Quant LetterNo. 47
169 items across 11 sections, as sent to readers on 1 May 2024. Paper titles open their ML-Quant page; ↗ goes to the source.
Quantitative-finance and ML-for-finance preprints from arXiv.
13 items
A new method using wavelet coefficients to study stock price jumps reveals features of volatility and suggests that many simultaneous price jumps in different stocks are due to an internal contagion mechanism.
6 shares3 citations todaySource ↗
A proposed model for log price effectively captures features of implied SPX volatility, providing a solution to the consistent modeling problem and demonstrating competence in joint calibration of the SPX/VIX Markets.
5 sharesSource ↗
An approach to enhanced indexation with sector constraints outperforms the S&P 500 over five years, including during the Covid pandemic, by optimizing the proportion of the portfolio invested in each sector.
3 sharesSource ↗
A study finds that high-frequency traders (HFTs) can lower a large trader's costs if the market temporary impact is relatively large, and that HFTs' repeated behavior makes the large trader's optimal strategy close to uniform trading.
2 sharesSource ↗
The use of the Kalman-Filter analysis to assess market efficiency shows significant portfolio returns in both emerging and developed markets, suggesting its effectiveness across various market types.
2 shares1 citation todaySource ↗
The article suggests a blockchain-based system, owned and managed by the community, to enhance transparency and efficiency in academic peer reviews, aiming to minimize scientific misconduct.
4 shares4 citations todaySource ↗
The paper highlights the strategic advantages of bug bounty programs for software vendors, demonstrating that they can boost profits, expedite software launches, and improve security by encouraging ethical hackers to identify and report vulnerabilities.
2 shares6 citations todaySource ↗
The article presents Elliptic2, a dataset of Bitcoin subgraphs, to help understand and combat money laundering in cryptocurrency.
9 shares28 citations todaySource ↗
The paper suggests using blockchain in Credit Guarantee Schemes to improve their performance and efficiency.
5 shares1 citation todaySource ↗
The study investigates the impact of incidents on trust and trading behaviors in centralized and decentralized cryptocurrency exchanges.
3 shares10 citations todaySource ↗
The study investigates the risk of multi-block MEV attacks in dynamic weight market making, proposing new protections supported by numerous simulations.
7 shares2 citations todaySource ↗
A model is created to predict potential targets of activist investment funds, using Russell 3000 index data and identifying key factors through the Shapley value method.
4 shares1 citation todaySource ↗
A new framework is presented for modeling the EFP spread in spot precious metals market making, using a nested Ornstein-Uhlenbeck process to balance profit and inventory risk.
4 shares3 citations todaySource ↗
Working papers in finance and economics from SSRN.
40 items
Research on 501 Chinese firms listed on the Hong Kong Stock Exchange reveals that hedging, particularly with interest rate derivatives, significantly reduces a firm's default risk, with state-owned enterprises less likely to default.
4 sharesSource ↗
A new method using probability density evolution and a machine learning algorithm has been proposed to assess the seismic reliability of aqueduct structures, showing improved reliability against random seismic effects.
2 sharesSource ↗
MatFlow, a web-based machine learning platform for materials science, supports both forward and inverse design of materials, with a specific application in bandgap prediction.
2 sharesSource ↗
A study defines hedge funds' systemic risk based on a banking index's sensitivity to extreme losses, finding that larger funds, use of leverage, and uncertain market conditions indicate higher systemic risk levels.
5 sharesSource ↗
A study of Japanese CFOs reveals that those with higher risk aversion and less personal portfolio diversification tend to invest less in risky projects, engage more in hedging with derivatives, and choose lower financial leverage.
4 sharesSource ↗
During the COVID-19 pandemic, Korean retail investors have significantly increased their involvement in the ETF market, particularly in derivative ETFs.
2 sharesSource ↗
A study suggests a reliable method for designing a rational seismic system for small and medium-span girder bridges using a real bridge database and machine learning algorithms.
3 sharesSource ↗
A new numerical method using the mimetic finite difference method is introduced for pricing American and European options, addressing the degeneracy of the Black-Scholes differential operator.
2 sharesSource ↗
A study investigates the uncertainty of predictions from pruned neural network models in the context of conformal prediction, with a focus on filter-level pruning in convolutional neural networks.
