Reinforcement Learning for Arbitrage
The study presents a new framework for statistical arbitrage that uses reinforcement learning to optimize asset coefficients and identify the best mean reversion strategies.
6 shares2 citations todaySource ↗
Quant LetterNo. 41
March 2024, Week 3
96 items across 8 sections, as sent to readers on 20 March 2024. Paper titles open their ML-Quant page; ↗ goes to the source.
- Items
- 96
- Papers
- 57
- Sections
- 8
- Issue
- No. 41
arXiv
Quantitative-finance and ML-for-finance preprints from arXiv.
14 items
Finance6
Deep Learning for Order Book Forecasting
The research applies deep learning techniques to predict mid-price changes for NASDAQ-traded stocks, providing a framework to evaluate the feasibility of these predictions.
4 shares15 citations todaySource ↗
Rough Volatility and Premium
The paper examines the effect of unpredictable risk on pricing in a rough volatility model, emphasizing the random nature of the market price of volatility risk.
3 shares1 citation todaySource ↗
Robust Utility Maximization
The study investigates maximizing utility in a frictionless market, introducing projectively measurable functions and proving the existence of an optimal investment strategy under certain conditions.
3 shares4 citations todaySource ↗
Optimal Portfolio Choice
The research looks at optimal portfolio choices in continuous time, considering the impact of transactions on prices and providing solutions to optimal portfolio and execution problems.
3 shares9 citations todaySource ↗
MeanField Game Market Entry
The paper explores optimal portfolio liquidation in games with unchangeable trading direction, proving the existence of a unique equilibrium in both mean-field and N-player games.
2 shares5 citations todaySource ↗
Economics2
Composite Likelihood for Gaussian Processes
The first article discusses a framework for understanding parametric continuous-time stationary Gaussian processes, applied successfully to models describing the random log-spot variance of financial asset returns, including cryptocurrency.
3 shares1 citation todaySource ↗
Signature Kernel Solver for Path-Dependent PDEs
The article introduces a new, verifiably effective kernel-based solver for path-dependent partial differential equations (PPDEs). This provides a practical alternative to Monte Carlo methods, especially for option pricing under rough volatility.
2 shares18 citations todaySource ↗
Crypto & Blockchain2
Physics & Finance
The study uses Kelvin waves and advanced math to link physics and financial engineering, aiming to solve complex problems like hedging losses in cryptocurrency trading.
6 shares3 citations todaySource ↗
Scaling Uniswap
The paper finds that cheaper, faster chains improve Uniswap v3 Protocol's performance and profitability, and suggests that issues with Automated Market Makers may stem from chain dynamics.
4 shares5 citations todaySource ↗
Historical Trending4
MeanField Microcanonical GD
The study presents a new model that improves the efficiency of sampling multiple data points simultaneously, particularly in high-dimensional financial time series.
9 sharesSource ↗
Capital Return Path-dependency
The research challenges the belief that profit rate and return rate on equity depend on divestments, capitalization path, and market interest rate, respectively, in periodic growth processes.
4 shares3 citations todaySource ↗
Large Order Trading with HFT
The study finds that faster or more accurate high-frequency traders may harm themselves but benefit the normal-speed informed trader in a market.
4 sharesSource ↗
Pairs Trading Graphical Matching Approach
The paper proposes a new method for pairs trading that selects pairs with strong cointegration but no shared assets, leading to lower portfolio variance, reduced trading costs, and better risk-adjusted performance.
4 sharesSource ↗
SSRN
Working papers in finance and economics from SSRN.
26 items
Quantitative8
Cryptocurrency Hedging Strategy
QuantPedia's article suggests a strategy to hedge cryptocurrency portfolios in cold storage using a Top 5 cryptocurrency index and BTC derivatives to reduce market risk.
3 sharesSource ↗
Weather Derivative Pricing Rainfall Model
The paper discusses the modeling and pricing of rainfall-based weather derivatives using the Markov Chain Analogue Year Mixed Exponential model and the Esscher transform.
4 sharesSource ↗
Seasonality in Cryptocurrencies
The study explores seasonality patterns in ten cryptocurrencies, noting lower trading volume and volatility during weekends and no consistent calendar effects on returns.
4 sharesSource ↗
Robust Linear Regression for Portfolio Optimization
The research presents a new optimization framework that minimizes unknown parameters and addresses estimation error in portfolio optimizations by focusing on the row sums of precision matrix estimates.
3 sharesSource ↗
Volatility Forecasting Deep Estimation
The article suggests using deep neural networks to estimate volatility models, aiming to improve volatility forecasting.
