Quant LetterNo. 52
151 items across 11 sections, as sent to readers on 5 June 2024. Paper titles open their ML-Quant page; ↗ goes to the source.
Quantitative-finance and ML-for-finance preprints from arXiv.
9 items
The study by Milionis et al. (2023) explores the expected losses and time-to-fill in Dutch auctions in decentralized finance. It considers factors such as starting price, volatility, decay rate, and average interblock time, and examines the balance between execution speed and quality.
3 sharesSource ↗
The article introduces the Distributional Refinement Network (DRN), a model that enhances predictive performance and interpretability in actuarial modelling by merging a baseline model with a flexible neural network.
3 shares1 citation todaySource ↗
A study finds the success of debt recycling, a strategy involving a second loan to pay off a mortgage faster, depends on factors like the initial mortgage-to-equity ratio and economic conditions.
7 shares2 citations todaySource ↗
A new deep learning model, HLOB, has been developed for predicting Limit Order Book mid-price changes, outperforming nine other models and offering new insights into information distribution in Limit Order Books.
6 shares21 citations todaySource ↗
A new method for assessing the efficiency of broker execution algorithms in reducing execution costs for large orders has been introduced, offering a robust framework for choosing the most cost-effective brokers.
6 shares2 citations todaySource ↗
Researchers have found that Large Language Models (LLMs) pretrained on relevant arithmetic datasets can predict human behavior more accurately than many existing cognitive models.
5 shares12 citations todaySource ↗
Early findings suggest that programmable automated market makers with concentrated liquidity capabilities are not most effective when focused solely on the current price.
5 shares1 citation todaySource ↗
The queue-reactive model has been expanded to include order sizes, types, and arrival rates, creating simulated markets that closely mirror real markets and exhibit similar volatility.
3 shares2 citations todaySource ↗
The paper explores the worst-case scenarios of distortion risk metrics for general distributions with limited information, with applications for entropies, weighted entropies, and risk measures.
2 shares4 citations todaySource ↗
Working papers in finance and economics from SSRN.
34 items
The paper introduces a new method for estimating volatility using intraday price durations, which is more accurate than previous methods.
4 sharesSource ↗
The article highlights the accuracy and promptness of SP Global Market Intelligence’s Dividend Forecasting dataset in capturing dividend revisions, giving investors an edge.
44 sharesSource ↗
The study shows that retail transaction data can be used to generate credit scores for unbanked individuals, leading to higher credit card approval rates.
3 sharesSource ↗
The paper discusses how AI technologies like chatbots and personalized recommendation systems can improve service quality and boost sales on ecommerce platforms.
4 shares10 citations todaySource ↗
The note explains Rolloos' approximation for the covariance between an asset and its realised volatility, a tool that can help monitor the term structure of implied asset-realised volatility covariance and guide trading decisions.
4 sharesSource ↗
Machine learning and big data analytics are being utilized to gain insights from underused IT Ops ticket data.
2 sharesSource ↗
Machine learning is aiding in predicting the adsorption energies of molecules on platinum group metals, vital for catalytic reactions.
2 sharesSource ↗
Retail industry customer satisfaction can be significantly improved through quality analytics using machine learning algorithms.
4 sharesSource ↗
Despite global financial integration, consumption volatility in emerging markets has risen due to sensitivity to world real interest rates.
3 sharesSource ↗
Monetary policy impacts sector-specific credit card spending unevenly, decreasing spending in some areas while increasing it in others.
3 sharesSource ↗
The article analyzes the risks and benefits of dispersion strategies in investment, and how they can predict future investment opportunities.
9 sharesSource ↗
The article discusses the role of machine learning in intelligent search and information discovery, emphasizing the need for transparent and trustworthy AI models.
2 sharesSource ↗
The article studies how the volatility of unrecognized tax benefits can indicate changes in tax-related risk, including audit, compliance, and credit risks.
2 sharesSource ↗
The article details the creation of a Hands-Free Mouse system for physically disabled individuals, utilizing facial recognition and machine learning techniques.
5 sharesSource ↗
The article explores the tendency of American investors to overinvest in domestic stocks due to information asymmetry.
2 sharesSource ↗
The article explores the use of data science models in economics, detailing their applications, strengths, limitations, and impact on sectors like finance, retail, energy, and healthcare.
2 shares2 citations todaySource ↗
The essay traces the evolution of finance as an academic field from the 18th century, discussing its core assumptions, ideas, and influence on economics and social sciences.
2 sharesSource ↗
The study reveals a positive and significant impact of market efficiency on capital allocation efficiency in decentralized crypto asset exchanges.