2 sharesSource ↗
The efficiency of semi-supervised learning in machine learning can be enhanced by using methods that prevent overfitting and eliminate the need for incremental labeling, such as implementing diversity strategies.
2 sharesSource ↗
A new method using machine learning has been created to analyze ultra-high-performance fiber-reinforced concrete.
2 sharesSource ↗
Risks vs Rewards: The article highlights the risks of using AI in finance due to a lack of understanding of its technical aspects.
2 sharesSource ↗
AI is being used to predict the success of crowdfunding campaigns based on textual descriptions.
3 shares4 citations todaySource ↗
A theoretical framework has been proposed for creating optimal interest rate models on DeFi lending platforms.
2 sharesSource ↗
The study suggests venture capitalists may manipulate startup valuations for their own fundraising benefits, questioning the accuracy of derived performance measures.
2 sharesSource ↗
The article highlights the legal gaps in data mining and how it affects national security due to unclear big data property rights.
2 sharesSource ↗
The article contrasts mean-variance and behavioral portfolio theories, focusing on investors' risk and return management.
3 sharesSource ↗
The article introduces a deep-learning AI model that surpasses human analysts in predicting stock prices a year ahead, using large data sets efficiently.
3 shares1 citation todaySource ↗
The article investigates the impact of US financial uncertainty on global credit conditions and economic slowdown, influenced by market expectations.
2 sharesSource ↗
The article addresses the ethical, societal, and security concerns of generative AI systems in finance, advocating for thorough testing, strong security, and regulatory involvement.
2 sharesSource ↗
The Gated Recurrent Straightthrough Unit (GRSTU), a new deep learning model, outperforms statistical jump models in identifying regime changes in the S&P500 index, especially with smaller datasets.
143 shares6 citations todaySource ↗
Nonlinear machine learning models are more effective than linear models in predicting future fund flows, with past flows and the Morningstar rating as key predictors.
2 shares1 citation todaySource ↗
A deep generative model can create high-quality, correlation-aware synthetic financial time series, useful in a GAN-assisted risk-based portfolio construction method.
2 sharesSource ↗
A crowdfunding campaign is underway to fund a book on portfolio construction and risk management, featuring a guide to a next-generation investment framework and Python code.
13 shares1 citation todaySource ↗
Various methods, including a proposed ensemble method, are used to detect and exploit relative mispricing in the German stock market, with some strategies yielding average portfolio returns of about 60 bps per month.
4 sharesSource ↗
Bitcoin ETFs have attracted over $20 billion in their initial weeks of trading, significantly influencing bitcoin prices and exceeding daily bitcoin production by fivefold.
3 shares4 citations todaySource ↗
Enhanced ESG transparency can decrease investment efficiency by overshadowing fundamental stock price information and increasing pressure on ESG performance.
10 shares2 citations todaySource ↗
A novel method using decision trees offers a reliable estimator of the conditional correlation between financial assets, aiding in financial market analysis and understanding asset behavior in various market conditions.
5 sharesSource ↗
Institutional investors tend to rollover index options and futures contracts when facing higher unrealized losses as expiration dates near, particularly during political uncertainty, but this strategy is unprofitable.
3 sharesSource ↗
New metrics of downside and upside aggregate implied correlation from options reveal that downside correlation risk has a significant negative premium, suggesting investors require a hedge against correlation spikes.
3 sharesSource ↗
Volatility targeting strategies often outperform simple investment methods, especially in equity, due to their trend-following nature.
150 sharesSource ↗
Institutional investors gain investment advantages through onsite meetings with portfolio firms, leading to more informed local stock trades.
195 sharesSource ↗
The long-term performance of stock returns in the Korean market from 1980-2021 is largely influenced by capital investment returns, which declined after 2000.
2 sharesSource ↗
Large-Scale Asset Purchases influence security prices directly and indirectly, with the direct impact being less for securities held by price-elastic investors.
2 sharesSource ↗
Sustainable investing impacts firms' equity costs through institutional portfolio rebalancing, with firms' cost of equity increasing as climate change price pressure decreases.
2 sharesSource ↗
A novel method has been created to predict stock returns, solving the low signal-to-noise ratio problem and applicable to all US stocks by linking characteristics and stock returns.
