Quant LetterNo. 31
109 items across 10 sections, as sent to readers on 3 January 2024. Paper titles open their ML-Quant page; ↗ goes to the source.
Quantitative-finance and ML-for-finance preprints from arXiv.
22 items
A new forecasting model is suggested that uses past returns and their squares to predict implied volatility surfaces and underlying asset returns for up to two years.
8 shares2 citations todaySource ↗
A formula is introduced for the conditional probability of a portfolio based on its optimal common drivers, aiding in dynamic risk management.
7 shares2 citations todaySource ↗
A paper finds that a simple shadow price strategy for maximizing long-term returns given average volatility is nearly optimal, but suggests alternative strategies for different risk aversions.
6 shares1 citation todaySource ↗
AI and machine learning are revolutionizing quantitative trading with advanced algorithms, including a new model, QTNet, that uses deep reinforcement learning to manage volatile financial data.
6 shares27 citations todaySource ↗
The study explores dynamic risk management of potential credit losses on a derivatives portfolio, using recent advancements in risk-averse Reinforcement Learning for option hedging.
5 shares2 citations todaySource ↗
The article discusses a new reinforcement learning-based model for analyzing real-world data, which is more applicable than the continuous-time model.
5 shares2 citations todaySource ↗
10B-Level Language Model for Asset Management: The article introduces Shai, a superior language model specifically designed for the asset management industry, providing practical financial insights.
4 shares5 citations todaySource ↗
Framework for Understanding Relationships: The article enhances the Constraint-based Causal Discovery algorithm to identify intricate causal relations between financial assets and variables, useful for factor-based investing and market dynamics comprehension.
4 shares8 citations todaySource ↗
The article introduces a SHAP-based explainability technique for financial forecasting, increasing transparency and confidence in the stock exchange market.
3 shares2 citations todaySource ↗
The study presents a computational framework for simulating large-scale agent-based financial markets, useful for machine learning applications and market microstructure research.
3 shares4 citations todaySource ↗
The research uses neural networks to predict exports of certain OECD countries and Iran from 2021-2025, suggesting that long-term export contracts are less impacted by crises like Covid-19, and should be considered in economic policies.
5 sharesSource ↗
A new machine learning model can predict gentrification using socioeconomic data and images, highlighting a significant connection between gentrification and schools.
5 sharesSource ↗
Large Language Models (LLMs) display human-like social behaviors but also have significant differences, necessitating further research for accurate human behavior emulation.
4 shares71 citations todaySource ↗
A new content recommendation model takes into account both user and creator engagement, suggesting that neglecting creator departures can lead to reduced total engagement, and provides two algorithms for improved performance.
4 shares6 citations todaySource ↗
A study found that LightGBM was the best for fraud detection when comparing anomaly detection and standard supervised learning methods, but it was more susceptible to distribution shifts, questioning the advantage of combining these two methods.
2 shares16 citations todaySource ↗
A study involving blockchain experts found that the technology could be successfully integrated into certain areas of the circular economy under specific conditions, despite some integrations being unlikely to work.
4 shares15 citations todaySource ↗
A new algorithm using limit order book data and a continuous output error model has been developed, providing accurate predictions of cryptocurrency returns and outperforming other models in accuracy and profit in a trading environment.
4 shares4 citations todaySource ↗
The paper introduces a jump risk dimension to the market price of risk for electricity swap contracts, improving previous models by accounting for jumps and mean-reverting behavior.
246 shares1 citation todaySource ↗
The research uses Tsallis Entropy in models with latent factors to optimally control and explore the state space, proving that the optimal state distribution is q-Gaussian, which can be used in creating robust statistical arbitrage trading strategies.
53 shares7 citations todaySource ↗
The article introduces a generative adversarial network (GAN) method for calculating volatility surfaces, showing that the GAN model is more accurate and faster than artificial neural network (ANN) methods.
50 shares3 citations todaySource ↗
The study highlights the effectiveness of robust risk-aware reinforcement learning in managing risks related to path-dependent financial derivatives, especially in hedging barrier options, proving robust strategies are superior.
56 shares19 citations todaySource ↗
The paper introduces a method to identify clusters and estimate cluster-specific regression parameters using Variational Inference (VI), which is ideal for financial forecasting in markets with different regimes and market change patterns.
