Artificial Intelligence in Trading and Portfolio Management
Opportunities and Risks
-
July 31, 2025
DownloadsDownload Article
-
AI implementation in the financial industry has progressed from experimentation to the successful deployment of high-value, strategic use cases. More firms are experimenting with the technology and applications are becoming more mature. Whereas larger organizations were once slow or reluctant to use generative AI — with applications restricted to smaller hedge funds or fintech companies — bigger firms are now showing a greater willingness to adopt AI for trading and portfolio management.
According to NVIDIA’s 2025 State of AI in Financial Services1 survey report, investment in AI infrastructure has increased significantly compared to the previous year. The survey also indicates signs of maturity in AI adoption and a decline in implementation challenges. More than half of the surveyed companies identified AI as critical to their success.
Whether your firm is a major financial institution, mid-sized bank, trading firm or a corporation making trading decisions, now is the time to start exploring ways to adopt AI to manage your portfolio and trading. After all, the earlier you start, the earlier you can identify relevant risks and opportunities. Below is an overview of how some firms in the finance and trading spaces are using AI technology — and risks to watch out for.
Opportunities
With ongoing market pressure to turn profits amid unprecedented macroeconomic and geopolitical uncertainty,2 many firms are seeking to do things faster and leaner. AI has the potential to not only transform a financial organization’s operations, but also to enhance its research capabilities, enabling faster and deeper analysis of large volumes of data. It allows analysts to cover a broader investment universe with deeper insights.3
But AI doesn’t only represent advantages in terms of greater efficiency. Firms can create multi-agent systems and automate workflows with agents tailored to their financial functions. AI models can be fine-tuned to better reason according to identified market dynamics and they can be integrated with a firm’s proprietary models, data and simulators to boost its competitive advantage. Sophisticated reasoning capabilities of generative AI combined with agentic frameworks can streamline the development, deployment and back-testing of investment strategies, thereby accelerating the research lifecycle and enabling investors to make informed, data-driven decisions in real time.
While generative-AI-driven trading and portfolio management is still in its early stages, it holds significant promise. Below is a list of ways that AI is already being used by firms:
- Data Augmentation: AI allows firms to integrate and synthesize large-scale multi-modal datasets to support existing analytical frameworks or human discretion (e.g., processing earning calls, filings, news and satellite images to extract features for quantitative models, or to generate insights for company and industry research).
- Latent Alpha Generation: Emergent trends, hidden correlations and causations can be identified, and the effects of market events in large volumes of unstructured data can be detected (e.g., tracing how supply chain disruptions or commodity price changes propagate across different sectors).
- Risk Management: Market shifts and potential risk events can be identified and volatility can be forecast using real-time data monitoring, anomaly detection and analysis of complex nonlinear patterns for predictive analytics. This approach enables the early identification of emerging risks.4
- “Quantamental” Investing: Quantitative and qualitative strategies can be integrated by systematically analyzing large datasets to identify and quantify contextual narratives.5 This strategy aims to offer qualitative interpretations and explanations for market movements or statistical patterns to improve returns.
- Thematic & Custom Portfolio Construction: Emerging trends can be identified and portfolios and thematic indices can be systematically constructed based on key themes such as geopolitical shifts or technological innovations. Further, portfolios can be rapidly tailored based on clients’ historical investment patterns and real-time market conditions.
- Synthetic Data Generation: Synthetic datasets can be generated when real datasets are unavailable, sparse or confidential. This data could be used for model training and help reinforce stress testing and back-testing across a broader range of market scenarios.
Risks
Quantifying the risks of using AI in trading is challenging because the technology is evolving rapidly and regulatory standards are developing differently across jurisdictions.6 However, there are areas that any financial organization that deploys the technology should monitor. Key risks include:
- Market Instability: Adoption of AI in trading can result in collective actions and higher concentration risk, thereby amplifying market shocks7 — accordingly, the Bank of England plans to8 closely monitor hedge fund and bank AI usage.
- Systematic Risk: Due to a reliance on a small number of AI service providers, firms could be vulnerable to making similar high-risk or suboptimal trading decisions.
- Latency: Some generative AI systems may be unsuitable for high frequency trading due to their inference latency relative to algorithmic trading.
As noted above, laws, regulations and industry standards around AI usage are in flux, and will likely continue to be, due to the rapidly advancing nature of the technology. So, in addition to monitoring key risks associated with automated portfolio management and trading, firms will need to stay abreast of evolving best practices in AI governance and ethics.
Conclusion
For some time now, AI has stoked the finance world’s imagination, inspiring organizations to consider how new applications might transform their operations and uncover new opportunities. But the past year has seen accelerating adoption, with more major companies in the trading and portfolio management space beginning to explore use cases.
The sooner a firm starts experimenting with AI and becomes familiar with its opportunities and risks, the better positioned it will be to leverage the cutting-edge technology. A trusted partner can help an organization implement and iterate on AI deployments — particularly in a space like finance, where competition is fierce and so much is at stake.
Footnotes:
1: NVIDIA, “State of AI in Financial Services,” NVIDIA (February 2025)
2: FTI Consulting, “Managing Geopolitical Risk & Change,” FTI Consulting (May 2025)
3: Andrew Chin, “Leveling the Divide Between Discretionary and Systematic Investing: How AI Enables Breadth and Depth,” The Journal of Portfolio Management (June 4, 2025)
4: Hamlin Lovell, “Future-Proofing Risk and Portfolio Management with AI and Gen AI,” HedgeNordic (December 2, 2024)
5: Alejandro Lopez-Lira et. al., “Thematic Scoring: Quantifying Contextual Narratives using Language Models,” Social Science Research Network (April 15, 2025)
6: Michael Horoho, Joe Knight and Kyle Wetzold, “AI Compliance in an Uncertain Regulatory Landscape: What General Counsel Need to Know,” FTI Consulting (March 20, 2025)
7: Bank of England, “Financial Stability in Focus: Artificial intelligence in the financial system,” Bank of England (April 9, 2025)
8: Meg Short and Philip Aldrick, “BOE Warns Deviant AI on Trading Floors Risks Triggering a Crash,” Bloomberg (April 9, 2025)
Related Information
Published
July 31, 2025
Key Contacts
Managing Director
Senior Director
Senior Managing Director, Leader of Securities, Commodities & Derivatives
Managing Director
Senior Managing Director, Leader of AI & Digital Transformation