2 sharesSource ↗
The article suggests three linked solutions to the absence of a comprehensive asset pricing theory, including a state-space representation, explanatory gain decomposition, and a shift in research from asset returns to asset prices.
2 sharesSource ↗
Research shows that climate and environmental policies significantly affect the relationship between a company's environmental impact and its credit risks and debt costs, with stricter policies increasing credit risk for polluting firms and reducing it for eco-friendly ones.
2 sharesSource ↗
The choice for a company to go public is influenced by the regional availability of equity financing, with firms rich in intangibles more likely to initiate IPOs when regional equity capital is plentiful.
2 sharesSource ↗
The article stresses the need for fair wealth distribution for the effective operation of a free-market economy, arguing that extreme inequality and a flawed banking system obstruct entrepreneurship and innovation, worsening wealth concentration and inequality.
3 sharesSource ↗
Economics working papers from RePEc's NEP field reports.
29 items
The article highlights the significant role of big data and AI in transforming the finance industry, stressing the importance of combining financial knowledge with data analytics.
26 sharesSource ↗
The study identifies consistent patterns in Bitcoin and Ether's volatility and volume, attributing these to algorithmic trading and futures market timings, and slower price changes on decentralized exchanges.
19 sharesSource ↗
The paper disputes traditional beliefs in capital asset pricing models, showing the absence of tangency due to kinks on efficient frontiers and the universal presence of kinks in portfolio choices.
16 sharesSource ↗
The study explores the structure of sectoral risk contagion in terms of volatility and jump, revealing strong inter-sector effects and the need for accurate risk contagion identification for effective regulation.
14 sharesSource ↗
The research suggests a memory-enhanced momentum strategy for commodity futures markets, proving its superiority over traditional strategies and its resistance to momentum crashes and market fluctuations.
12 sharesSource ↗
A study shows that XGBoost machine learning models are more effective than logistic regressions in predicting credit risk in Brazilian wholesale firms.
30 sharesSource ↗
Interpretable machine learning models and the financial stress index can accurately predict systemic financial risk, with stock and money markets being the most influential.
26 sharesSource ↗
A UK study reveals cryptocurrency investors often lack diversification, lean towards high-risk investments, and show a correlation with gambling behaviors.
12 sharesSource ↗
Correlation-based filter selection methods can enhance the efficiency and accuracy of detecting financial fraud in large datasets.
11 sharesSource ↗
Including risk-neutral volatility skewness and kurtosis in volatility forecasting models is more effective than extrapolation, which may hinder the performance of risk-neutral kurtosis.
11 sharesSource ↗
Machine learning analysis of Chinese data from 1993-2016 shows credit is a better output predictor than money, but its predictive power has decreased post-2007 due to financial development.
28 sharesSource ↗
The article categorizes machine learning applications into four types based on reuse strategies and offers guidance for their development and deployment.
23 sharesSource ↗
The article presents a package named ddml for double/debiased machine learning in Stata, supporting estimators of causal parameters for five econometric models.
20 sharesSource ↗
A study using machine learning to predict stock returns found that the LSTM classifier is superior to RF and SVM, and these methods outperform a random choice classifier, questioning the random walk and efficient market hypotheses.
17 sharesSource ↗
The research suggests two control function methods to improve machine learning accuracy when training and prediction samples differ, reducing prediction error and selection bias.
17 sharesSource ↗
A model using five machine learning algorithms is created to automate candidate selection for Social Safety Net programs, with a random forest-based algorithm proving most accurate.
16 sharesSource ↗
A multi-output regression model is proposed for better supply chain forecasting, using variables from different hierarchical levels to generate reliable predictions.
13 sharesSource ↗
The study presents statistical models and machine learning methods for early software effort estimation, with Random Forest providing the best estimates for development time.
13 sharesSource ↗
The article suggests using machine learning to create big data-driven macroeconomic fan charts, aiding in complex decision-making for public policy.
16 sharesSource ↗
The article explores the application of machine learning in the COLIEE competition, using data augmentation and advanced language models to enhance legal information extraction and prediction.
13 sharesSource ↗
The article investigates the use of machine learning for profitable short-term stock market investments, especially in day-trading.
23 sharesSource ↗
The piece examines the application of factorial and cyclical models in macroeconomic and sectoral modelling, focusing on changing economic conditions and investment demand.