2 sharesSource ↗
VIX Forecasting Illusion
The paper uses daily volatility measures to forecast stock market volatility, finding inconsistent results with different evaluation metrics.
2 sharesSource ↗
Superkurtosis in Trading
Traditional risk measures may underestimate losses in intraday trading, posing a risk to financial market stability.
2 sharesSource ↗
Mergers Economic Impact in Africa
Mergers and acquisitions can help companies survive economic crises, affecting market share, shareholder wealth, employment, and economies of scale.
2 sharesSource ↗
Financial18
Arbitrage Efficiency
The study suggests that portfolios can use asset mispricing to increase efficiency, particularly during high-sentiment periods.
8 sharesSource ↗
Timing Volatility Premiums
The research shows that managing a portfolio based on volatility risk premium timing strategies can improve long-term performance, especially during periods of high volatility.
3 sharesSource ↗
Equity Options Trading
The study uses a multi-asset model to show that components of informed trading can predict high-volatility events in equity options.
3 sharesSource ↗
Bond vs Equity ETFs Liquidity
The research reveals that bond ETFs and equity ETFs have different trading dynamics, with bond ETFs offering more hidden liquidity and dark trading volume due to the lack of transparency in bond markets.
2 sharesSource ↗
ChatGPT Trading Strategy
The paper shows that ChatGPT can use Twitter news to generate profitable stock tickers for day trading, demonstrating the AI's ability to turn non-specific news into firm-specific mispricing signals.
4 sharesSource ↗
Merger Arbitrage with Machine Learning
Machine learning can enhance the profitability of merger arbitrage trades, offering valuable financial insights and substantial economic benefits for investors.
2 shares1 citation todaySource ↗
Improved Volatility Forecasting
Enhancing the Heterogeneous Autoregressive Regression model with new methods for deriving volatility estimators from option price data improves daily stock volatility forecasts.
2 sharesSource ↗
Short-Selling Hedge Funds Success
Hedge funds involved in short-selling show superior performance and unique trading patterns, often trading against retail trading trends, contributing to their exceptional performance.
2 shares1 citation todaySource ↗
The Forking Effect in Cryptocurrency
Cryptocurrency forking events don't significantly impact bitcoin's returns but increase its volatility, which remains high for three days post-fork.
2 sharesSource ↗
Investor Reactions to Alt Data
Investors only significantly respond to alternative data in financial decisions if it aligns with previous financial reports, according to a study.
3 sharesSource ↗
Firm Characteristics and Equity Options
A research found that firm characteristics greatly affect equity option prices, and machine learning can enhance option pricing by pooling similar stock information.
2 sharesSource ↗
Short Selling and Info Risk Pricing
Short selling influences the pricing of the probability of informed trading (PIN), especially after good news and among small firms, a study reveals.
2 sharesSource ↗
Reducing Carbon Footprint of Index
An index-based portfolio's carbon footprint can be reduced by over 50% with low active risk by focusing on low emission stocks, according to the authors.
2 sharesSource ↗
Valuation of Levered Equity and Debt Shield
The note argues that discounting the debt tax shield at the cost of debt capital gross of corporate tax has several advantages, including not needing to forecast speculative tax shields due to future net borrowings.
2 sharesSource ↗
Retail Investor Trading Effects
Google Trends data shows that inexperienced retail investors can negatively affect the cost of capital, future performance, and value of real options, especially for smaller firms with less institutional ownership.
2 sharesSource ↗
Bitcoin Inflation Hedge
Research suggests that Bitcoin's returns significantly increase after a positive inflation shock, indicating its potential as an inflation hedge.
2 shares7 citations todaySource ↗
Public Firm Ownership
A new dataset, derived from regulatory filings, offers a detailed view of public firms' ownership structures and institutional managers' investment holdings, surpassing the limitations of commercial databases.
2 sharesSource ↗
Insider Trading Dynamics
A study on insider trading in the U.S. shows that most of it happens during periods of high information asymmetry, with trading during low asymmetry periods showing a strong self-selection bias.
2 sharesSource ↗
RePEc
Economics working papers from RePEc's NEP field reports.
17 items
Finance10
Global Volatility Impact
Global factors significantly influence the local volatility persistence in equity indices of 17 developed economies.
24 sharesSource ↗
Currency Hedging Strategy
A new dynamic currency hedging strategy for global investors is introduced, which is more stable, robust, and risk-reducing than other methods.