2 sharesSource ↗
The paper suggests a supervised classification method for detecting anomalies in insurance companies' asset data reporting, emphasizing the advantages of machine learning in data quality management.
2 sharesSource ↗
The study shows that local natural disasters significantly influence global portfolio flows, leading to a reduction in net flows to equity mutual funds in affected countries, especially those with high climate risk exposure.
2 sharesSource ↗
The article presents the pathdependent volatility model that uses historical returns to calculate volatility, and introduces a neural network approximation of the VIX to address computational issues in pricing and calibrating SP 500 and VIX options.
73 sharesSource ↗
Wisconsin Investment Board's recent Bitcoin ETFs investment indicates a move towards mainstream cryptocurrency acceptance, despite debates about Bitcoin's role as an inflation hedge and concerns about market volatility and regulation.
3 sharesSource ↗
The article reveals that idiosyncratic volatility rises with the number of listed firms due to creative destruction, but no similar effect is found from nonlisted firms, suggesting a unique economic dynamism role for listed firms.
2 sharesSource ↗
A study found no evidence of price bubbles in the third or fourth trading period of the EU Emissions Trading System, suggesting it may be more efficient than thought.
11 sharesSource ↗
Gamification of stock trading can both harm retail traders by reducing returns and increasing volatility, and benefit the market by lowering costs and risks for liquidity providers.
2 sharesSource ↗
Companies with higher financial leverage see larger stock returns following a positive inflation surprise, as per a study of nonfinancial Eurozone firms from 2020-2022.
3 sharesSource ↗
A currency cross-section study found that only a small fraction of nontradable factors, mostly related to volatility, uncertainty, and liquidity, have a significant risk premium, not macro variables.
2 sharesSource ↗
A study reveals that smaller insurance firms heavily invest in finance bonds for diversification, but diversify into other sectors as they expand.
2 sharesSource ↗
A study from 2009 to 2020 shows that Euro-area households prefer pricier, lower risk-adjusted return mutual funds, particularly equity funds, and tend to follow past returns.
2 shares1 citation todaySource ↗
A new method for estimating the liquidity premium in Italian government bonds indicates that higher liquidity leads to higher prices, and the liquidity drop during the debt crisis was longer but less severe than during the Covid-19 pandemic.
2 shares1 citation todaySource ↗
The article assesses European financial data compilation methods, finding the direct method, though more expensive, aligns better with statistical standards.
2 sharesSource ↗
The research explores asset demand in a noniid returns model linked to an ESG factor, concluding that ESG assets offer risk diversification, but identifying the ESG factor's effect can offset the advantage.
2 sharesSource ↗
The paper examines Italian banks' bond issuance post the 2007-08 financial crisis, highlighting a decrease in bond issuance and a greater dependence on alternative funding and monetary policy measures.
2 sharesSource ↗
The study evaluates the impact of the ECB's Pandemic Emergency Purchase Programme (PEPP), revealing it significantly lowered the yield on eligible green-labelled bonds, indicating such programmes can boost climate-friendly investments.
2 sharesSource ↗
Economics working papers from RePEc's NEP field reports.
23 items
A study reveals a new model, considering non-trading and trading periods, has enhanced the prediction of stock market volatility in the Taiwan Futures Exchange.
27 sharesSource ↗
Not a Free Lunch: A study challenges the belief that diversification is risk-free, stating it can reduce returns if expected returns are estimated, based on historical returns of the S&P500 constituents.
17 sharesSource ↗
A study suggests gold is a strong hedge against risk in the Pakistani stock exchange, especially post-COVID-19, and recommends investors diversify their portfolios.
16 sharesSource ↗
Option signals from leveraged ETFs are strong predictors of the ETFs performance, especially during economic downturns, and can be used to create a trading strategy with an average abnormal performance of 1.13% per month.
15 sharesSource ↗
Companies led by CEOs who adopt value-investing strategies with their cash flows generally yield higher stock returns.
14 sharesSource ↗
The European Central Bank's unconventional monetary policies, like quantitative easing, boost stock prices and reduce market volatility.
14 sharesSource ↗
Despite increasing public information, the European Insider Trading and Market Manipulation Regulation has led to greater stock price volatility due to lower quality information.
12 sharesSource ↗
During the COVID-19 pandemic, Foreign Institutional Investors drove market returns in India, while Domestic Institutional Investors contributed to market volatility.
12 sharesSource ↗
The study finds ensemble boosting tree models to be more effective and economically valuable in predicting volatility in China's crude oil futures than traditional models.