45 shares3 citations todaySource ↗
Working papers in finance and economics from SSRN.
26 items
The article introduces a formula for calculating the conditional probability of a portfolio based on its optimal common drivers, offering new risk metrics.
13 sharesSource ↗
The study presents a method for identifying change points in time series data, including financial data, using a trained neural network, offering new tools for financial market analysis.
6 sharesSource ↗
Reshaping Index Construction: The authors present Machine Beta, a method that uses statistical factors and non-linear mechanisms to correct biases in market capitalization-weighted benchmarks, aiming to achieve lower tracking errors and outperform these benchmarks.
7 sharesSource ↗
The study uses risk-averse Reinforcement Learning for managing potential credit losses on a derivatives portfolio, proving its effectiveness through a numerical study for a portfolio consisting of a single FX forward contract.
2 shares2 citations todaySource ↗
The paper demonstrates the use of explainable artificial intelligence in empirical asset pricing, showing enhanced predictive power and investment performance when incorporating insights from explainable AI into model refinement.
2 shares3 citations todaySource ↗
A new method using genetic programming to build the best mean-variance portfolio has been suggested, which significantly improves the Sharpe ratio and outperforms other machine learning techniques.
3 sharesSource ↗
The article shows how the XGBoost machine learning algorithm can predict next-day volatility jumps based on firm-specific news, leading to improved portfolio performance.
2 sharesSource ↗
Sell-side analysts using alternative data in their analyses generate more accurate earnings forecasts and earn higher trading commissions.
2 sharesSource ↗
The paper introduces a new method to estimate daily mutual fund trades in individual stocks using daily stock prices, returns, and quarterly fund holdings, showing high accuracy for larger trades.
3 shares2 citations todaySource ↗
The research indicates that machine learning can enhance earnings forecasts, especially for small firms and longer horizons, and that investors' expectations align with the best machine forecast.
2 sharesSource ↗
The CNNLSTM hybrid machine learning approach enhances prediction accuracy in emerging markets' connected economic sectors.
2 sharesSource ↗
The growing use of algorithmic tools in financial advisory prompts questions about the sufficiency of current regulatory frameworks.
2 sharesSource ↗
The paper discusses the potential of quantum machine learning as an efficient alternative to classical machine learning in financial risk management.
2 sharesSource ↗
The article presents new risk management frameworks for high probability forex trading, providing practical strategies and models for traders.
2 sharesSource ↗
Momentum and Low Risk Anomalies: The new model expands the traditional Capital Asset Pricing Model (CAPM) by factoring in idiosyncratic tail risk, explaining momentum in stock returns and low risk anomalies.
3 shares1 citation todaySource ↗
The research shows that uncertainty in commodity futures markets influences hedging behavior, with swap dealers increasing their hedging demand and commodity producers decreasing their activity during uncertain times.
8 sharesSource ↗
The article suggests a new simulation scheme for the multifactor OrnsteinUhlenbeck stochastic volatility model that is simpler to use, quicker to run, and offers better error control.
4 sharesSource ↗
A study introduces a saddlepoint approximation for credit portfolio losses in continuous time models, providing a more efficient algorithm that greatly improves on recursive methods.
3 shares2 citations todaySource ↗
The stochastic volatility model shows that certain stochastic discount factors can cause a bubble in wealth processes and derivatives, but not in stocks or risk-free bonds.
4 sharesSource ↗
Market Implications: Financial innovations like Exchange-Traded Funds and smart beta products, modeled as composite securities, simplify trading for investors and attract more factor investors.
2 sharesSource ↗
The study analyzes cashflows in private investment strategies using a comprehensive dataset, demonstrating the effectiveness of the Yale model and suggesting improvements.
4 sharesSource ↗
The study shows that large fund trades stabilize the corporate bond market but can introduce fragility during illiquid markets, with bond return volatility and fund size inversely related.
2 sharesSource ↗
Research shows non-U.S. stock returns are more influenced by U.S. stock returns than U.S. mutual fund price pressure, highlighting cross-border arbitrage barriers.
2 sharesSource ↗
Short sellers are found to be more informed than margin traders in leveraged investing, as they predict returns more effectively.
2 sharesSource ↗
A study finds that investors form separate beliefs on cash flow level and growth, explaining half of the anomaly portfolios' deviation from the CAPM.
