Updates

This month brought a feature we have wanted to ship for a long time. The financials frame now includes a beta version of data point linking, where each value in the income statement, balance sheet, or cash flow statement can be traced back to its exact location inside the original filing. This gives full transparency on where every number comes from and lets analysts audit the data without ever guessing. The feature is powered by a new Filings section we built in parallel, which allows users to skim through company filings directly inside the terminal without having to open a browser or jump to an external SEC viewer. The two shipped together because the linking infrastructure relies on the same parsing and rendering layer that drives the Filings section.

The risk factor model also received a significant upgrade. After a process of fine tuning a language model specifically for financial summaries, the quality of the output improved sharply. The model now captures nuances that were previously lost and produces summaries that are clearer, more structured, and more useful for analysts. We are not disclosing the model choice or the exact training steps at this stage, but the results are extremely interesting. The risk axes themselves were also enhanced this month, so that more companies fit cleanly into the framework, covering macro risk, operational risk, financial risk, and other categories.

A new Tools section has been added to the terminal, organized around two pillars: calculators and exports. The calculators cover present value, future value, interest compounding, option pricing, refinancing analysis, and several other standard financial computations, all accessible without leaving the application. On the export side, Excel and PowerPoint templates are now available as pre formatted files ready to be opened and filled in, which removes the friction of starting from a blank page and saves significant time on layout work. The current version requires manual input, but the longer term goal is for the terminal to populate these templates directly with company data, so the export becomes a one click operation.

In parallel, work continues on strengthening the data layer and bringing it to the server with full reliability. The target is to have it available next month. Once that ships, we will be in a position to package and launch the first beta version of AtlasEQ, made available to a selected group of users for early performance and reliability testing.

The portfolio module is now live in beta. Users can import their trades from a CSV file exported from their broker, connect it to the platform and immediately see all their holdings. From there, the module displays performance against a benchmark, giving a clear view of how the portfolio is doing. A holding table lets you review and edit positions directly. This is a first version and further development is needed for deeper portfolio analysis, but the foundation is solid and usable today.

On the valuation side, we went well beyond a simple DCF. The platform now supports three valuation models: a Discounted Cash Flow model, a Residual Income model, and an Economic Profit model. Each model comes with a full set of assumptions that can be edited either as general parameters or on a per-year basis, giving analysts fine-grained control. For the cost of capital, three approaches are available: CAPM, a simplified method, and the Fama-French three-factor model. On top of that, a stress test module lets you define different scenarios and see how they affect the output valuation directly within the application, whether it is a recession, a rate hike, or any other event you want to simulate.

Cash flow statement parsing also got an overhaul. The label matching logic for operating, investing, and financing activities is now significantly more reliable across different filing formats.

This was a big month. We added initial ESEF filing support, bringing European companies into the platform alongside US-listed securities. It's a first implementation and there's still plenty of work ahead to refine and extend it across all functionalities, but it works well through the financials pipeline and flows down to other features. A unified fetcher handles both SEC and ESEF data through the same channel, with rate limiting and proper error handling built in.

The backend moved to PostgreSQL. Cover page data, supplemental filings, debt maturity schedules, and segment revenue breakdowns all come from the database now. We also built a Holdings API to replace what was previously local 13F database access.

On the desktop side, we shipped a native macOS title bar using PyObjC with trapezoid-shaped navigation buttons. It was a pain to get right with Qt, but the result feels much more native. Excel export for financial statements also went live with proper styling and a cover page.

The analytics frame also received an important upgrade this month: data coverage now extends up to ten years, doubling the previous limit and giving analysts a much longer historical window to work with when studying trends and cycles.

We replaced the old Machine Learning frame with a proper Analytics module. The new frame covers profitability analysis, efficiency analysis, and solvency analysis, with dedicated charts for each category and export options to save the results. At this stage, the module supported up to five years of historical data. We also introduced automatic IFRS taxonomy detection, so the app can now handle international accounting standards alongside US GAAP.

The entire terminal got a new theme option: midnight blue. We centralized all color definitions into a theme system, which made it possible to switch themes consistently across every frame and component.

Excel export also got a redesign, and we reworked the fetching flow so the app can handle identical tickers across different markets without confusion.