20 sharesSource ↗
The role of Large Language Model, ChatGPT, in portfolio management is found to significantly improve asset selection and diversification compared to random selection.
18 sharesSource ↗
The research reveals that the performance of machine and deep learning models in forecasting metal futures varies significantly based on the metal choice, sample period, and inputs.
17 sharesSource ↗
The study finds the Decision Tree model to be the most accurate in predicting employee turnover based on various factors, using machine learning algorithms.
17 sharesSource ↗
The research finds the logistic regression model to be more effective than the random forest model in predicting landslides caused by rainfall events in India using machine learning models.
14 sharesSource ↗
The FS-RS-ML framework uses machine learning to enhance the accuracy of credit risk predictions for small and medium-sized businesses in supply chain finance.
25 sharesSource ↗
Machine learning can improve the precision of trade direction classification in corporate bond markets, with trading conditions and information environment being key factors.
22 sharesSource ↗
Machine learning can forecast global stock returns, with firm-specific traits remaining profitable even after considering transaction costs, particularly for long-term predictions.
18 sharesSource ↗
A data mining-based research classification model effectively categorizes scientific literature, as demonstrated in digital workplace accessibility studies for the disabled.
12 sharesSource ↗
The K-Nearest Neighbours algorithm outperforms the Naive Bayes method in storm warning accuracy, achieving a significant accuracy rate of 68.20%.
12 sharesSource ↗
A new deep learning financial strategy has been created that could enhance risk management and provide an average annual benefit of 1.21 million CNY for a typical Chinese aluminum firm.
16 sharesSource ↗
A novel deep learning framework for time series classification has been developed, improving the prediction of Bitcoin market crashes by 15.8% compared to six other models.
14 sharesSource ↗
The study shows that European companies with lower carbon emissions have higher stock returns, indicating that carbon risk affects equity prices.
14 sharesSource ↗
The research reveals that companies with diverse gender representation on their boards have lower expected risks, positively impacting risk management and financial performance.
6 sharesSource ↗
The study indicates that firms with gender diversity have less expected risk and improved financial performance, highlighting the importance of diverse corporate boards.
1 sharesSource ↗
The general machine-learning papers the letter carried in 2023-25.
20 items
The research identifies a link between state-space models and Transformers in deep learning, leading to the creation of a faster language modeling architecture, Mamba-2.
300 shares1,953 citations todaySource ↗
The paper suggests improved methods for training rectified flows in diffusion models for image and video generation, surpassing current distillation methods in low function evaluation settings.
49 shares118 citations todaySource ↗
The research introduces S3Gaussian, a self-supervised method for efficient 3D reconstruction of street scenes without 3D annotations, showing superior performance on the Waymo-Open dataset.
17 shares123 citations todaySource ↗
The Exploratory Preference Optimization (XPO) algorithm has been introduced for online exploration in Reinforcement Learning from Human Feedback (RLHF), potentially enhancing language model training.
7 shares115 citations todaySource ↗
A robot for organic farms uses deep learning to identify and treat weeds with a patented, organic method, achieving up to 98% accuracy.
6 shares3 citations todaySource ↗
Single Image 3D: Unique3D, a new image-to-3D framework, creates high-quality 3D meshes from single-view images, surpassing other similar frameworks in geometric and textural details.
6 shares181 citations todaySource ↗
Vocals and Motions: A new task uses a multimodal transformer model to generate 3D body motions and singing vocals from textual lyrics, ensuring a realistic blend of vocals and human motions.
5 shares7 citations todaySource ↗
Shape Editing: ParSEL, a system that allows controllable editing of high-quality 3D assets from natural language, uses Analytical Edit Propagation (AEP) to extend initial edits into complete editing programs.
5 shares14 citations todaySource ↗
Sequence Modeling: Mamba, a neural network architecture that doesn't use attention or MLP blocks, provides faster inference and better performance in language, audio, and genomics than Transformers.
1,099 shares9,205 citations todaySource ↗
Music Optimization: The DITTO framework allows control of pre-trained text-to-music diffusion models at inference-time, enabling music generation without fine-tuning.
252 shares97 citations todaySource ↗
A survey of self-evolution approaches in Large Language Models (LLMs) provides a conceptual framework and insights to improve these models and scale towards superintelligence.
161 shares83 citations todaySource ↗
A study shows a time asymmetry in Autoregressive Large Language Models' ability to learn natural language, explained by sparsity and computational complexity.
154 shares22 citations todaySource ↗
Bilingual LLM Series: MAP-Neo, an open-source bilingual language model with 7B parameters, performs comparably to state-of-the-art Large Language Models, with full details provided for reproduction.