2 sharesSource ↗
The authors discuss the effect of transaction costs on ESG-aware portfolios, concluding that simple liquidity constraints can reduce market impact while maintaining a good ESG profile.
3 sharesSource ↗
The general machine-learning papers the letter carried in 2023-25.
9 items
The study introduces a hash table with learned probes for neural graphics primitives, providing a balance of size and speed, and outperforming previous index learning methods.
8 shares35 citations todaySource ↗
Implications and Interventions: The use of strategies by users can initially benefit data-driven platforms, but can eventually corrupt data and affect decision-making, highlighting the need for reliable algorithms.
13 shares8 citations todaySource ↗
SparseProp, an event-based algorithm for simulating and training large-scale spiking neural networks, is introduced, offering reduced computational cost and efficient training.
8 shares15 citations todaySource ↗
Google's Gemini, a Multimodal Large Language Model, is evaluated across 12 commonsense reasoning datasets, showing competitive reasoning abilities and the need for further model advancements.
6 shares29 citations todaySource ↗
A Transfer Causal Learning framework has been developed to improve the accuracy of causal effect estimation in scenarios with limited data, such as rare medical conditions.
15 shares4 citations todaySource ↗
A new dataset combining numerical stock data with qualitative text data for sentiment extraction is presented, achieving over 60% accuracy for the Dow Jones Industrial Average.
23 shares1 citation todaySource ↗
PIXART-$\alpha$, a Transformer-based text-to-image model, generates high-quality images at a low cost, reducing CO2 emissions and offering a cost-effective solution for the AIGC community.
166 shares967 citations todaySource ↗
Angelopoulos et al.'s method provides valid inference on the association between unobserved response and covariates, regardless of the quality of the pre-trained machine learning model, unlike Wang et al.'s method.
49 shares13 citations todaySource ↗
Motivations, Challenges, Future Directions: The combination of Foundation Model (FM) and Federated Learning (FL) enhances AI research by increasing data availability and improving performance and convergence speed.
20 shares142 citations todaySource ↗
Papers that shipped their code, from the Papers with Code feed (2023-25).
7 items
Customizable Hotword ASR System: The paper explores a model that merges the precision of AED-based models, the efficiency of NAR models, and the ability to customize for enhanced performance.
1,880 shares
GPU-based Language Model Serving: The article presents PowerInfer, a fast Large Language Model inference engine designed for personal computers with a single consumer-grade GPU.
4,843 shares
Multilingual Visual Text Generation and Editing: The paper introduces AnyText, a benchmark for assessing the accuracy and quality of visual text generation, using the AnyWord3M dataset.
428 shares
The study involves a comprehensive comparison of chaos forecasting methods, testing 24 methods on a database of 135 low-dimensional systems using 17 forecast metrics.
309 shares
The article provides an overview of Google's Gemini model class, the first to rival OpenAI's GPT series in multiple tasks.
114 shares
The article delves into the difficulties of running large MoE language models on consumer-grade hardware due to limited accelerator memory.
280 shares
The article presents a unique approach to depicting human lives in a format that structurally mirrors language.
152 shares
Repositories the letter featured.
7 items
The article explores the application of BERT in Financial Sentiment Analysis.
1,220 shares
The article conducts a baseline analysis of Frequency Interpolation Time Series (FITS).
14 shares
The article shares handwritten notes and source code from the author's Machine Learning Simulation YouTube videos.
561 shares
Practical Business Python offers code notebooks and examples for educational purposes.
1,948 shares
Mixtral8x7B models can now be operated on Colab or personal desktops.
1,037 shares
Ollama has launched a text-based terminal client for user ease.
253 shares
A detailed list of engineering blogs has been compiled for reference.
26,784 shares
Industry news: funds, hiring, markets and regulation.
7 items
Despite volatile trading conditions, investors in DE Shaw's largest hedge fund experienced nearly a 10% return in 2023.
3 shares
Kelly Young has been appointed as the new CEO of Acadian Asset Management, while also continuing her role in the firm's Executive Committee.
3 shares
The article discusses the roles and strategies required to earn a significant income as a quant researcher.
3 shares
Steve Cohen's Point72 Asset Management is among investors shorting shares of British company Hargreaves Lansdown.
3 shares
Top trading professionals are engaging in discussions about future trends, opportunities, and potential risks in the industry.
