Evidence-backed thematic investment signals
DataHex Thematic Studio
Turn an investment theme into an evidence-backed, investable signal. Thematic Studio leverages AI to score a market’s companies against a theme you define, grounds every score in verbatim quotes from primary-source company filings, and carries it through construction, backtesting, deployment, drift monitoring, due diligence and deep-dive research in one self-service workflow.
The opportunity
The thematic classification challenges investment firms face
Thematic and custom indexing is among the fastest-growing segments in capital markets. Yet the work underpinning it, deciding which companies are genuinely exposed to a theme, is costly and inefficient. It invites subjective bias, is hard to scale across thousands of names, and hard to defend to investors and regulators. Thematic Studio closes that gap.
Classification is subjective
Deciding which companies are exposed to a theme is manual analyst work, inconsistent between people, with no record of why a name was included.
Exposure proxies are crude
Sector schemes don’t map to themes, and revenue-segment "purity" misses indirect and strategic exposure and relies on patchy disclosure.
Credibility is under scrutiny
Investors and regulators increasingly challenge whether constituents are genuinely exposed. A basket that can’t show why each name belongs is a reputational risk.
Construction is slow & costly
Research, methodology, backtesting and rebalancing are bespoke and resource-heavy, constraining how many themes a team can take to market.
The result: fewer themes shipped, baskets that are hard to defend, and thematic products that struggle to earn investor trust.
Outcomes
What DataHex Thematic Studio gives you
Trust
Defensible inclusion
Every AI-generated score is backed by verbatim source evidence, captured for audit.
Depth
Beyond keywords
Four exposure dimensions that revenue tags and sector codes cannot capture.
Speed
Theme to signal
A scored universe in hours of compute, not weeks of analyst reading.
Coverage
Any market, any language
Model-agnostic, multilingual AI reading source data across markets.
Control
Cost and governance
Pre-estimated run costs, point-in-time method, role-based access.
How it works
From a plain-language thesis to a monitored signal, in six steps
One AI-assisted, self-service workflow takes a theme from idea to deployed, monitored signal. The platform reads and indexes the corpus you select for the run, and each stage hands its output to the next.
Illustrative use case
Testing AI and robotics signals across global markets. An analyst tests investment signals for exposure to artificial intelligence and robotics across global markets. A one-line thesis works, and so does a multi-clause brief spanning sub-sectors, value-chain tiers, and exclusions.
Define the theme, in plain language
Build the thesis yourself, criterion by criterion, or let the AI assistant draft it from your description and any research you upload. The thesis becomes a structured, versioned rubric: theme name and definition, in-scope and out-of-scope boundaries, per-dimension evidence criteria, sub-themes, and weights. A guided, bounded workflow, not an open-ended chatbot. You review, refine, and approve before anything is scored.
Theme name, structured investment prompt, and scope, reviewed and approved before any company is scored.
Score the universe on four dimensions
Score an entire market or a custom watchlist, from the top 10 to the top 30 or more. For each company the engine retrieves the most relevant passages from its filings using hybrid semantic and keyword search with reranking, then scores exposure on a minus 3 to plus 3 scale across four fixed dimensions: direct business exposure, strategic alignment, indirect exposure, and market perception. A deterministic, weighted rubric produces the final score. The model never invents the number.
Composite score and rank per company, with the volume of evidence behind each score.
Ground every score in evidence
Every dimension carries its evidence quotes, its reasoning, and a confidence level. Each quote is checked against the source filing, and a quote that cannot be verified reduces the score's confidence rather than passing silently. Scores link back to their source documents and the exact passages cited, so a portfolio manager can ask why a company scores as it does and see the answer.
Left, a validated quote carries the score. Right, the same card when a quote cannot be verified: the validation mark is absent and confidence drops, rather than the quote passing silently.
Construct and backtest the index
Rank by exposure, apply minimum-score and top-N rules, and set the constituents and weights you choose. Then take the signal back through history, point-in-time, with dividend-adjusted and split-adjusted returns and the look-ahead controls that keep a backtest honest, so a theme is validated before a cent is committed.
Point-in-time backtest of the AI and Robotics thematic signal against a market benchmark.
Illustrative example only. Backtested and past performance are not indicative of future results. Nothing on this page is investment advice.
Deploy the signal
Approved signals run in production on a schedule and are delivered through a clean, point-in-time API, ready to consume as an index or as an alpha signal. Agent-ready: built for the way investment teams and their AI agents work, grounding outputs in governed data and primary-source evidence rather than generic assumptions.
Monitor drift
Companies pivot and exposure changes after inclusion. Deployed signals are monitored continuously, with portfolio drift tracked against your thresholds and position-concentration, sector-concentration, and compliance checks applied. Choose how to act, and every deployment, rebalance, and alert is recorded on an activity timeline.
Auto-Rebalance
Proposes a rebalance for your approval when drift crosses a threshold.
Notify-Only
Alerts you when drift crosses a threshold, so you act manually.
Every constructed and deployed index lands in the Thematic Data Catalog, where teams track performance, drift, and constituents in one place, and request access, export constituents and scores, or integrate through the API.
Who it's for
Built for the teams behind thematic products
Exchange index teams
Launch proprietary thematic indices with methodology that stands up to scrutiny.
Analysts and researchers
Cover more ground and go deeper, with evidence-linked scores and citations on every name.
Portfolio managers and quants
Trade thematic exposure as a reproducible signal, delivered through a clean API.
Ready to see DataHex Thematic Studio in action?
We will walk your team through a theme end to end, from thesis to scored universe to backtest, using a theme that matters to you.