115 shares88 citations todaySource ↗
Libraries or Librarians?: The article explores bibliotechnism, arguing that while large language models can generate new text and references, they may not necessarily possess consciousness or intelligence.
106 shares36 citations todaySource ↗
The paper introduces the Model Openness Framework, a system designed to rate machine learning models on their transparency and completeness to promote responsible AI practices.
74 shares57 citations todaySource ↗
The study presents a unified approach to reinforcement learning from human feedback for large language models, offering theoretical guarantees and practical effectiveness.
67 shares75 citations todaySource ↗
The research investigates the evolution of information handling by chemical automata, suggesting that simple self-replicators could evolve into complex molecular machines.
61 shares2 citations todaySource ↗
The paper introduces Nearest Neighbor Speculative Decoding, a language modeling approach that improves generation quality and speed by incorporating real-world text spans into language model generations.
59 shares28 citations todaySource ↗
Papers that shipped their code, from the Papers with Code feed (2023-25).
7 items
The article presents new models based on T5Base and T5Large that enhance pretraining speed without needing extra computational power.
10,006 shares
The article outlines the difficulties the finance sector encounters when incorporating Large Language Models due to proprietary data and specialized knowledge.
752 shares
The article discusses the increasing interest in Retrieval Augmented Generation techniques due to the evolution of Large Language Models.
624 shares
GPT4 Turbo: AutoCoder, a Large Language Model, has surpassed GPT4 Turbo and GPT4o in the Human Eval benchmark test, marking a significant achievement in AI development.
507 shares
Target Speech in Noisy Environments: A new enrollment interface has been introduced that captures a brief, noisy binaural sample of a target speaker when the wearer looks at them for a few seconds.
412 shares
Memory for Language Models: The article discusses the evolution of mammalian brains, focusing on their ability to store vast amounts of knowledge and integrate new information without forgetting, even in hostile and changing environments.
291 shares
Experts Architecture: Yuan 2.0M32, a variant of Yuan2.0 2B, utilizes a mixture of experts architecture with 32 experts, of which only two are active.
126 shares
Repositories the letter featured.
10 items
QIS package offers tools for visualizing and analyzing financial data in quantitative investment strategies.
91 shares
Kats is a toolkit for analyzing time series data, detecting changes, and predicting future trends.
4,813 shares
The article explores the use of optimization analytics in creating and testing the best portfolios.
17 shares
The article presents an open-source library for managing time series data in Python.
2,094 shares
Tigramite is a Python package designed for determining cause and effect in time series data.
1,219 shares
The article explores methods for automating the identification of chart patterns.
9 shares
The piece presents Odoo, an open-source application aimed at facilitating business growth.
35,371 shares
The article lists free software network services and web applications that can be hosted on personal servers.
182,020 shares
The article details a Python scanner capable of detecting and illustrating common chart patterns.
68 shares
The piece introduces a quick, keyboard-centric terminal user interface for interacting with large language models.
1,340 shares
Industry news: funds, hiring, markets and regulation.
20 items
TickLab is revolutionizing the finance sector with an AI-driven quant hedge fund and E.D.I.T.H, a tool for financial analysis.
6 shares
Bequant, a digital assets firm, has launched RiskQuant, a comprehensive risk management toolset that includes prime services, liquidity access, and administrative tools.
5 shares
Taula Capital, a new hedge fund firm, has ceased accepting new funds on its first trading day after launching with $5bn in assets under management.
4 shares
Glacial Capital and TRSE are suing Buenos Aires province over a $77m debt default from bonds issued in 2011.
4 shares
Liquidnet has introduced Roll Seeker, a tool designed to improve the efficiency of workflows related to the execution of futures rolls.
3 shares
Avelacomm has created two new low latency routes to improve trading links between Seoul, Hong Kong, and Singapore.
2 shares
US hedge fund managers have cut their domestic stock exposure following a drop in US stock market indexes, says Morgan Stanley's prime brokerage division.
2 shares
Murex has teamed up with StarQube to offer a data-driven portfolio management system via the integrated MX.3 platform.
2 shares
The Teacher Retirement System of Texas and other institutional investors have called on hedge fund managers to set cash hurdles before levying management fees.
2 shares
Digital asset investment products have seen their fourth straight week of inflows, totaling $15bn year-to-date, as per a CoinShares report.
2 shares
Hedge funds including Millennium Management GLG and Gladstone Capital Management have been surprised by a wave of takeover bids for UK companies they were short selling, leading to a rethink of such strategies.