The Tracker is one of the features we are most proud of. We integrated a wide range of databases at once, connecting directly to FRED, EIA, Census Bureau, NOAA, BLS, Eurostat, and more. The result is access to a large number of indicators across many countries, all centralized in a single access point within the application.

An economic release calendar tracks upcoming data publications so you can stay ahead of GDP, inflation, unemployment, and trade releases from both the US and Europe. The layout is fully customizable, letting each user decide how to organize and arrange the data views to match their workflow.

We also added a cross-currency matrix that displays the latest exchange rates at a glance, along with demographic indicators covering population, education, and health metrics. The whole module was built with a searchable event selector and parameter control bar, so finding and visualizing data is fast and straightforward.

User accounts are now live. We built a full authentication module with session handling, login/logout, and user preferences that sync to the server hosted on OVH, a French provider we chose to keep infrastructure local and secure. Your settings follow you across sessions.

The settings panel gives users control over the default state of the terminal, from chart preferences to display options and other defaults, so the application feels tailored from the moment you open it. Account management, customization, and privacy controls are all accessible from one place.

A major milestone: the app can now run against a remote server. We set up an OVH backend and added a toggle to switch between local and server mode. Company search, data retrieval, and all the heavy lifting can now happen server-side.

On the charting side, Bollinger Bands and MACD indicators shipped with full configuration options. Credit ratings fetching got more reliable through LEI-based lookups instead of relying on ticker matching alone.

The peer comparison module came together this month. Users can now search for and select peers, with an intelligent selection logic that suggests relevant companies. A bottom table shows the current company alongside its peers with synced scrolling. FINRA short interest data and institutional ownership analysis from SEC 13F filings also went live. Performance charts support both total return and price performance views.

A major addition was the chart builder module. It lets users create custom charts directly from the peer statistics data, with more than eight chart types available. You can pick any combination of metrics, change colors, switch axes, and put any data together in the same view. Once built, charts can be exported to Excel in a clean, formatted layout. The whole idea is that the data is already there from the peer analysis, so you should be able to visualize it however you want without leaving the application.

On top of that, we built a lightweight chart library from scratch for the main charting engine. Multiple chart types, technical indicators, hover detection, color schemes, and a proper Y-axis manager. It is written to be efficient, fast to render, and flexible enough to handle whatever we throw at it. Saving and reusing chart configurations through templates makes the whole workflow feel smooth.

The KeyStats calculator became the single source of truth for financial metrics. Shares outstanding, EBITDA, gross profit, margins: everything flows through one calculation layer now.

Insider transaction data went in with centralized fetching and caching. The dashboard gained a new section pulling headlines directly from SEC filings. We also implemented incremental quarterly data fetching, which means the app only requests what it does not already have.

The first working version of the financial statements engine. Income statements, balance sheets, and cash flow statements, all pulled directly from SEC EDGAR via XBRL.

We built a centralized SEC API rate limiter to stay within EDGAR's usage guidelines, implemented proper sign logic for income statement items, and got quarterly row ordering and indentation working correctly. It was rough around the edges but the data pipeline was solid.

This month was all about research, understanding how the SEC works, how EDGAR serves filings, and how XBRL data is structured. The goal was to learn as much as possible about the system before building on top of it.

The first prototype was essentially a live parser of the SEC database. It could pull filings and display financial data, but with serious limitations: only the most recent filing, limited year coverage, and poor handling of the thousands of different XBRL tags companies use to report the same line items. Rough, but it proved the concept.

The Profile frame was the first real feature. Company overview, sector classification, key officers, and a summary of the business. Alongside it, we built the Ratios frame, calculating profitability, liquidity, and leverage metrics from the raw filing data.

Honestly, at this point the app was still mostly a wrapper around API calls. But this was also about learning, getting deeper into Python, understanding how to structure a real application, and figuring out how to turn raw data into something usable. Every piece built here laid groundwork for what came later.

Everything starts somewhere. In January we laid down the core architecture: the navigation system, the dashboard structure, the frame management that lets you switch between different views without losing state.

The goal from day one was to build a professional research tool I could actually use myself, powered by the best data sources available. That meant getting the fundamentals right before adding any features. Window management, data flow, caching strategy: all the invisible stuff that makes or breaks the user experience.