Market Trough Prediction
The research uses machine learning to identify that the volatility of options-implied risk and market liquidity are key factors causing market lows, challenging simpler models.
19 sharesSource ↗
Quant LetterNo. 112
September 2025, Week 2
75 items across 4 sections, as sent to readers on 13 September 2025. Paper titles open their ML-Quant page; ↗ goes to the source.
- Items
- 75
- Papers
- 67
- Sections
- 4
- Issue
- No. 112
arXiv
Quantitative-finance and ML-for-finance preprints from arXiv.
27 items
Finance10
Optimal Exit in AMMs
The study investigates the best way for a liquidity provider to withdraw liquidity in an automated market, balancing fees and potential losses, and offers insights into dynamic liquidity provision.
14 shares3 citations todaySource ↗
Chaotic Bayesian Models
The paper presents a risk model that combines heavy-tailed priors with chaotic dynamics to predict volatility clustering, fat tails, and extreme events, providing a dual perspective for systemic risk analysis.
12 sharesSource ↗
Deep Learning for Options
The study introduces a deep learning framework for pricing options based on market-implied volatility surfaces, providing an efficient and scalable method that improves with more data.
9 shares3 citations todaySource ↗
Hedging Asset Portfolio Options
The research examines the viability of hedging options with a cheaper, related asset when the main asset is expensive to trade, concluding that trading the wrong asset can be a viable option under certain conditions.
9 sharesSource ↗
Environmental Performance & Tax Avoidance
Research shows that firms with better environmental performance, especially those under financial stress, are more likely to evade taxes, based on a study of 567 FTSE All Share listed companies from 2014 to 2022.
7 shares10 citations todaySource ↗
Joint Volatility Surface Calibration
The article suggests a calibration framework for complex option pricing models that simultaneously fits market option prices and variance term structure, enhancing the precision of model-implied variance term structures.
6 sharesSource ↗
FinZero: Multi-modal Forecasting
Multi-modal Forecasting: FinZero, a pre-trained model fine-tuned by the Uncertainty-adjusted Group Relative Policy Optimization method, is introduced in the article, enhancing the accuracy, adaptability, and scalability of financial time series forecasting.
6 shares5 citations todaySource ↗
Financial & Sustainability Ratios in Sector Mapping
The article visualizes the performance of Spanish fish and meat processing companies using compositional data methodology and principal-component analysis biplot, focusing on long-term solvency, energy, waste and water intensity, and gender employment gap.
5 shares2 citations todaySource ↗
Optimal Investment in Stochastic Factor Model
The article discusses optimal investment and consumption in an incomplete stochastic factor model, offering a comprehensive characterization of the problem's well-posedness and an efficient numerical algorithm for computing the value function.
5 shares3 citations todaySource ↗
Economics6
Environmental Law Enforcement in Amazon
The Real-Time Deforestation Detection System (DETER) in the Brazilian Amazon has reduced homicides by 15%, showing the social benefits of preventing deforestation.
53 shares2 citations todaySource ↗
Comparing Wellbeing Measures
The validity of international wellbeing rankings like the World Happiness Report is questioned due to varying survey responses and rankings across countries, indicating a need for more research on cross-country wellbeing comparability.
9 shares2 citations todaySource ↗
AIs Impact on Expert Markets
Large language models (LLMs) in expert services have pros and cons, with human markets being more efficient, but LLMs potentially reducing trust and overshadowing experts' preferences, while also improving experts' communication of their goals.
8 shares3 citations todaySource ↗
Life Satisfaction Preferences Inequality
A UK study suggests societal well-being policies should focus on fairness rather than personal risk, as most people prioritize collective values over individual life satisfaction.
7 shares1 citation todaySource ↗
Food Security Longitudinal Data
A study analyzing 40 years of US data on household-level food security offers insights into how recessions and policy changes impact different demographics.
6 shares1 citation todaySource ↗
Basic Chemicals Production Decarbonization
A study on decarbonizing the chemicals industry indicates that significant investment and a favorable investment environment are needed, with full decarbonization possibly extending into the latter half of the century.