22 sharesSource ↗
TimeFrequency Correlation in Oil Markets
Short-term trading significantly influences price volatility in crude oil futures markets, especially between mature and emerging markets.
19 sharesSource ↗
OR Insights on System Dynamics
Operational research methods are used to analyze risks and dynamics in financial and economic systems.
18 sharesSource ↗
Return and Volatility in NFT Markets
The relationship between NFTs and conventional currencies is weak, but NFTs have increased influence during the Covid-19 crisis.
16 sharesSource ↗
Cryptocurrencies in Stock Markets
High volatility persistence is found in stock and cryptocurrency markets, with Tether being the most effective diversifier during market turmoil.
14 sharesSource ↗
GARCH-MIDAS Volatility Models Evaluation
The research suggests that GARCH-MIDAS models may overstate the influence of macro-variables in forecasting total variance, warning of data-mining bias.
12 sharesSource ↗
Commodity Financialization Convenience Yield
The paper reveals that soybeans' convenience yield is affected by financial markets volatility and global macroeconomic variables, supporting the concept of commodity financialization.
12 sharesSource ↗
Insider Silence Strategy Performance
The study concludes that insider trading strategies, particularly buying insider purchases and selling insider sales, yield higher profits over longer holding periods.
12 sharesSource ↗
Portfolio Optimization Numerical Integration
The paper introduces an efficient numerical integration method for Mean-Variance portfolio optimization, demonstrating its effectiveness in various investment scenarios.
10 sharesSource ↗
Statistical3
Machine Learning Workflow for China's Financial Crisis
Researchers have developed a multistep workflow using machine learning to predict China's systemic financial crises, successfully identifying six high-risk periods from 1990 to 2020.
15 sharesSource ↗
Forecasting Gold Price
The paper suggests using the eXtreme Gradient Boosting (XGBoost) machine learning model and Shapley additive explanations (SHAP) for accurate forecasting and interpretation of gold price fluctuations, surpassing other advanced models.
13 sharesSource ↗
Panel Data Nowcasting: Price-Earnings Ratio Prediction
Price-Earnings Ratio Prediction: The article talks about using structured machine learning regressions to predict corporate earnings. This method, which uses mixed-frequency time series panel data, has shown better results than traditional forecasting methods.
20 sharesSource ↗
Historical Trending4
LeadLag Relationship Between Futures and Cash Index
Research shows that CSI 300 index futures typically precede the cash index by 0-5 minutes, but this varies with market conditions.
21 sharesSource ↗
Mixed-Frequency Volatility Model
The MF-MoP model, based on predictability momentum, is more effective than GARCH and Realized GARCH models in predicting financial asset volatility.
17 sharesSource ↗
Bayesian Comparison of GARCH and Stochastic Volatility
Stochastic volatility models, particularly the SV-M model, are more effective than GARCH models in modelling inflation rates across 18 developed countries.
17 sharesSource ↗
Jumps and Margin Level Study
The SE-SVCJ model, combined with a GPD, provides a more accurate forecast of margin levels in the stock index futures market.
14 sharesSource ↗
Papers with code
Papers that shipped their code, from the Papers with Code feed (2023-25).
11 items
Trending5
LLM Data Interpreter
The article explores the effectiveness of Large Language Model (LLM) based agents and provides a GitHub link to the code.
36,459 shares
Generalization Beyond Overfitting
The study suggests examining the generalization of neural networks on small, algorithmically generated datasets, with the code on GitHub.
2,517 shares
Decompiling Binary Code
The article announces the release of the first open-access decompilation LLMs, pretrained on C source code and assembly code, with the code on GitHub.
1,198 shares
GSPMD: Scalable Parallelization
Scalable Parallelization: The paper presents GSPMD, a compiler-based parallelization system for machine learning computations, with the code on GitHub.
654 shares
pyvene: PyTorch Models Library
PyTorch Models Library: The article highlights the significance of interventions on model-internal states in AI, including model editing and robustness, with the code on GitHub.
279 shares
Rising6
Fast InnerProduct Algorithm
The article discusses the Freepipeline Fast Inner Product (FFIP), a new algorithm and hardware architecture that improves upon the fast inner product algorithm (FIP) introduced by Winograd in 1968.
213 shares
Chronos: Time Series Language
Time Series Language: The article introduces Chronos, a new framework designed for pre-trained probabilistic time series models.
165 shares
Sequoia: Speculative Decoding
Speculative Decoding: The paper presents Sequoia, a new algorithm designed for speculative decoding that is scalable, robust, and hardware-aware.