2 shares
LA-based hedge fund firm, Canyon Partners, has closed two collateralised loan obligations worth over 1bn, including a €550m European CLO and a 500m US CLO.
2 shares
The European Energy Exchange has established a subsidiary in Japan to strengthen ties with local stakeholders, including traders and authorities.
1 shares
GoldenTree Asset Management has closed a 500m collateralised loan obligation, bringing the total issued under its GLM strategy to over 14bn through 26 CLOs.
1 shares
TikTok is now offering salaries that are higher than those typically offered by hedge funds.
1 shares
Senior engineers Mark Franklin and Luke Pebody have departed Balyasny to join Point72 and Jane Street respectively.
1 shares
Citadel's main fund, Wellington, has grown to over $14.5bn in assets, doubling its size in the last five years.
1 shares
JPMorgan is expanding its use of artificial intelligence significantly.
0 shares
A weak yuan and unstable economy in China are causing a surge in demand for foreign assets, leading to a record number of global fund launches.
0 shares
Junior bankers are finding that skills in automation are becoming increasingly important.
0 shares
Episodes on markets, quant methods and economics.
10 items
Adam Butler and Rodrigo Gordillo of ReSolve discuss market volatility and Carry strategies in their latest podcast.
19 shares
Meb discusses his startup investments and the concept of The Bear Market in Diversification in his recent podcast episode.
6 shares
A podcast by Meera Chandan, Patrick Locke, and James Nelligan focuses on US payrolls, CPI, FOMC, and G10 central bank meetings in June.
5 shares
Francis Diamond, Fabio Bassi, and Aditya Chordia discuss European rate markets, the May ECB meeting, and upcoming elections in their podcast.
5 shares
The North American Equities market continues to experience significant volatility.
4 shares
On World Ocean Day, Alex Burr, Ellie Piscopiello, and Shichen Zhao discuss the economic implications of water scarcity and the potential technological solutions.
3 shares
Tucker Scott of Bergen Park Capital shares his experience managing the Templeton Mutual Fund and his top picks for undervalued foreign stocks on a talk show.
3 shares
Natasha Kaneva predicts a 10% increase in Brent oil prices by September due to a shift in oil demand from petrochemicals to crude-rich products.
2 shares
Jay Barry and Afonso Borges discuss the medium-term outlook for Treasury supply and potential risks to the Treasury market ahead of upcoming first-tier data.
2 shares
Brendan Ahern discusses the current state of China's stock market and economy, including the impact of the property crisis, upcoming U.S. election, and regulatory changes.
2 shares
Posts from quant and economics blogs and newsletters.
6 items
The first article shares successful trading strategies for the EURNZD forex pair.
5 shares
The second article presents custom trading techniques for the AUDCAD forex currency pair.
5 shares
The third article explores the detrimental effects of confirmation bias on trading decisions.
2 shares
The article highlights the need for innovative ways to analyze changing price data in financial markets.
2 shares
The piece focuses on the ongoing quest for better indicators to comprehend dynamic price data in financial markets.
2 shares
The article underscores the importance of a traditional technical indicator in the fast-paced financial trading world.
1 shares
Talks, lectures and tutorials.
5 items
The BTRM webcast discusses the effects of Basel 3.1 on bank capital planning and management, and how to adapt efficiently.
0 shares
Peter Carr stresses the need for a quantitative understanding of finance.
0 shares
Alok from Jane Street's Research team shares his positive experience of the company's collaborative culture in their New York office.
34 shares
The article highlights a trend in companies preferring to hire specialists rather than generalists.
2 shares
The article underscores the importance of cherishing every moment with children, as they grow up quickly.
4 shares
Posts from quant researchers on X.
7 items
The article explores how sell-side data arbitrage influences stock dividend divergence investment strategies.
4 shares
Python Asset Pricing Library: PyAnomaly, a new Python library, aids asset pricing research by supporting firm characteristics generation, portfolio construction, and regression.
3 shares
The article evaluates short-term mean-reversion indicators for global equities, noting that the most successful indicators use current price measures in the trading range.
1 shares
The article shares Keaton Hamm's advanced linear algebra lecture notes for data science from the University of Arizona.
1 shares
The article explores AlphaSimplex's insights on the factors driving the growth of managed futures ETFs.
1 shares
The article examines the varying performance of highly shorted stocks during squeeze and nonsqueeze days.
0 shares
The article presents Nevergrad, a gradient-free optimization platform by Facebook AI Research, capable of single/multi-objective optimization.
0 shares