Uncovering Price Anomalies with AI Reasoning
A new framework uses artificial intelligence to break down complex price checks into understandable steps, mirroring human audit processes.
Science
Beyond SVI: Machine Learning Uncovers New Volatility Models
![Symbolic regression identified six expressions-[latex] f_1, \dots f_6 [/latex]-that surpass the performance of standard SVI, demonstrating the potential for simpler, more interpretable models to achieve superior results on the efficient frontier and hinting at the limitations of even sophisticated algorithms when confronted with the inherent complexity of the data.](https://arxiv.org/html/2603.21892v1/images/efficient_frontier.png)
Researchers are leveraging symbolic regression to automatically discover parametrizations of implied volatility, achieving performance comparable to-and sometimes exceeding-the widely used SVI model.
The Rise of AI Compute Futures
As demand for artificial intelligence surges, a new market for tradable compute power is emerging to manage price risk and volatility.
From Lab to Live: Scaling AI Trading Strategies
A new modular infrastructure, FinRL-X, aims to streamline the notoriously difficult process of deploying research-driven quantitative trading algorithms into real-world markets.
Spotting the Fake: New Method Detects AI-Generated Images Without Training
Researchers have developed a computationally efficient technique for identifying images created by artificial intelligence, bypassing the need for traditional training datasets.
Beyond the Abstract: An Agent for Discovering Truly Novel Research
A new multi-agent system, NoveltyAgent, automates the process of identifying genuinely new contributions in academic literature, moving beyond simple keyword comparisons.
Smarter Grids, Fewer Losses: AI Spots Electricity Theft
A new machine learning framework leverages spatial and temporal data to significantly improve the detection of electricity theft in modern power grids.
When Financial AI Gets It Wrong: A New Benchmark for Truthfulness
![The study demonstrates that the proposed algorithm achieves superior performance across all evaluated methods, consistently minimizing the error function defined as [latex] E = \sum_{i=1}^{n} |y_i - \hat{y}_i| [/latex], where [latex] y_i [/latex] represents the actual value and [latex] \hat{y}_i [/latex] the predicted value for the i-th data point in the dataset.](https://arxiv.org/html/2603.20252v1/images/best_scores_bar_chart.png)
Researchers have created a rigorous new test to expose how easily financial question-answering systems, even those powered by knowledge graphs, can be misled and generate inaccurate responses.
Decoding Market Signals: A New Approach to High-Frequency Trading
Researchers have developed a novel model that dynamically focuses on different time scales to improve the prediction of rapid changes in order flow and volatility.
Decoding Market Signals with Intelligent Agents
A new multi-agent system uses the power of language models to automatically discover and refine the underlying logic driving financial markets.