9 sharesSource ↗
The study reveals the influence of two types of investor sentiments - general market-wide and individual stock-specific - on asset valuation models.
8 sharesSource ↗
The paper suggests a new investment strategy for endowments and foundations, involving long only positions in two optimized long/short funds, to surpass the usual stock/bond split.
7 sharesSource ↗
Evidence from Markov-Switching Model: The study explores the relationship between return-volume in French SMEs and large firms, finding varying causal relationships based on market conditions and company size.
7 sharesSource ↗
The research investigates the factors affecting gold prices in Malaysia, including GDP, inflation rate, interest rate, unemployment rate, and exchange rate.
7 sharesSource ↗
Financial Conditions Index for South American Economies: The study introduces an International Financial Conditions Index for South American economies to evaluate their financial status, especially after the 2022 Russian invasion of Ukraine.
7 sharesSource ↗
The research notes a shift in international consumer price index inflation comovement since 2008, with global factors playing a larger role in national inflation rates.
6 sharesSource ↗
The study uses a dynamic factor model to examine the role of the national factor in US house price fluctuations, hinting at a possible national bubble since 2014.
5 sharesSource ↗
The general machine-learning papers the letter carried in 2023-25.
18 items
The article discusses the current research on large language models (LLMs) in continual learning (CL), focusing on the challenges of integrating these models into dynamic data distributions and task structures.
31 shares351 citations todaySource ↗
The study examines how large language models (LLMs) respond to prompts designed to generate coherent responses from nonsensical inputs, finding that the success of this manipulation depends on the length and complexity of the target text.
16 shares14 citations todaySource ↗
The research introduces an algorithm for learning mixtures of Gaussians in multiple dimensions using diffusion models, marking a significant advancement in unsupervised learning tasks.
14 shares44 citations todaySource ↗
The paper presents a study on learning from human demonstrations using a bimanual system with multifingered hands, introducing a low-cost teleoperation system and a novel hardware adaptation using two prosthetic hands with touch sensors.
10 shares159 citations todaySource ↗
The article proposes a method for predicting resource needs in Virtual Network Functions while preserving data privacy, using Federated Transfer Component Analysis.
10 shares1 citation todaySource ↗
A new framework is introduced that models Reinforcement Learning from Human Feedback as a Markov decision process, using an algorithm that learns from preference data.
9 shares143 citations todaySource ↗
The paper introduces Kangaroo, a self-speculative decoding framework that uses a small model to increase token acceptance rate and minimize drafting steps.
9 shares61 citations todaySource ↗
3D Painting: The article presents Make-it-Real, a new approach that uses Large Language Models to enhance the visual authenticity of 3D objects and streamline the 3D content creation process.
8 shares20 citations todaySource ↗
The paper introduces a framework that improves the synthesis of novel views for unseen scenes, using strategies like Adaptive Cost Aggregation, Spatial-View Aggregator, and Consistency-Aware Fusion.
5 shares9 citations todaySource ↗
Poisoning Attacks on Text-to-Image Models: The article introduces Nightshade, an attack on text-to-image models that can disrupt their functionality, potentially serving as a defense against web scrapers.
3,299 shares141 citations todaySource ↗
The research presents a framework for optimizing programs using large language models, achieving a mean speedup of 6.86, outperforming average individual programmers.
671 shares164 citations todaySource ↗
Understanding Human Activity: The paper presents Ego-Exo4D, a large-scale video dataset and benchmark challenge featuring human activities from various perspectives, aimed at improving first-person video understanding.
179 shares603 citations todaySource ↗
Text-Driven 3D Gaussian Splatting Editing: The article proposes GaussCtrl, a text-driven method for editing 3D scenes, offering quicker editing and improved visual quality through multi-view consistent editing.
75 shares146 citations todaySource ↗
Time Series: The article presents Motion Code, a new framework for time series classification and forecasting on noisy data, which improves the accuracy of time-dependent signal detection.
21 shares2 citations todaySource ↗
The paper offers a guarantee on the stability of bagging for any machine learning model, regardless of data distribution or algorithm properties, supported by empirical results.
21 shares31 citations todaySource ↗
The study explores latency attacks on deep learning applications, specifically object detection, and introduces a framework, Overload, that can generate such attacks, posing a threat to systems with limited resources.
