ML Pricing for Multidimensional Passport Option
The study proposes a discrete-time solution for pricing passport options in multi-dimensional Black-Scholes markets using two machine learning methods.
6 sharesSource ↗
Quant LetterNo. 10
August 2023, Week 1
67 items across 5 sections, as sent to readers on 2 August 2023. Paper titles open their ML-Quant page; ↗ goes to the source.
- Items
- 67
- Papers
- 54
- Sections
- 5
- Issue
- No. 10
arXiv
Quantitative-finance and ML-for-finance preprints from arXiv.
15 items
Finance6
Profiting from Commodity Futures Curve Dynamics
The research uses the Nelson-Siegel framework to model commodity futures prices and develops profitable investment strategies based on slope changes.
5 shares7 citations todaySource ↗
Option Smile Volatility & Implied Probabilities in Earnings Announcements
The paper investigates the effect of earnings announcements on stock price volatility, revealing that investors pay a significant premium to hedge against announcement-induced uncertainty.
4 sharesSource ↗
Electricity Intraday Model
The article presents a statistical model for daily electricity prices, which can simulate all products at once, replicating the Samuelson effect and price correlation structure, and shows its application in storage valuation.
3 shares6 citations todaySource ↗
Valuing Passport Options in Lévy Model
The study investigates the valuation of an unusual derivative called the American passport option, formulates the pricing equation, and proves that the option value is a viscosity solution of variational inequality.
2 sharesSource ↗
Causal Inference in Banking, Finance, and Insurance Survey
The paper reviews 37 studies on the use of causal inference in banking, finance, and insurance from 1992 to 2023, categorizing them and discussing the statistical methods used, while highlighting that this application is still in its infancy.
3 shares8 citations todaySource ↗
Crypto & Blockchain3
Interactive Alpha Mining
Alpha-GPT is a new mining paradigm that uses human-AI interaction and a unique algorithm to understand quantitative researchers' ideas and generate efficient trading signals.
13 shares81 citations todaySource ↗
Bitcoin and Litecoin Analysis
A study confirms the comparison of Bitcoin to gold and Litecoin to silver, highlighting Bitcoin's superior value storage capacity compared to Litecoin.
10 shares12 citations todaySource ↗
Shifting Influence in Cryptocurrency Market
A cryptocurrency market analysis shows Ethereum as the market leader, with influential cryptocurrencies gaining more power over time, despite the increasing number of cryptocurrencies.
5 sharesSource ↗
Historical Trending6
Efficient Learning in OTC Markets
The study uses deep reinforcement learning to balance hedging and skewing in a game between liquidity providers and takers in an over-the-counter market, introducing a new algorithm to impose constraints on the game's equilibrium.
139 shares23 citations todaySource ↗
Modular Networks and Technological Sophistication
The research uses partial information decomposition to find that industries with more complexity have small-world topologies, and countries and industries with a well-connected core and specialized modules have higher economic efficiency.
70 shares5 citations todaySource ↗
Impact of HFT Front-Running on Traders
The paper concludes that high-frequency traders always front-run and large traders benefit when there is enough high-speed noise trading and the high-frequency trader's prediction is unclear.
34 shares3 citations todaySource ↗
Lévy Models: Calibration and Pricing
Calibration and Pricing: The paper investigates the impact of multivariate Lévy models' structures on calibration and pricing, using various methods to assess their fit with market data and pricing of exotic derivatives.
32 shares2 citations todaySource ↗
Systemic Risk Measures: A Mixed-Integer Programming Approach
A Mixed-Integer Programming Approach: The research proposes a mixed-integer programming problem to compute systemic risk measures in systems with complex clearing mechanisms.
13 shares13 citations todaySource ↗
Leverage, Growth, and Asset Price Bubbles
The paper introduces a new theoretical framework to understand asset price bubbles in dividend-paying assets, examining a macro-finance model with a positive feedback loop between capital investment and land price.
11 shares10 citations todaySource ↗
SSRN
Working papers in finance and economics from SSRN.
27 items
Quantitative12
Computing Multi-Eigenpairs with Tensor Networks
The paper introduces a tensor-neural-network-based machine learning method to accurately compute multi-eigenpairs of high dimensional eigenvalue problems.
2 sharesSource ↗
ML Control for Counterfactual Forecasting
The Machine Learning Control Method (MLCM) is a new tool for analyzing causal effects without a control group, recently used to study COVID-19's impact on income inequality in Italy.
2 sharesSource ↗
ML vs. Econometrics for Real GDP Predictions
Machine learning algorithms are more accurate than OLS models in predicting GDP growth, but understanding economic causality requires knowledge in economics and econometrics.
2 sharesSource ↗
Funding Constraints and Liquidity of VIX Futures
The study suggests that funding constraints significantly impact the trading costs of VIX futures, with leveraged traders playing a key role.
3 sharesSource ↗
Crop Diversification Strategies for Farmers
The research reveals that poor and nonpoor farmers adopt different crop diversification strategies in response to environmental and economic challenges.