3 shares
Quants and technologists are expected to spend more time physically in the office this year.
2 shares
Radcliffe Capital Management consistently adopts strategies to identify significant mispricing, with the firm's principals investing alongside its clients.
2 shares
Episodes on markets, quant methods and economics.
7 items
The episode discusses the use of large language models (LLMs) programmatically, which are now accessible to all through affordable API options despite their high cost.
3 shares
Quantamental Approach to ESG: Liz Simmie, co-founder of Honeytree Investment Management, shares insights on their ESG-focused ETF, BEEZ, and the state of ESG and active management.
4 shares
A busy professional shares four tips on managing a hectic schedule that includes family, work, two YouTube channels, and various hobbies, admitting that things don't always go as planned.
2 shares
Low on Juice: Sean McGovern, VP of Research at McAlinden Research Partners, discusses the recent increase in frozen concentrated orange juice futures and the factors driving costs higher.
3 shares
Erik Townsend and Patrick Ceresna of MacroVoices discuss with Jim Bianco about the FOMC's monetary policy change, with Bianco suggesting that peak yields are still to come.
3 shares
Richard Vague's book A Brief History of Doom explores the cycle of economic crises over the past 200 years, attributing them to fluctuations in private sector debt.
3 shares
Thomas Hayes, Chairman of Great Hill Capital, discusses investment strategies and opportunities in small and mid-cap sectors, emerging markets, and China's economy.
10 shares
Posts from quant and economics blogs and newsletters.
5 items
Supertrend Indicator Trading Strategy outlines how to use the supertrend indicator in trading.
2 shares
Day Trading Statistics 2024 The Truth explores the current state and future projections of day trading statistics.
6 shares
The article shares Bloomberg's annual survey results, forecasting a 6.2% increase for the S&P500 index by the end of 2023.
1 shares
The article reiterates Bloomberg's annual survey's prediction of a 6.2% rise in the S&P500 index by December 2023.
1 shares
The author examines the creation of new indicators and the criteria for assessing their quality, citing various statistical significance tests and resources.
1 shares
Talks, lectures and tutorials.
5 items
The YouTube Short educates on portfolio management, systematic trading, and optimization techniques used by junior managers in multistrategy hedge funds.
0 shares
HKML EduTech's video introduces Information Coefficient in quantitative trading, demonstrating Python code to calculate forward returns and measure the IC.
3 shares
The YouTube Short explains the impact of portfolio turnover on trading costs and strategy effectiveness, and how to utilize turnover insights to enhance trading strategies.
0 shares
5 Key Questions Answered: The video offers guidance on transitioning from academia to quant finance, including tips on resume building, firm research, self-promotion, and choosing between buy-side and sell-side roles.
24 shares
Avoiding Model Failure: The article cautions against using Out-of-Sample testing over Out-of-Time testing in financial modeling, highlighting the risk of information leakage and model failure.
19 shares
Posts from quant researchers on X.
14 items
A Neurocomputing magazine article introduces a framework to prevent data leaks in machine learning for the finance sector.
6 shares
Microsoft's InsightPilot, powered by LLM, automates data analysis to simplify data exploration.
3 shares
Language Model for Asset Management: China Asset Management Ltd researchers have created Shai, a language model for the asset management industry.
2 shares
ManGroup explores the use of generative AI in investment management.
1 shares
The article explores the identification of cause-and-effect relationships from observational time series data without the need for stationarity adjustments.
1 shares
The article presents Functime, a new time series model designed for production use, featuring automated feature extraction and panel set capabilities.
1 shares
The article introduces tclf, a new trade classification algorithm compatible with scikit-learn, designed for use in financial markets.
1 shares
The article examines the potential impact of Large Multi Media and Large Language Models on traditional business analytics, questioning their current scope.
1 shares
Google unveils TSMixer, a new forecasting model, in its TimeSeries Thursday series.
0 shares
MarketGuided Stock Transformer: The article presents the MASTER MarketGuided Stock Transformer for predicting stock prices.
0 shares
Ethereum Prediction: The TimeSeries Thursday series applies TimesNet timeseries prediction to Ethereum.
0 shares
The article provides a guide on constructing a Transformer with Attention in Python without training.
0 shares
A study reveals that despite a decline among G10 currencies, FX carry returns remain robust in exotic currencies after the Global Financial Crisis.
0 shares