This paper presents a novel study on harnessing Large Language Models' (LLMs) outstanding knowledge and reasoning abilities for explainable financial time series forecasting. The application of machine learning models to financial time series comes with several challenges, including the difficulty in cross-sequence reasoning and inference, the hurdle of incorporating multi-modal signals from historical news, financial knowledge graphs, etc., and the issue of interpreting and explaining the model results. In this paper, we focus on NASDAQ-100 stocks, making use of publicly accessible historical stock price data, company metadata, and historical economic/financial news. We conduct experiments to illustrate the potential of LLMs in offering a unified solution to the aforementioned challenges. Our experiments include trying zero-shot/few-shot inference with GPT-4 and instruction-based fine-tuning with a public LLM model Open LLaMA. We demonstrate our approach outperforms a few baselines, including the widely applied classic ARMA-GARCH model and a gradient-boosting tree model. Through the performance comparison results and a few examples, we find LLMs can make a well-thought decision by reasoning over information from both textual news and price time series and extracting insights, leveraging cross-sequence information, and utilizing the inherent knowledge embedded within the LLM. Additionally, we show that a publicly available LLM such as Open-LLaMA, after fine-tuning, can comprehend the instruction to generate explainable forecasts and achieve reasonable performance, albeit relatively inferior in comparison to GPT-4.

Nvidia saw its valuation soar to $1tn on Tuesday, making it the fifth most valuable American company and one of the first major corporate beneficiaries of the hype around AI. The chipmaker has been a major and in some cases dominant player in several industries for years. But no development has raised its profile – and its potential windfall – as much as the current excitement around generative AI. Nvidia has been around for 30 years. The company got its start in 1993 building graphics processing units (GPUs) for video games.

Nvidia Corp stock has soared about 26 percent, taking it closer to a market value of $1 trillion after the chip designer's stellar revenue forecast showed that Wall Street has yet to price in the game-changing potential of artificial intelligence. Thursday's surge added to a more than two-fold rise in the stock this year and was set to increase Nvidia's value by about $196bn to nearly $951bn, putting it on course for the largest single-day value gain for a US firm. That market capitalization makes Nvidia twice the size of the second-largest chip firm, Taiwan's TSMC. In the United States, it trails only the trillion-dollar value companies Apple Inc, Alphabet Inc, Microsoft Corp and Amazon.com The rosy earnings also sparked a rally in the chip sector and for AI-focused firms, lifting stock markets from Japan to Europe.

Large Language Models (LLMs) demonstrate exceptional performance in textual understanding and tabular reasoning tasks. However, their ability to comprehend and analyze hybrid text, containing textual and tabular data, remains underexplored. In this research, we specialize in harnessing the potential of LLMs to comprehend critical information from financial reports, which are hybrid long-documents. We propose an Automated Financial Information Extraction (AFIE) framework that enhances LLMs' ability to comprehend and extract information from financial reports. To evaluate AFIE, we develop a Financial Reports Numerical Extraction (FINE) dataset and conduct an extensive experimental analysis. Our framework is effectively validated on GPT-3.5 and GPT-4, yielding average accuracy increases of 53.94% and 33.77%, respectively, compared to a naive method. These results suggest that the AFIE framework offers accuracy for automated numerical extraction from complex, hybrid documents.

Multi-task learning (MTL) aims at achieving a better model by leveraging data and knowledge from multiple tasks. However, MTL does not always work -- sometimes negative transfer occurs between tasks, especially when aggregating loosely related skills, leaving it an open question when MTL works. Previous studies show that MTL performance can be improved by algorithmic tricks. However, what tasks and skills should be included is less well explored. In this work, we conduct a case study in Financial NLP where multiple datasets exist for skills relevant to the domain, such as numeric reasoning and sentiment analysis. Due to the task difficulty and data scarcity in the Financial NLP domain, we explore when aggregating such diverse skills from multiple datasets with MTL can work. Our findings suggest that the key to MTL success lies in skill diversity, relatedness between tasks, and choice of aggregation size and shared capacity. Specifically, MTL works well when tasks are diverse but related, and when the size of the task aggregation and the shared capacity of the model are balanced to avoid overwhelming certain tasks.