4 sharesSource ↗
Miscellaneous6
Optimal Transport Distances for Financial Time Series
The article introduces a new, faster, and more efficient algorithm for generative AI in financial decision-making applications, improving the computation of the nested optimal transport distance.
12 sharesSource ↗
Fair Machine Learning with Protected Attributes
The paper suggests a new framework for promoting fairness in machine learning, especially in regression tasks and situations with multiple protected attributes, by reducing the link between model predictions and protected attributes.
9 shares2 citations todaySource ↗
Interpretable Deep Learning for Insurance Pricing
The paper presents the Actuarial Neural Additive Model, a transparent deep learning model for insurance pricing that provides superior prediction accuracy and full transparency in its internal workings.
6 shares1 citation todaySource ↗
Market-Driven Equilibria for Solar Panels
The study finds that a differentiated market encourages optimal investment in distributed solar panels, while a single-product market results in under-investment.
5 sharesSource ↗
Linear Pricing for Load Management in Day-Ahead Markets
The paper presents a new linear pricing mechanism for retail electricity that can control congestion and direct customer consumption without requiring bi-directional communication or customer bidding programs.
4 shares1 citation todaySource ↗
Linear Pricing for Load Management in Energy Markets
The article suggests a new linear pricing mechanism for retail electricity that can manage congestion and guide energy consumption, removing the need for bi-directional communication or customer bidding programs.
4 shares1 citation todaySource ↗
Historical Trending5
AI Impact on Occupations
AI is primarily used in work activities for information gathering and writing, particularly in knowledge-based roles like computer and administrative support.
3,971 shares27 citations todaySource ↗
Generative Models for Financial Forecasting
A new method for simulating financial market data transforms limit order book data into an image format and uses diffusion models to predict future states, showing top performance on LOB-Bench.
20 shares2 citations todaySource ↗
Causal Perspective on CAPM
A novel approach to simulating limit order books in financial markets converts the data into an image format and applies diffusion models to predict future states, surpassing previous methods in certain metrics.
20 shares2 citations todaySource ↗
Optimal Paths in Dynamic Optimization
A study suggests the Capital Asset Pricing Model should be viewed as associational, not causal, with beta reflecting market capture of underlying drivers, and risk management should focus on declared causal paths instead of fixed factors.
14 sharesSource ↗
Characterizing Optimality in Dynamic Settings
A new method for studying optimal paths in dynamic optimization problems uses a locator function to identify and assess the stability of steady states, allowing for comparative statics without solving the entire dynamic program.
13 sharesSource ↗
RePEc
Economics working papers from RePEc's NEP field reports.
30 items
Finance6
Enhanced EM Portfolios with AATS
The rise of algorithmic trading and passive investing has caused greater losses during market downturns, but an Automated Adaptive Trading System could help stabilize emerging markets in such times.
27 sharesSource ↗
Efficient Volatile KSE-30 Equities Identification
Machine learning has been utilized to pinpoint assets causing downward trends in the Pakistan Stock Exchange, suggesting a portfolio optimization plan for effective asset allocation.
25 sharesSource ↗
Dynamic Correlations in Risk Parity Portfolio Optimization
Using expected shortfall as the risk measure in risk parity portfolio optimization reduces sensitivity to volatility shocks, decreases portfolio turnover during market unrest, and enhances risk-adjusted returns considering fat-tailed returns.
16 sharesSource ↗
Adaptive Market Hypothesis & Sharpe Ratio Strategies
The research finds that trading strategies based on the Sharpe Ratio are more profitable than the buy-and-hold strategy in global markets, supporting the Adaptive Market Hypothesis.
15 sharesSource ↗
Novel Window Analysis for HFT
The study introduces a new method for assessing decision-making efficiency over time, using the Whale Optimization Algorithm, and applies it to foreign exchange investment strategies and utility companies.
11 sharesSource ↗
Monitoring Poverty in Data-Deprived Lebanon
The paper uses a new data augmentation technique to study poverty in the Middle East and North Africa, specifically Lebanon, using alternative data sources when traditional income data is scarce or unavailable.
10 sharesSource ↗
Statistical5
BRM for Incomplete Data Prediction
The BRM method is introduced for analyzing missing data patterns, using ensemble models to reduce data imputation, and showing better predictive performance for various models.
