Start with the reported information.
Public company filings provide the financial inputs. The source material is the starting point for the research workflow.
Source recordsMOAT / AN ANTEOMNIO IN-HOUSE PRODUCTYears of filings.
Years of filings.
One coherent
research view.
A platform for researching Indian stocks and mutual funds, built around historical information and long-term analysis.
The challenge was to connect what companies report with what a researcher needs to examine: a consistent history, a modelled estimate and the context to interpret it.
Explore the engineering ↓- PROJECT TYPE
- In-house software product
- DOMAIN
- Investment research
- ENGINEERING FOCUS
- Data, calculation and presentation
01 / THE PROBLEMThe difficult work
The difficult work
starts before the comparison.
A useful research view needs more than a collection of numbers.
Historical filings arrive in different structures. Labels vary between companies and reporting periods. Before a model can work through years of earnings, those records need a consistent meaning.
Repeating that preparation in separate spreadsheets makes it harder to maintain the same method as new information arrives. MOAT brings the preparation, calculation and presentation into an application workflow.
The product is designed for examining a business over time. That purpose shaped an interface centred on historical research rather than a constantly moving intraday display.
02 / THE SYSTEMFour stages.
Four stages.
A distinct job for each.
Separating the stages makes the system easier to reason about. An unexpected result raises a specific question: did the source change, was a field interpreted incorrectly, or did the model produce it?
Give the model a consistent structure.
Line items and reporting periods need a common representation before calculations can compare the history.
Structured historyApply the valuation method.
The model uses historical earnings to calculate an estimate. The documented approach works through a decade of earnings.
Calculated resultMake the comparison understandable.
Present estimated value in relation to market price, with the distinction between reported information and model output kept clear.
Research view03 / THE FIRST ENGINEERING CHALLENGEDifferent labels.
Different labels.
Comparable meaning.
A calculation cannot resolve a mistake in what an input represents. Normalisation is therefore a substantial part of the application, not a formatting step added at the end.
The system needs a consistent way to represent financial line items across companies and years. The important distinction is between changing a label and establishing that two records mean the same thing.
This is a recurring custom-software problem: a clean interface depends on the decisions made while preparing the information beneath it.
04 / THE SECOND ENGINEERING CHALLENGECalculate when
Calculate when
the information changes.
Working through a decade of figures on every page request would repeat the same calculation for each visitor. The archive describes a different approach: compute results when a filing arrives and cache them for subsequent views.
That separates the cost of preparing a result from the act of reading it. The research page can retrieve an available calculation instead of making the visitor wait for the full model to run.
The design still depends on updating calculated results when their inputs change. Caching is a responsibility to manage, not evidence that a result is always current.
05 / THE PRODUCT EXPERIENCELet the interface
Let the interface
serve the research.
MOAT brings stocks and mutual funds into the same product. This case study focuses on the company-filings and valuation workflow; it does not imply that a fund and an individual business are modelled identically.
Make the comparison legible.
The research question is the relationship between a modelled estimate and market price. The page needs to make those different kinds of information understandable.
Support considered review.
The documented interface omits live ticking prices and intraday charts. That is a product choice aligned with the intended research workflow.
Keep the model in perspective.
A calculated estimate is an output of a method. It should be read with its assumptions and source information, rather than treated as a fact established by the interface.
WHAT THIS CASE STUDY ESTABLISHESA view into
A view into
the engineering.
This is Anteomnio's own product, described through its purpose and documented architecture. We have not published usage figures, independently validated model accuracy, investment returns or a measured performance benchmark here.
The transferable work is the system design: bring inconsistent inputs into a common structure, separate calculation from presentation and make a complex result easier to examine.
Explore OmnioPHP, the framework used for MOAT ↗What does your team
keep rebuilding in a spreadsheet?
If the same information needs collecting, reconciling and calculating every week, tell us how that work happens today. We can help assess whether it belongs in a dedicated application.