In Meta's quarterly earnings call in April, chief executive Mark Zuckerberg was on the defensive. The metaverse, the vision of a globe-spanning virtual reality that he had literally bet his multibillion-dollar empire on creating, had been usurped as the new hot thing by the growing hype around artificial intelligence (AI). Critics had even noticed Meta itself changing its tune, highlighting the difference between a November statement from Zuckerberg, in which he described the project as a "high-priority growth area" and a March note that instead focused on how "advancing AI" was the company's "single largest investment". Not so, said the world's richest millennial. "A narrative has developed that we're somehow moving away from focusing on the metaverse vision, so I just want to say upfront that that's not accurate. "We've been focusing on AI and the metaverse for years now, and we will continue to focus on both … Building the metaverse is a long-term project, but the rationale for it remains the same and we remain committed to it." But more than 18 months after Facebook changed its name to Meta – demonstrating Zuckerberg's firm belief that "the metaverse will be the successor of the mobile internet" – the future he promised seems no closer to existence than it did backthen. Reams of concept art, tech demos and prototype devices have given way to little meaningful progress. The company has even struggled to actually define what it is hoping to build: in a lengthy blogpost published last May, Nick Clegg, the former UK deputy prime minister who is now Meta's president of global affairs, described the ambition only in vague terms, despite elaborating across 8,000 words how it would nonetheless change the world. "The metaverse is a logical evolution.

Meta is not pivoting away from its signature product, the metaverse. Or at least that's what the Meta chief executive, Mark Zuckerberg, is arguing. Despite reports that sales teams at Meta have spent less time pitching the metaverse to advertisers, Zuckerberg claimed on the tech firm's latest quarterly earnings call that it's business as usual over at the company formerly known as Facebook. "A narrative has developed that we're somehow moving away from focusing on the metaverse vision, so I just want to say upfront that that's not accurate," the CEO said. But neither is the virtual reality world the only product Meta has bet its future on, Zuckerberg argued: "We've been focusing on both AI and the metaverse for years now, and we will continue to focus on both."

Qualcomm has agreed to acquire an Israeli fabless chipmaker called Autotalks, and according to TechCrunch, the deal will cost the company around $350 to $400 million. Autotalks creates chips and vehicle-to-everything (V2X) communication technologies dedicated towards boosting road safety for both ordinary and driverless vehicles. In its announcement, Qualcomm said that Autotalks' "production-ready, dual mode, standalone safety solutions" will be incorporated into the Snapdragon Digital Chassis, its set of cloud-connected assisted and autonomous driving technologies. Nakul Duggal, senior VP of automotive for Qualcomm Technologies, Inc., said in a statement: "We have been investing in V2X research, development and deployment since 2017 and believe that as the automotive market matures, a standalone V2X safety architecture will be needed for enhanced road user safety, as well as smart transportation system... We share Autotalks' decades-long experience and commitment to build V2X technologies and products with a focus on solving real-world road user safety challenges. We look forward to working together to deliver global V2X solutions that will help accelerate time-to-market and enable mass market adoption of this very important safety technology."

Wow." Phone calls with law professors about regulatory actions don't normally start with unprompted expressions of amazement, but regulatory actions don't normally come like this. Anne Witt, professor of law and member of the EDHEC Augmented Law Institute, had been expecting to have a very different conversation when we spoke last Wednesday. But then, just minutes before we were due to talk, the UK's competition regulator blocked Microsoft's attempted $68.7bn acquisition of megadeveloper Activision Blizzard, the sprawling corporation behind games including Candy Crush Saga, World of Warcraft, Tony Hawk's Pro Skater and, most importantly, Call of Duty. Britain's Competition and Markets Authority (CMA) is just one of a number of international regulators which was investigating the proposed acquisition. In the US, the Federal Trade Commision (FTC) had already sued to block the takeover in December, with the case due in court later this year. The European Union is investigating, and has given itself a deadline of 22 May to make a decision, while Australia has paused its own investigation while it engages with overseas regulators. One of those regulators had already given the deal a pass. In March, the Japan Fair Trade Commission ruled that it was "unlikely to result in substantially restraining competition", and approved it to go ahead. Japan's justification for allowing the merger was also behind Witt's expectation it would be approved. "For 30 years or so, competition agencies, very much influenced by the US school, have taken the view that'vertical mergers' are rarely dangerous," she explained, once the shock had worn off. "If you have a'horizontal merger' – if Microsoft had bought up a competitor – it is very evident that that will have a direct effect on competition, because it eliminates one player in the market.

Microsoft Corp beat Wall Street's quarterly revenue and profit estimates on Tuesday, driven by growth in its cloud computing and Office productivity software businesses, and the company said artificial intelligence products were stimulating sales. The company forecast that revenue in its main segments for the current quarter would match or top Wall Street targets. Shares gained 8.3% in after-market trading following a report by the Redmond, Washington-based technology company that profits were $2.45 a share in the fiscal third quarter, beating Wall Street estimates of $2.23, according to data from Refinitiv and up 10% from the same quarter last year. In regular trading, fears about earnings had sent Microsoft down 2.2%, making it the biggest drag on the S&P 500 on Tuesday ahead of its report. Revenue rose 7% to $52.9bn in the quarter ended March, inching past the average analyst estimate of $51.02bn, according to Refinitiv.