20 sharesSource ↗
New Momentum Strategy for Equity Premium
A new N-MDIS strategy using machine learning is proposed for predicting equity premium, proving more accurate than previous strategies with logistic regression.
19 sharesSource ↗
Product Market Competition and Zero-Leverage
The study reveals that firms are more likely to adopt zero-leverage policies as product market competition increases, especially in firms with high earnings volatility.
18 sharesSource ↗
News Sentiment Impact on Risk Management
The impact of news sentiment on stock return volatility is reevaluated, showing that both positive and negative firm-specific and macroeconomic news significantly affect intraday stock return volatility.
16 sharesSource ↗
Stochastic ML for Production Technologies
A version of Stochastic Gradient Boosting is proposed to estimate production possibility sets, reducing overfitting in Data Envelopment Analysis, and showing competitive performance compared to other methods.
16 sharesSource ↗
Machine Learning7
Machine Learning for M&A
A study found that machine learning models are more effective than traditional methods in predicting Chinese corporate merger and acquisition activities.
28 sharesSource ↗
Tail Risk Management
New probabilistic deep learning frameworks have been proposed for estimating financial risk measures, improving capital allocation for financial institutions.
27 sharesSource ↗
Bond Market Volatility in China
Machine learning methods can accurately predict Chinese stock market volatility using the volatility of long-term treasury bond contracts.
24 sharesSource ↗
Lot Streaming and Scheduling
The article proposes a new algorithm and machine learning model to improve efficiency and accuracy in the Lot Streaming and Scheduling Problem with stochastic product arrival times.
16 sharesSource ↗
Dynamics in Chinese Financial Markets
The paper introduces a new machine learning technique for analyzing and modeling complex time series, providing a potential alternative to the Box-Jenkins method in financial modeling.
13 sharesSource ↗
Monetary Policy Frictions and Nonperforming Loans
The article uses machine learning to create a monetary policy frictions index from financial news, revealing a significant positive impact on Chinese commercial banks' nonperforming loans.
12 sharesSource ↗
Housing Market Connectedness
The paper uses machine learning to study the global housing market, finding that the US market is the main source of systematic shocks and its interest rate is the most influential factor.
10 sharesSource ↗
Deep Learning2
Oil Price Forecasting: Machine Learning vs. Deep Learning
Machine Learning vs. Deep Learning: Deep learning methods have been found to be more effective than traditional machine learning in predicting oil prices, particularly during crises.
31 sharesSource ↗
Multifrequency Data Fusion Model for Carbon Price Prediction
The newly introduced MFF-CPPM model in China has shown higher accuracy and flexibility in predicting carbon trading prices compared to standard models.
10 sharesSource ↗
Historical Trending10
Predicting VIX Trends
The article discusses a study that uses machine learning to predict the CBOE Volatility Index, with weekly jobless claim data as a significant factor.
23 sharesSource ↗
Stock Price Prediction in Eurozone
The study reveals that traditional machine learning models outperform deep learning models in predicting Eurozone banking sector stock prices.
13 sharesSource ↗
AI Capability and Performance
The research indicates that AI capability directly affects firm performance, with a data-driven culture and AI infrastructure playing crucial roles.
5 sharesSource ↗
Climate Discussions on Social Media
The article emphasizes the need for communication and a comprehensive approach to address climate change, using machine learning to analyze social media discussions on the subject.
4 sharesSource ↗
Dark Patterns in Retail Investors
The study investigates the use of dark patterns in retail investment, and how behavioral sciences and AI can improve regulation and protect investors.
2 sharesSource ↗
Young Informal Workers in the EU: Labor Market Dynamics
Labor Market Dynamics: The research profiles young informal workers in the EU27, aiming to understand how Covid-19 has impacted youth employment informality.
2 sharesSource ↗
Resource Management in Cloud Computing: AI for DevOps
AI for DevOps: The article discusses how artificial intelligence can improve resource management in cloud environments, boosting DevOps workflows' performance and efficiency.
2 sharesSource ↗
EGovernance and Citizen Participation: Literature Review
Literature Review: The review investigates the link between e-governance initiatives and citizen participation, emphasizing the need for interdisciplinary research for effective evaluation.
