Sentim8: Moving Beyond the ‘What’ to Understand the ‘Why’ in Investment Commentary
In the evolving landscape of asset management, the ability to quickly generate accurate and insightful investment commentary has become increasingly important. While traditional natural language technologies have made significant strides in automating investment performance reporting, a crucial piece has been missing: the explanation of why these performance or market patterns occur. This is where Sentim8 breaks new ground.
The Current State of Investment Commentary Technology
Traditional language technologies have served the investment industry well in specific capacities:
- Natural Language Generation (NLG) converts data into readable narrative
- Natural Language Understanding (NLU) helps systems comprehend human inputs
- Natural Language Query (NLQ) enables data retrieval through conversational interfaces
- Natural Language Processing (NLP) powers the basic analysis of text and data
These technologies excel at producing commentary that tells you *what* happened: “Fund X outperformed its benchmark by 2.3% in Q3 2023,” or “Small-cap equities showed strong performance in Asian markets.” There are combinations of business rules and proprietary algorithms, patents (in some cases) and software licenses.
The Sentim8 Difference: Understanding the ‘Why’
Sentim8 builds upon these foundational technologies but takes a quantum leap forward by incorporating:
1. Advanced Large Language Models (LLMs)
We leverage state-of-the-art LLMs, carefully fine-tuned for financial analysis, to process vast amounts of market data and identify complex patterns and relationships that traditional NLP might miss.
2. Explainable AI Generation
Unlike black-box solutions, Sentim8 provides transparent reasoning for its analysis. Every insight is backed by clear evidence and logical progression, allowing investment professionals to validate and trust the output.
3. Retrieval-Augmented Generation (RAG)
Our RAG implementation ensures that every piece of analysis is grounded in verified, current market data. This means commentary isn’t just generated from training data but is continuously enhanced with real-time market information.
4. Sophisticated Prompt Engineering
We’ve developed proprietary prompt engineering techniques that guide our AI to focus on relevant market factors and maintain consistency with established financial analysis frameworks.
5. Add-in Sentiment Analysis techniques
This allows the ability to include contextual elements of the structured data and the unstructured data to couple both together with algorithms to create a synergy with something very positive, positive, neutral, negative or very negative. And check whether the story is viable (using correlation).
6. System Developer toolkits
The industry no longer must learn proprietary software code or another software platform. Opensource tools like Python can have far greater capabilities in the hands of Developers and business users (using an LLM, for example)
Real-World Application: A Hybrid Approach to Truth
What truly sets Sentim8 apart is its flexible approach to sourcing truth. We understand that different organizations have unique needs and trusted sources. Our platform:
1. Integrates Multiple Sources of Truth
- Connects with established financial data providers like Bloomberg, LSEG, B
- Incorporates respected research or market analysis from firms like Deloitte, JPMorgan Insights
- ‘Internal Documentation’: Seamlessly blends your proprietary data. For example, the explanation(s) for strategies, trade rationale or includes elements of internal research (from an email directory output)
2. Ensures Data Security
- Maintains strict data segregation
- Provides flexible deployment options to meet compliance requirements
- Agnostic of the choice of the leading common secure-private cloud hosting platforms
- Encryption of data elements in both directions. RESTful API
- Cites sources and input parameters.
- Data never leaves your environment
From Data to Insight: The Sentim8 Process
Here’s how Sentim8 transforms raw data into meaningful commentary:
1. Data Ingestion and Verification
- Collects performance data from client systems/sources. Drag and drop or batch process. JSON file format conventions
- Validates against trusted external sources
- Ensures data consistency and accuracy (same source, source of the truth)
2. Context Analysis
- Examines broader market conditions
- Identifies relevant economic indicators
- Considers instrument/segment/asset / sector-specific trends
3. Sentiment Assessment
- Analyzes market sentiment across multiple sources
- Evaluates analyst opinions and market reactions
- Considers historical sentiment patterns
- Adds a score of 1-5 to allow the user to create the story linking the structured data and the unstructured data
4. Insight Generation
- Synthesizes data points into a coherent narrative
- Explains causative factors behind investment performance
A Representative Result: Comprehensive Investment Commentary
For an equity portfolio, instead of simply stating that “Technology sector holdings contributed 1.5% to overall performance,” Sentim8 might explain:
“Technology sector holdings contributed 1.5% to overall performance, driven by strong semiconductor demand in Asia. This growth was supported by three key factors: increasing AI chip requirements, supply chain normalization in Taiwan, and favourable currency movements. Market sentiment remains positive, with 73% of analysts maintaining buy ratings, though some concerns exist about potential regulatory headwinds in Q4.”
For a fixed-income corporate bond portfolio, instead of simply stating that “Technology sector holdings contributed 1.5% to overall performance,” Sentim8 might explain:
“Technology sector holdings contributed 1.5% to overall performance, driven by strong semiconductor demand in Asia. This growth was supported by three key factors: increasing AI chip requirements, supply chain normalization in Taiwan, and favourable currency movements. Market sentiment remains positive, with 73% of analysts maintaining buy ratings, though some concerns exist about potential regulatory headwinds in Q4.”
What about an overview of the economic and market conditions? Sentim8 would offer:
UK Market Analysis Q2 2024
—————————————-
The UK fixed-income market is currently navigating a complex landscape shaped by a cautious monetary policy stance from the Bank of England (BoE), inflationary pressures, and geopolitical uncertainties. The BoE’s cautious approach to easing monetary policy is evident in its series of gradual rate cuts, with the base rate expected to reach 3.5% by the end of 2025. This contrasts with the more aggressive easing measures adopted by the Federal Reserve and the European Central Bank. The BoE’s cautious stance is driven by a projected rise in inflation to 3% in early 2025, fueled by higher energy prices and a persistent high level of services inflation. This inflationary pressure is causing real yields to remain low, which could potentially discourage investment in fixed-income instruments. In the government bond market, the yield trends are influenced by the BoE’s monetary policy and inflation expectations. The yield curve, while currently upward-sloping, could potentially flatten if the BoE continues to cut rates. This could impact the returns on longer-dated bonds. Furthermore, the UK government’s borrowing and issuance activity could increase due to the growing host of spending priorities competing for limited fiscal space, as highlighted in the Autumn Budget.
The future of investment commentary lies not just in reporting numbers but in understanding and explaining market dynamics. Sentim8 represents the next evolution in financial technology, combining traditional Natural Language capabilities with other techniques and advanced AI to deliver deeper and more relevant insights to the consumer(s).
By bridging the gap between raw data and meaningful insight, Sentim8 empowers investment professionals to better understand and communicate market dynamics, leading to more informed investment decisions and being able to provide clearer client communications in the form of process automation. Do more with less.
It is scalable and understandable. Each component is delivered based on its most appropriate use. An LLM can be used for some components, business rules and logic for others, and natural language for others.
Click here for a fuller example of a Fixed Income investment performance commentary