Key Takeaways
- Perpetuals.com, in partnership with Younet AI, has launched Forgentiq.ai, an on-premises AI platform built for hedge funds, proprietary trading firms and digital asset managers.
- The platform addresses data sovereignty by letting firms run large language models (LLMs) on their own infrastructure — keeping proprietary trading data out of third-party cloud environments.
- The agentic AI infrastructure targets alpha generation through quantitative research, real-time market analysis and trading strategy development, processing both structured and unstructured data on-site. Perpetuals.com is betting that hedge funds won’t send their most sensitive trading data to the cloud — and it has built an on-premises AI platform around that conviction. Forgentiq.ai, launched in partnership with Younet AI, gives quant firms a way to run purpose-built LLMs against their own proprietary data without it ever leaving their infrastructure. That’s a meaningful distinction in a sector where execution records, position data and research signals are the IP.
Perpetuals.com Addresses Data Sovereignty With Forgentiq.ai Launch
Perpetuals.com Ltd. (NASDAQ: PDC) announced Forgentiq.ai as a purpose-built AI infrastructure platform for hedge funds, proprietary trading firms and digital asset managers. Built in collaboration with Younet AI — which licensed the underlying Forgentiq.ai technology — the platform provides secure, on-premises LLMs for quantitative research, market analysis and trading strategy development. Perpetuals.com plans to deploy it internally against its own market microstructure data and trading strategy IP before rolling it out to external clients on a subscription basis.
The Imperative of Data Sovereignty in AI-Driven Trading
Hedge funds handle some of the most sensitive data in finance — execution records, position data, research signals, portfolio models. That data is core IP, and sending it to a centralised cloud AI provider creates real security and regulatory exposure. Forgentiq.ai operates directly within a client’s own systems, so firms get advanced AI capabilities without surrendering control of their most valuable assets.
The platform also takes aim at the “black box” problem common in complex AI systems, where the path from data input to trading decision is opaque even to the people running it. Regulators including the Bank for International Settlements have flagged the limited explainability of complex AI models as a systemic concern, particularly in high-stakes financial applications. By keeping models on-premises and within the firm’s governance perimeter, Forgentiq.ai gives compliance teams more visibility into model behaviour — potentially easing obligations under MiFID II, MiCA, DORA and EMIR, which Perpetuals.com already adheres to for its Kronos X platform.
AI’s Multi-Faceted Edge in Alpha Generation and Risk Management
Hedge funds are deploying AI across three main areas to generate alpha and sharpen risk management. Alpha — an investment strategy’s excess return relative to a benchmark — is the whole game for active managers, and AI’s ability to process large data volumes at speed is increasingly central to competing for it.
The first area is predictive analytics and algorithmic trading. AI algorithms can execute trades at optimal times, analyse historical data trends and incorporate real-time market signals to improve execution and reduce costs. They can also automate backtesting of trading strategies against historical data, letting managers refine their approaches with greater precision.
The second is alternative data combined with natural language processing (NLP). The vast majority of new enterprise data is unstructured — research reports, earnings transcripts, regulatory filings, news and social media sentiment. Converting that into structured decision signals is a significant challenge. AI-driven NLP tools let hedge funds analyse these data sets in real time, surfacing market sentiment and uncovering signals that may not yet be priced in. The global alternative data market is expected to grow significantly through the end of the decade, according to industry analysts.
The third is risk management. Machine learning models can identify volatility patterns in historical data and help funds adjust strategies in real time. AI-powered stress testing and scenario analysis give portfolio managers deeper insight into how positions might behave under different market conditions — flagging rising correlations or emerging risks faster than manual processes allow.
Navigating the Risks: Explainability, Overfitting and Bias
The upside is real, but so are the risks. The most pressing is limited explainability. Deep learning models and LLMs can produce outputs that even their developers struggle to interpret, and regulators require transparency into how AI models reach their conclusions. Without it, model risk increases and compliance becomes harder to demonstrate. Explainable AI (XAI) frameworks aim to surface the logic behind model outputs — accuracy, fairness, transparency and decision pathways — but this remains an active engineering challenge rather than a solved one.
Overfitting is another significant risk. AI models, particularly deep learning systems, can learn the noise in historical data rather than genuine patterns. The result: strong backtest performance that falls apart in live markets. This optimisation bias — where a model is tuned so tightly to past data that it can’t adapt to new conditions — is mitigated through techniques like cross-validation, out-of-sample testing and keeping model complexity in check. If you’re building agentic trading workflows, the hidden costs of AI automation pipelines are worth auditing before you scale.
Algorithmic bias is a third area of concern. Bias can enter AI systems through unrepresentative training data or assumptions baked in during model development. Models trained on historical creditworthiness data, for example, can reinforce discriminatory lending patterns — locking out underrepresented groups or creating feedback loops that distort market behaviour. Robust data quality checks, diverse datasets and continuous monitoring are not optional extras here; they’re operational requirements.
The Future of Quant: Human-AI Collaboration and Adaptability
AI adoption in hedge fund management is moving past the experimental phase into something more structured — focused on compliant, explainable and reliable deployment. The firms getting real traction aren’t just stacking datasets and tools; they’re building centralised data foundations, compliant workflows and governance frameworks that let AI augment human judgment rather than operate around it.
The convergence of traditional quantitative methods with modern AI techniques — think LangChain-powered research agents or LlamaIndex pipelines over proprietary data stores — is producing more adaptive trading strategies. But it demands continuous iteration as market conditions shift. The likely end state is hybrid: AI provides speed and pattern recognition at scale, while human expertise drives strategic direction and keeps risk exposure in check. For firms navigating how to select the right AI infrastructure for this kind of work, our guide to choosing a generative AI provider in 2026 is a useful starting point. Firms that invest in data quality, advanced analytics and strong governance now will be better positioned as competition intensifies. For more on AI agents and automation tools, visit our AI Agents section.
Originally published at https://autonainews.com/perpetualscom-launches-forgentiqai-on-premise-llms-for-hedge-fund-alpha/
Top comments (0)