From data to narrative signals
We built Moaripanoraune to answer a narrow question. Can we treat market narratives as data without pretending they remove uncertainty. The answer is partial. We can represent narratives as labels derived from text. We can relate those labels to price and flow behaviour. We can show when those relationships hold and when they break. We cannot turn that into a promise about outcomes, and we do not try. Our pipeline starts with data ingestion. We collect curated text, price, and flow inputs and align them on a common time axis. We then run natural language models to identify themes, entities, and sentiment like features, but we do not stop there. We cluster related text into narrative candidates, assign working labels, and expose them to analysts for review. In parallel, we run behavioural models that describe price and flow regimes without reference to the text. The interesting part is the mapping layer. Here we look at how often certain narrative labels appear before, during, or after particular behavioural regimes. We test different windows, asset sets, and thresholds. Sometimes we see patterns that look stable over long stretches. Sometimes we see relationships that collapse around specific events. We record both. When a mapping looks fragile or breaks, we flag that explicitly rather than smoothing it away in a new calibration. In practice, this gives teams a shared lens on narratives rather than a single score to follow. Research can dig into text snippets and label histories. Risk can see where narrative based views align or conflict with existing frameworks. Compliance can review documentation, limits, and disclaimers. Throughout, we repeat the same points. Results may vary. Past performance does not guarantee future results. Use these tools as one input among many in your own decision process.
How Moaripanoraune approaches AI for market narratives
This page explains how we think about AI for financial market research in concrete terms. We start from our own frustration with vague explanations of market moves, then describe how we use models to classify and track market narratives across text, price, and flow data without pretending that any signal is definitive. We focus on traceable workflows, explicit limits, and patterns that analysts, risk teams, and compliance can all interrogate before they influence decisions.
-
Aggregate curated text, price, and flow data into a consistent, timestamp aligned pipeline.
-
Apply narrative classification models to text and expose editable, human readable labels.
-
Relate narrative labels to price and flow regimes under different historical windows.
-
Highlight potential narrative shifts and regime breaks for analyst review, not auto action.
-
Log overrides, comments, and decisions so governance teams can review the full context.
-
Regularly recalibrate models and mappings as language, liquidity, or structure evolve.
Transparent narrative detection approach
We design every feature so a sceptical analyst can ask where a signal came from, see the underlying text, price, and flow inputs, and reconstruct the path from raw data to narrative label without relying on hidden steps or unexplained scores.
Iterative, documented model evolution
We assume our first model specification is wrong or incomplete, so we log experiments, compare alternatives, and keep a written history of why we promote or retire particular narrative features over time.
Separation of analysis and decision making
We treat narrative outputs as inputs to your research, risk, and compliance processes, not as instructions, keeping a clear boundary between analytical pattern detection and any decision your team may take.
Governed, cautious use of AI signals
We operate from Ireland with attention to data protection, audit trails, and clear disclaimers that results may vary and that past performance does not guarantee future results for any use of our tools.
Scope, limits, and next steps
If you have read this far, you probably want specifics on scope, not marketing lines.
This section states the boundaries in clear terms.
Moaripanoraune focuses on AI assisted analysis of financial market narratives using text, price, and flow data. We do not manage money, execute trades, or act as an intermediary. We do not provide personalised recommendations, and nothing on this site or in our tools should be taken as tailored advice. Our outputs are analytical artefacts intended for teams who already operate under established governance, with their own professional advisers.
We make no promise that any particular configuration, dataset, or workflow will
produce a given outcome. Results may vary based on factors such as data quality,
coverage, configuration choices, and how your team interprets and uses the outputs.
Past performance does not guarantee future results, even when a relationship appears
stable in historical tests. You should treat all examples and scenarios as
illustrative only.
We operate from Ireland and handle data with attention to applicable regulation and
client standards. We maintain audit trails, logging, and documentation so that
internal and external reviewers can understand how models were configured and how
narrative labels evolved. If you need more detail about privacy, data protection, or
legal terms, review our dedicated pages on those topics or contact us with specific
questions.