20 shares31 citations todaySource ↗
The authors introduce Confidence-Triggered Detection (CTD), a method that improves tracking speed and accuracy in real-time object tracking by skipping object detection in certain frames.
19 shares29 citations todaySource ↗
The research examines the geometry of binary hyperdimensional computing (HDC), presents a new encoding method, Laplace-HDC, and discusses potential solutions to the limitations of binary HDC in encoding image spatial information.
16 shares8 citations todaySource ↗
Papers that shipped their code, from the Papers with Code feed (2023-25).
13 items
The article suggests a new method for assessing the ability to generate personalities.
19,582 shares
The article presents OpenELM, a new advanced open language model.
1,747 shares
The article highlights the impressive performance of InternVL 1.5, which scored highest in almost half of the benchmarks tested.
988 shares
The paper explores the application of Language Models in assisting the proof of mathematical theorems.
777 shares
The study introduces a gating strategy with pretrained LoRA adapters, enabling the XLoRA model to generate unique deep layer combinations for problem-solving.
506 shares
The article delves into the creation of node chains to facilitate a structured thought process.
199 shares
AutoCrawler is a framework that utilizes the hierarchical structure of HTML for a more comprehensive understanding in two stages.
147 shares
ConsistentID is a system that enhances precision by merging facial features and descriptions, and uses an ID-preservation network for consistency in facial areas.
128 shares
The authors encourage further research into the capabilities of versatile multimodal foundation models for practical applications.
92 shares
The article discusses the issue of large language models failing to effectively use information in long contexts, a problem termed as the 'lostinthemiddle' challenge.
79 shares
The piece highlights how SnapKV enhances decoding speed, generation speed, and memory efficiency when handling inputs of 16K tokens.
59 shares
The article outlines the difficulties in finding a cost-effective teleoperation system for a dual-arm setup with multi-fingered hands, and the scarcity of multi-fingered hand hardware with touch sensing.
37 shares
The article introduces a new differentiable solver for wide-baseline two-view pose, which is used to train the backbone end-to-end.
31 shares
Repositories the letter featured.
10 items
The article investigates the application of deep learning in algorithmic trading via different experiments.
1,405 shares
The piece talks about a Python framework created for testing trading strategies and studying financial markets.
694 shares
The article highlights a high-frequency trading system for futures, using machine learning and deep learning methods.
225 shares
This article showcases a set of finance-related instances where machine learning is applied.
92 shares
The article explores a large-scale time-series machine learning system using Polars for simultaneous feature extraction and predictions on panel data.
899 shares
The article explores a Python library that enables the repetition of specific operations.
5,974 shares
The piece highlights distfit, a Python library created for adjusting probability density.
337 shares
The article introduces a Python-oriented framework for creating and testing Ethereum Virtual Machine's smart contracts.
2,596 shares
The article outlines techniques for analyzing central bank announcements.
59 shares
The article reveals the codebase for FOMCNLP, which was approved for the ACL 2023 main event.
35 shares
Industry news: funds, hiring, markets and regulation.
20 items
Walleye Capital's CIO Andrew Carney and CTO Mark Zeldis are retiring, as announced by Business Insider.
7 shares
CNBC reports that ultra high net worth individuals have drastically reduced their hedge fund investments from 12% to 2% over 16 years.
5 shares
Private equity M&A partners have transitioned from Travers to a US law firm located in London.
4 shares
Clear Street, a financial services firm in New York, has launched futures clearing and execution services on US exchanges for global clients.
4 shares
The International Business Times Singapore Edition discusses the influence of AI and Quant Trading on the financial sector in an article titled Beyond Algorithms.
4 shares
Hong Kong hedge fund firm ActusRay Partners has increased its assets to over $700m in 2023 despite challenging conditions.
3 shares
Elliott Investment Management has bought a significant stake in Sumitomo, a favourite Japanese trading house of Warren Buffett.
3 shares
Financial services compliance software provider COMPLY has named Michael J Stanton as its new CEO.
3 shares
Solo Capital Partners' founder, Sanjay Shah, insists the cumex trades accused of defrauding Danish tax authorities of £1.4bn were legal.
3 shares
According to a study by Ocorian, 92% of alternative fund managers use AI for risk and compliance procedures.
3 shares
Banks expect a rise in private equity transactions.