2 sharesSource ↗
Private Bonds and Corporate Investment
The research finds that increased availability of private activity bonds leads to a rise in corporate investment and employment.
10 sharesSource ↗
Optimal Securities Screening
The article explores how asset owners can create asset-backed securities to raise funds from knowledgeable liquidity suppliers, potentially benefiting from their expertise.
72 sharesSource ↗
Pricing Power of Perpetual Futures
The study presents a new method for pricing power perpetual futures with stochastic volatility, enhancing the existing deterministic volatility framework, particularly for crypto trading.
2 sharesSource ↗
Illusions in Equity
The research investigates employees' understanding of equity-based compensation in venture-backed startups, revealing frequent misinterpretations and exploitable market illusions.
2 sharesSource ↗
Overnight GARCH-Itô Models
The paper presents a unified factor overnight GARCH-Itô Models model for estimating and predicting large volatility matrices, suggesting a weighted least squares estimation procedure with a nonparametric factor volatility estimator.
2 sharesSource ↗
Probabilistic Forecasting of Electricity Prices
Research improves probabilistic electricity price forecasting using artificial neural networks, achieving similar results to benchmarks but with lower computational cost.
2 shares5 citations todaySource ↗
Machine Learning for Exam Fraud Detection
A study successfully uses machine learning to detect exam cheating, enhancing the credibility of the examination system.
2 sharesSource ↗
Financial15
Portfolio Choice with Transaction Costs
CARA investors use a constant trading speed to balance their portfolio, taking into account trading costs and execution risks, to optimize past trades and future investment opportunities.
33 shares4 citations todaySource ↗
NLP in Accounting and Finance: Performance Comparison
Performance Comparison: Despite the success of deep learning in Natural Language Processing, traditional machine learning and rule-based methods are still prevalent in accounting and finance, with deep learning performing best overall.
2 shares2 citations todaySource ↗
Expanding Fama-French Model with Industry Beta
The news-based stock pricing model (NBSPM) performs better than the five-factor Fama-French model (FF5M) for US equity sector ETFs, but adding industry beta to FF5M improves its accuracy, though not as much as NBSPM.
10 sharesSource ↗
Deep Learning for Corporate Bonds
A U.S. corporate bonds market asset pricing model shows that maximizing the Sharpe ratio performs better for individual bonds, with significant excess returns shown in out-of-sample annual SDF portfolio Sharpe ratios.
7 shares1 citation todaySource ↗
Big Data Displacing Fund Managers - Fund Managers Displaced by Big Data
The use of big data, like satellite imagery tracking firms' parking lots, can diminish the stock picking abilities of active mutual funds, implying that big data could replace high-skill workers in finance.
2 shares4 citations todaySource ↗
Legal Institutions and Fragile Financial Markets
Equity liquidity varies across countries due to funding constraints and legal institutions, with firms in countries with extensive disclosure requirements experiencing fewer liquidity shocks, leading to higher firm value in volatile markets.
4 sharesSource ↗
Liquidity and Asset Bubbles - Asset Bubbles and Liquidity
Asset price bubbles can be created by the interaction between limited participation and credit lines, benefiting arbitrageurs and liquidity providers, but not regular stockholders due to increased stock volatility.
3 sharesSource ↗
Volatility Dynamics in Cryptocurrencies - Cryptocurrency Volatility Dynamics
The volatility dynamics of cryptocurrencies, which have unique statistical properties, are vital for portfolio managers and traders, with high or low connectedness periods linked to specific crypto market and macroeconomic events.
3 shares2 citations todaySource ↗
Deep Learning for Price Trend Prediction in ETF Markets
The channel and spatial attention convolutional neural network (CSACNN) uses deep learning to predict financial market trends, performing as well or better than models using only time series data.
181 sharesSource ↗
Jump-Diffusion Model for Climate Risk Assessment
A stochastic asset pricing model assesses climate risk at the firm level, examining the impact of climate-related risk factors on stock return volatility and market return correlations.
78 sharesSource ↗
Solvability of Differential Riccati Equations in Algorithmic Trading
A differential Riccati equation (DRE) with indefinite matrix coefficients can solve two algorithmic trading problems using a constant absolute risk-aversion (CARA) utility function.
2 sharesSource ↗
Dollar Asset Holding and Hedging by Institutional Investors
Global institutional investors' holdings of U.S. dollar securities have increased six-fold in the last 20 years, maintaining high hedge ratios despite fluctuations in covered-interest rate parity.
2 sharesSource ↗
Rebalanced Portfolio Optimization without Trade-offs
Regular intraperiod portfolio rebalancing strategies offer a unique solution to portfolio optimization problems without requiring utility or risk tradeoffs.
2 shares1 citation todaySource ↗
Funding Constraints, Crisis, & Price Discovery
Financial crises and funding constraints affect the pricing dynamics between spot and futures markets, discouraging informed investors and impacting the likelihood of informed trading and large trading in futures markets.
2 sharesSource ↗
Sentiment Mispricing and Excess Volatility in Institutional Investors
Sentiment-driven investors and benchmark-focused institutions can lower stock market prices and create new countercyclical patterns in stock volatility.