Who typically uses Moaripanoraune and how
You probably want to know where our tools fit, where they do not, and how they interact with
the structures you already have. This section summarises typical use patterns without
pretending they are universal or prescriptive.
Research teams
Research teams use Moaripanoraune to add structure to qualitative narratives they already track. They feed in approved text sources, review model proposed labels, and compare narrative shifts with price and flow regimes around key events. The goal is not to replace judgement. It is to see whether the stories discussed in meetings have consistent data behind them, and where that support is weak or missing.
Risk and governance
Risk and governance teams focus on how narrative driven views interact with existing frameworks. They look at audit trails, label histories, and documented breaks in relationships between narratives and behaviour. This helps them question whether a view rests on repeatable patterns or on recent anecdotes. It also supports reviews that must explain, after the fact, why a particular narrative was treated as credible at a given time.
Compliance and oversight
Compliance teams care about boundaries. They review how data is sourced, how outputs are framed, and how limitations are communicated. We make it clear that results may vary and that past performance does not guarantee future results. We do not provide personalised advice, and we do not handle client funds or execute trades. Our role is to supply structured analytical context that sits inside your existing rules, not to define those rules.
What this page covers
Technical overview and practical boundaries
This page is for when you want the mechanics, not the headline. We use AI models
to classify and predict shifts in market narratives by combining three types of
data. Text from curated sources such as official remarks, structured news, and
key transcripts. Price series across selected instruments. Flow style data where
available, such as volume, depth, or positioning proxies. All of it is timestamp
aligned so we can ask concrete questions about what changed, when, and around
which events. The core models focus on narrative detection in text. They group
documents into clusters, assign labels, and score how strongly each piece of
text expresses a theme. Those labels then sit next to price and flow regimes,
which we derive from separate models that look only at behaviour. We are not
trying to compress everything into a single number. We are trying to see how
text based narratives and observable behaviour relate, where they line up, and
where they do not. Analysts remain in control. They can edit labels, exclude
sources, adjust windows, and comment on odd results. Every change is logged.
Over time, that log becomes a record of how your internal view of key narratives
evolved. Results may vary, and past performance does not guarantee future
results. Our tools do not handle client funds, execute trades, or provide
personalised advice. They provide structured context and pattern documentation
for teams who already run their own research and governance processes.
How the workflow fits together
Our approach looks straightforward on a slide. In reality, each step hides trade offs. This section breaks those steps down so you can see where judgement enters, where models dominate, and where we deliberately leave room for disagreement inside your team.
Data selection and alignment
We begin with your approved sources. Official remarks, curated news, research notes, and other text that already fits your governance standards. We add price and flow style data for instruments you care about. Everything is timestamped, normalised where appropriate, and stored with enough metadata that you can trace a narrative label back to a specific item and context, not just a generic category.
Narrative label design
Text models propose narrative clusters and working labels. Analysts then review these proposals, merge or split clusters, and refine names so they match internal language. This step matters because it anchors the system in the way your team actually talks about markets. The models provide structure. Humans decide which structures are meaningful and which are artefacts of the algorithm or the data sample.
Relating narratives to behaviour
Separately, we classify price and flow regimes. These models know nothing about the text. They only see behaviour. We then test how often certain narrative labels appear around particular regimes under different horizons. Some relationships will look strong and persistent. Others will be weak, noisy, or obviously unstable. We document both so you can see where your favourite stories have data behind them and where they do not.
Outputs, limits, and governance
Finally, we package outputs in a form that fits your processes. That might mean dashboards, exports, or periodic reports that highlight where narrative signals changed, where mappings weakened, and where models are uncertain. We make clear that these are analytical artefacts, not instructions. Results may vary, and past performance does not guarantee future results. Your governance processes stay in charge.
Logging and audit history
Across all steps we log choices. Data source changes, model updates, label edits, and threshold tweaks all leave a trail. This audit history lets you answer questions like why a signal disappeared, when a definition changed, or how a particular narrative came to dominate internal discussions. That record is often more useful for governance and onboarding than any static documentation alone.