2 sharesSource ↗
Bank Performance Determinants: Future Research
Future Research: The paper analyzes factors influencing banks' performance, proposing new research areas related to digital transformation, artificial intelligence, and COVID-19's impact.
1 sharesSource ↗
Online Gig Work: Structural Analysis
Structural Analysis: The study tests the Work Need Satisfaction Scale's (WNSS) suitability for online gig workers, suggesting modifications to the scale to better reflect online platform work.
1 sharesSource ↗
Machine learning
The general machine-learning papers the letter carried in 2023-25.
10 items
Historical Trending10
QLASS: Language Agent
Language Agent: QLASS is a new method for training language agents, using a step-by-step approach to generate annotations and enhance performance with minimal supervision.
189 shares19 citations todaySource ↗
LLM Benchmarks
Platinum benchmarks are proposed to test the reliability of large language models, showing that even advanced models can struggle with basic tasks.
57 shares54 citations todaySource ↗
ToddlerBot: Humanoid Platform
Humanoid Platform: ToddlerBot is a low-cost, open-source humanoid robot designed for scalable policy learning and AI research, focusing on data collection and policy execution.
49 shares21 citations todaySource ↗
Masked Autoencoders
MAETok is an autoencoder designed to enhance high-resolution image synthesis by learning a semantically rich latent space, leading to quicker training and increased inference throughput.
38 shares94 citations todaySource ↗
World Dynamics
NutWorld is a new framework that converts monocular videos into dynamic 3D Gaussian representations in one go, allowing for high-quality video reconstruction and real-time applications.
30 shares8 citations todaySource ↗
RAG Systems Evaluation with UAEval4RAG
UAEval4RAG is a new framework that assesses the capability of retrieval-augmented generation systems to manage unanswerable queries, emphasizing the significance of component selection and prompt design.
29 shares10 citations todaySource ↗
Stochastic Gradient Descent for Optimization
A novel gradient descent algorithm with stochastic elements is introduced for solving nonconvex optimization problems, demonstrating its global convergence and efficiency through numerical examples.
28 shares6 citations todaySource ↗
Training Dynamics in Linear Networks
The study offers a theoretical analysis of gradient descent dynamics in deep linear networks, explaining the wider is better effect and hyperparameter transfer effects in large-scale random data training.
26 shares17 citations todaySource ↗
DeepSeek R1 and Generative AI
DeepSeek's cost-effective reasoning model, DeepSeekR1, is compared to OpenAI's models, with a focus on its innovative use of Mixture of Experts, Reinforcement Learning, and engineering in Generative AI.
26 shares43 citations todaySource ↗
Rankify: Python Toolkit for Retrieval
Python Toolkit for Retrieval: Rankify, an integrated open-source toolkit, is launched to improve retrieval and re-ranking methodologies, providing consistency, scalability, and ease of use within a retrieval-augmented generation framework.
24 shares11 citations todaySource ↗
GitHub
Repositories the letter featured.
8 items
Finance3
Simple Market Making
The article guides on executing basic market making strategies on the Kalshi platform.
113 shares
Claudable Web Builder
The article presents Claudable, an open-source web builder that utilizes local CLI agents for product building and deployment.
2,104 shares
ZeroShot TextToSpeech System
The article explores a high-level, manageable, and efficient zero-shot text-to-speech system.
6,568 shares
Trending5
SM Blocks
The article Shadcn marketing blocks explains the application of Shadcn's marketing techniques and tools.
1,717 shares
AS AI
AgentScope AgentOriented Programming for Building LLM Applications discusses the use of AgentScope in creating LLM applications through agent-oriented programming.
10,696 shares
Codebuff
Generate code from the terminal offers a tutorial on generating programming code directly from the terminal.
351 shares
AOTF Memento
Official Code of Memento Finetuning LLM Agents without Finetuning LLMs reveals the official code for optimizing LLM agents without modifying the LLMs.
1,250 shares
PlutoPrint
A Python Library for Generating PDFs and Images from HTML powered by PlutoBook presents a Python library that enables the conversion of HTML into PDFs and images, facilitated by PlutoBook.
874 shares