44 sharesSource ↗
RePEc
Economics working papers from RePEc's NEP field reports.
12 items
Finance7
Algorithmic trading and block ownership
Algorithmic trading decreases the probability of block ownership initiation in U.S. public companies by 3.5%, deterring sophisticated investors from gathering information.
17 sharesSource ↗
Clustering Zero-Inflated Time Series
A novel clustering method for high-dimensional zero-inflated time series data has been developed, utilizing a modified thick-pen transform and an efficient iterative clustering algorithm, proven effective through simulations and real datasets.
15 sharesSource ↗
Heterogeneous Tail Common Factor Modeling
The proposed Factor-HGH model, which handles non-Gaussian errors, shows promise in modeling financial factors and asset returns, especially for cryptocurrencies with highly heterogeneous tails.
14 sharesSource ↗
Factor Models for Large, Incomplete Data Sets
A method combining two algorithms outperforms the standard EM algorithm when data has a grouped factor structure, as shown using the Federal Reserve Economic Data.
14 sharesSource ↗
Portfolio optimization with factors
A new model that minimizes historical data noise and uses machine learning for future predictions enhances portfolio optimization.
23 sharesSource ↗
VaR and ES forecasting in large portfolios
Two novel methods for estimating Value-at-Risk (VaR) and Expected Shortfall (ES) in large portfolios surpass existing techniques, as per backtesting and scoring results.
19 sharesSource ↗
Fama-French Model and Machine Learning
A seven-factor model, including the Hurst exponent and momentum factors, boosts the average R-squared by 7% in the A-share market, with SVM and random forests being the best performing machine learning algorithms.
20 sharesSource ↗
Machine Learning3
Global Economic Policy Uncertainty Index
The research introduces a new global economic policy uncertainty index, combining Principal Component Analysis and Random Matrix Theory, which surpasses current models in detecting global events without requiring additional economic data.
25 sharesSource ↗
Comparing Explainable Machine Learning Methods
The article explores different explainable artificial intelligence methods for data-driven insurance issues, highlighting the need for accuracy and interpretability in choosing a machine-learning model to improve prediction transparency and reliability.
24 sharesSource ↗
Predicting Corporate Fraud with Machine Learning
The study applies a machine learning model using the GONE framework to predict corporate fraud in China, revealing that the Random Forest model is superior and that exposure variables are vital for accurate prediction.
20 sharesSource ↗
Deep Learning2
High-Frequency Trading Volume Prediction with Neural Networks
A neural network study successfully predicted trading volumes of the CSI300 futures index using data from one to thirty minutes ahead, with no improvement from incorporating nearby futures or spot trading volume series.
23 sharesSource ↗
Hawkes Model Parameter Estimation with Recurrent Neural Networks
A study using a recurrent neural network to estimate parameters of a Hawkes model based on high-frequency financial data found it to be faster and comparably accurate to traditional methods, suitable for real-time volatility measurement.
14 sharesSource ↗
Papers with code
Papers that shipped their code, from the Papers with Code feed (2023-25).
5 items
Rising3
Enhancing Document-level Event Extraction
The article delves into the complexities of extracting event arguments from entire documents, highlighting the difficulties of long input and cross-sentence inference compared to sentence-level extraction.
62 shares
Factuality Detection in Gen AI
The paper introduces FacTool, a system designed to identify factual inaccuracies in text produced by large language models such as ChatGPT, irrespective of the task or field.
62 shares
TabR Unleashing Retrieval-Augmented Tabular Deep Learning
The piece investigates unexplored facets of the attention mechanism that greatly impact the performance on tabular data issues.
50 shares
Trending2
Gorilla Model with APIs
Large Language Models (LLMs) have recently made significant strides in areas such as mathematical reasoning and program synthesis.
5,312 shares
LoraHub: CrossTask Generalization with LoRA Composition
CrossTask Generalization with LoRA Composition: Lowrank adaptations (LoRa) are frequently utilized to modify Large Language Models for new assignments.
97 shares
GitHub
Repositories the letter featured.
8 items
Finance5
Open Source ML Framework for All
The software presents a new open-source framework for machine learning that is user-friendly.
176,583 shares
MOSEK Portfolio Optimization Cookbook
The software offers additional resources for the MOSEK Portfolio Optimization Cookbook.
27 shares
Time Series & Spatio-Temporal Forecasting Papers
This repo showcases a collection of academic papers on Time Series and SpatioTemporal Forecasting/Prediction, sorted by model type.
378 shares
Private Document Interaction with EmbedAI
The article unveils an app that enables secure interaction with documents using GPT, preventing data breaches.
2,385 shares
Trending3
File Sync
The article introduces an open-source tool designed for continuous file synchronization.
53,485 shares
Twitter Scraper
The software introduces a 2023 Twitter API scrapper that facilitates extensive data scraping with authorization support.
126 shares
ML Visualizations
The article announces the upcoming revision of the book Machine Learning in 2024, with rewards for error detection.
1,054 shares