135 shares
D Gaussian Splatting Optimization
The article introduces RAINGS, a new optimization strategy for training 3D Gaussians from random point clouds, following a detailed study of SfM initialization and 1D regression tasks.
99 shares
Monolithic Preference Optimization
The article highlights the crucial role of supervised finetuning (SFT) in achieving successful convergence in preference alignment algorithms for language models.
70 shares
Sparse MixtureofExperts Implementation
The article presents ScatterMoE, a GPU-based implementation of the Sparse MixtureofExperts (SMoE) model.
60 shares
GitHub
Repositories the letter featured.
10 items
Finance5
DiCE for Diverse Explanations
The article provides a guide on generating different counterfactual explanations for any machine learning model.
1,251 shares
Jupyterquant: Dockerized Research Environment
Dockerized Research Environment: The piece offers a tutorial on setting up a Jupyter quant research environment using Docker.
31 shares
Chronosforecasting: Language Models for Forecasting
Language Models for Forecasting: The article introduces Chronos, a pre-trained language model specifically developed for probabilistic time series forecasting.
195 shares
LlamaGym: Finetuning LLM Agents
Finetuning LLM Agents: The article explains the process of improving LLM agents through online reinforcement learning methods.
725 shares
Ubicloud: Portable Cloud Services
Portable Cloud Services: The piece presents a new free, open-source cloud service in its public beta version, featuring elastic compute block storage and managed Postgres.
2,512 shares
Trending5
bigAGI
A new AI suite, powered by the latest LLMs, offers features like AI personas and voice response, with deployment options on-premises or in the cloud.
3,235 shares
Mindgraph
A prototype has been developed that uses AI to generate and query an ever-growing knowledge graph.
552 shares
Podcasts
Episodes on markets, quant methods and economics.
2 items
Related2
Quantitative Investing
Jim O'Shaughnessy, founder of O'Shaughnessy Asset Management, discusses the importance of data in improving portfolio returns in an interview with Barry Ritholtz.
5 shares
Fed Signals
Ira Jersey, Bloomberg Intelligence's lead US interest rate strategist, explores the Federal Reserve's communication methods and the influence of government fiscal policy on the economy.
3 shares
Videos
Talks, lectures and tutorials.
4 items
Quantitative4
Generating Correlation Matrices
The video on YouTube explains the Onion Method, a technique used to create random correlation matrices for statistical analysis.
0 shares
Python AI Finance
Python Quants GmbH is hosting an information session about their Certificate in Python for Finance program.
1 shares
Maximizing Internship
A video offers tips on maximizing internships in the quant field, highlighting the significance of networking and diligence.
22 shares
Maximum Ignorance
A blog post explores the difficulty of estimating probabilities in scenarios with no previous data, using a surgeon's failure rate as an example.
284 shares
X / Twitter
Posts from quant researchers on X.
12 items
Quantitative6
Machine Learning in Finance
The article reviews recent studies on the use of machine learning in asset pricing and corporate finance.
11 shares
Research in Finance
The article recaps the week's research papers on topics such as empirical asset pricing, machine learning, macro options, and industry insights.
6 shares
Stock Option Hedging
The article explores the growing impact of delta hedging on stocks due to increased stock option volume, highlighting a study on expected hedging demand and its strong return predictability.
5 shares
Natural Resource Equities vs. Futures
Man Group suggests that investing in natural resource equities offers better diversification benefits than futures.
3 shares
Interview with Clifford Asness
In a recent interview, Clifford Asness discusses the 60/40 portfolio diversification strategy and market efficiency.
3 shares
Unified Time Series Model
The pre-trained LM UNITS is a comprehensive time series model capable of performing tasks like classification, forecasting, imputation, and anomaly detection.
2 shares
Miscellaneous6
Event Risk Impact on Shortdated Options
A study explores the effect of event risk on short-term options, deriving higher-order moments for option portfolios and identifying substantial macroeconomic event risk.
2 shares
ESG Integration Across Asset Classes
ManGroup provides insights on incorporating Environmental, Social, and Governance (ESG) elements into systematic investing across various asset classes.
1 shares
Macroeconomics Textbook Updates
A new textbook on macroeconomics is now accessible in eBook format.
0 shares
Price vs. Accounting
The paper suggests that accounting-based information is more precise for predictions longer than a month, contrary to prior studies favoring price-based indicators.
0 shares
Market Timing
Jonathan Kinlay explores the frequently misunderstood ability of market timing.
0 shares