3000 index stocks are trading below their 2021 highs, indicating a potentially unstable bull market.
9 shares
Litigation Finance: Eva Shang, CEO of Legalist, talks about litigation finance, AI in deal sourcing, and the launch of a Government Receivables strategy on The Meb Faber Show.
8 shares
JP. Morgan economists discuss the H2 outlook for Emerging Markets and the potential impact of the US election on EM economies and fixed income markets.
8 shares
CoFounder at CoVenture: Ali Hamed, CoFounder of CoVenture, shares his transition from a baseball player and agriculture major to a key figure in the startup ecosystem on a podcast hosted by Steve Curley and Chris Cannon.
7 shares
The 'At Any Rate' podcast discusses the Frontier Markets Fixed Income Conference in London, featuring insights from various financial sector speakers.
7 shares
A podcast features Raphael Brun-Aguerre, Aditya Chordia, and Meera Chandan discussing the implications of the upcoming French lower house elections on FX and rates markets.
5 shares
Carry Trade Deleveraging: Patrick Locke and James Nelligan from Global FX Strategy assess the impact of the week’s FX carry trade deleveraging and share their views on FX after the ECB, BoC, and US payrolls in a podcast.
5 shares
Paul Katz of the Jain Family Institute talks about his work in developing tools to help Brazilian leaders increase social wealth, emphasizing the role of history, research, and networking in non-academic success.
4 shares
Posts from quant and economics blogs and newsletters.
10 items
The article provides an in-depth analysis of the process of adjusting and correcting sample covariance estimates and the representation of linear dependency in covariance matrices.
5 shares
The article investigates the representation of linear dependency in covariance matrices and the process of modifying sample covariance estimates.
5 shares
The article delves into the complex topic of sample covariance estimates and their role in representing linear dependency in covariance matrices.
5 shares
The article scrutinizes the representation of linear dependency in covariance matrices and the methodology of adjusting sample covariance estimates.
5 shares
The article concentrates on the process of adjusting sample covariance estimates and the way covariance matrices depict linear dependency.
5 shares
The article talks about the correction of sample covariance estimates and how linear dependency is represented in covariance matrices.
5 shares
The piece reveals that the inverse of a covariance matrix shows the conditional linear dependence between variables.
5 shares
The post discusses the mathematical side of covariance matrices and their inverses, and how they affect linear dependency.
5 shares
The article explains the role of covariance matrices in showing linear dependencies, particularly in wide data matrices.
5 shares
The piece emphasizes the significance of understanding the entries of the inverse of the covariance matrix in terms of conditional linear dependence.
5 shares
Posts from quant researchers on X.
7 items
Man Group explores the possibility of generating excess returns through climate-focused investing.
3 shares
A new study criticizes the traditional mean-variance approach and recommends using hierarchical clustering methods for building investment portfolios.
2 shares
A recent study introduces a corporate bond valuation model that effectively predicts yield fluctuations and future bond returns.
2 shares
The article explores the substantial influence of nonfarm payrolls on macroeconomic yields.
1 shares
The study presents a novel tree-based asset clustering technique for predicting high-return stocks in various macroeconomic conditions.
1 shares
IBM has launched a new open-source ZeroShot Forecasting Model, Tiny Time Mixers (TTM), for time-series analysis.
0 shares
Terrence Tao, a Fields medalist, discusses the revolutionary effect of AI on the field of mathematics.
0 shares