2 shares
Hedge funds have significantly increased their tech stock purchases, the highest in over a year.
2 shares
Digital asset investment products saw their biggest outflows since March, as per CoinShares' report.
2 shares
US financial assets saw a strong recovery in 2023, with a notable decrease in inflation and mortgage rates from 2022.
2 shares
EEX Group's total revenues grew by 19% in 2023, due to increased trading volume across all assets and regions.
2 shares
Profits are increasing for trendfollowers due to the surge in cocoa prices and the decline in yen value.
2 shares
A compilation of the week's top stories on Environmental, Social, and Governance (ESG) is presented in one article.
1 shares
Financial News (FN) has published a distinguished list of noteworthy individuals or companies.
1 shares
The first quarter of the year saw a significant drop in the launch of new multistrategy hedge funds, accounting for less than 10% of all new funds.
1 shares
Financial News (FN) delves into the lighter, more entertaining side of the City's financial district.
1 shares
Episodes on markets, quant methods and economics.
10 items
Economy Risks: The ReSolve team and Jared Dillian discuss financial stress, diversification strategies, and future global financial market trends in a recent episode.
11 shares
Catherine LeGraw of GMO’s Asset Allocation Team talks about the best asset allocation opportunity in 35 years and the potential for high returns.
10 shares
Selloff Alert: Jonny Goulden and Saad Siddiqui discuss the impact of recent market developments on the EM fixed income asset class in a 25 April 2024 podcast.
8 shares
Dollar Rising: FX strategists discuss the potential impact of the dovish BoJ meeting on JPY, potential FX intervention risk, and upcoming US event risks for the dollar.
6 shares
LGIM's 13th annual Active Ownership report outlines their 2023 work on nature, biodiversity, and artificial intelligence, and their focus for 2024.
5 shares
Dr. Ed Yardeni discusses global investment strategies and asset allocation analyses with Bloomberg Radio host Barry Ritholtz.
4 shares
Dr. Michael De Groot's book suggests the global economic turmoil of the 1970s was instrumental in ending the Cold War due to economic interests surpassing national security.
3 shares
Gregory Shearer predicts a positive future for copper prices but warns of possible short-term volatility in a podcast recorded at the Cesco Week event in Chile.
3 shares
Bilal, from Macro Hive, discusses potential Federal Reserve cuts in 2024, overlooked aspects of China's economy, and Europe's strategic issues in a recent podcast.
2 shares
Alex Gurevich, founder and CIO of Honte investments, discusses persistent inflation, currencies, precious metals, and energy markets with MacroVoices hosts Erik Townsend and Patrick Ceresna.
2 shares
Posts from quant and economics blogs and newsletters.
8 items
The article tests forex trading of US dollar news announcements using a Monte Carlo simulation instead of a standard back test.
7 shares
The article uses a Monte Carlo simulation to evaluate the effectiveness of trading forex news announcements about the US dollar.
7 shares
The article uses a Monte Carlo simulation to test forex trading of US dollar news announcements, aiming to profit from large market shifts.
7 shares
The article discusses the difficulty of profiting from day trading strategies due to the vast number of available strategies.
6 shares
The article highlights the importance of robust trading strategies for successful trading in the EURAUD market.
4 shares
The Barnard chapter of the American Association for University Professors and faculty have expressed a vote of no confidence in President Laura Rosenbury.
0 shares
Columbia University students are staging a protest on the West Lawn, calling for divestment as per their proposal.
0 shares
The article underscores the risks and financial implications of engaging in high-speed activities.
0 shares
Talks, lectures and tutorials.
2 items
The article highlights the nuanced differences between quantitative research and model development, noting that these tasks are typically carried out by quants, not to be mistaken for quant devs.
40 shares
The article delves into the significant workload associated with data management and transfer.
0 shares
Posts from quant researchers on X.
6 items
The article gives a detailed analysis of the operations and structure of Jane Street.
0 shares
The paper delves into CRISPRGPT, a tool for automating gene-editing experiments, and discusses possible hurdles.
0 shares
Robin Wigg's article provides a thorough review of Jane Street's business and operations.
0 shares
Article: The article uncovers the solution to the ChatGPT problem that was identified in Nigeria.
0 shares
Article: The article provides ManGroup's guidance on the correct allocation to cryptocurrencies.
0 shares