Systematic equity coverage for a concentrated family office, without the institutional headcount

The challenge

A research process that could not scale with the universe it was asked to cover

The office maintained a concentrated public equity portfolio while tracking a North American long list exceeding five hundred names. With one investment lead and a part-time analyst sharing all research responsibilities, systematic coverage across that universe was not achievable with the tools in place. Technical screens for resistance breakouts and double bottoms were run manually across a small fraction of the watchlist each week, while the majority of names went unexamined. Fundamental data from FactSet and Bloomberg was extracted and transferred to spreadsheets by hand. Each new investment case required assembling a fresh document from disparate sources, and as the watchlist expanded, the proportion of names receiving any systematic review declined.

Automation at this scale required a reliable data foundation before anything else could be built. We began by constructing the ticker master, historical price series, and a fundamental cache drawing OHLC data, key metrics, and financial ratios into a single structured source. That data layer became the substrate for every subsequent capability: the pattern scan engine, the fundamental visualization suite, and the vendor export pipeline built downstream.

The screening desk

Nine pattern scans, each with annotated charts and tunable detection parameters

On the data foundation, we built a parallel scan engine operating across S&P 500, TSX, and configurable custom lists. Nine technical patterns run concurrently: clustered resistance breakout, consolidation breakout, cyclical double bottom, cup-and-handle, bull flag, 52-week high breakout, golden cross, ascending triangle, and relative-strength rank against the index. Each confirmed match returns a scored result alongside an annotated price chart: resistance levels and breakout markers for breakout patterns; first low, second low, and neckline annotations for double bottoms; base range and volume confirmation for consolidation setups. All detection parameters are adjustable from the scanner interface, including lookback period, breakout window, secondary low recency, and minimum separation in months between pattern lows.

Technical pattern results are the starting point for analysis, not the end of it. Selecting any match opens the twelve-graph fundamental suite for that name. Valuation charts track enterprise value against market cap and plot EV/EBIT, EV/EBITDA, and P/E alongside price-to-book, price-to-cash-flow, and price-to-sales multiples. Earnings quality is covered through EPS diluted, revenue and net income, EBITDA, and revenue growth rate over time. Return metrics include return on equity, return on total capital, return on assets, and price return against total return. Capital structure charts track diluted shares outstanding, book value per share, tangible book value, and total debt. Every series is shown alongside a rebased S&P 500 price return for relative context. Charts export as high-resolution PNGs for use in investment memos and committee presentations.

The same visualization suite is also available as a standalone research tool, independent of any scan result. An analyst can enter any ticker or comma-separated list of names and receive the complete twelve-graph fundamental report for each, covering any of four time horizons from one to ten years. This makes the charts equally applicable to deep-dive work on names already under consideration as to the evaluation of new pattern matches.

The export desk

Vendor data and investment rationale assembled automatically for each qualifying name

Built downstream of the scanner on the same data foundation, the export layer eliminates the manual assembly work that previously preceded each investment decision. Nightly integrations with FactSet and Bloomberg populate standardized Excel workbooks per ticker and per scan run, drawing in EBITDA, market cap, enterprise value, net debt, leverage ratios, free cash flow yield, and sector-relative multiples. When a name passes both technical and fundamental review, the system assembles a downloadable rationale pack automatically: the annotated pattern chart, the twelve-graph fundamental snapshot exported at presentation resolution, vendor metrics spreadsheets, and an editable thesis scaffold structured around the key investment considerations. The investment lead reviews the assembled materials, edits the thesis, and makes the capital allocation decision. No position is approved without explicit human sign-off.

The export layer is a distinct workflow from the scanner, not a feature within it. Its purpose is to convert a qualified match into a file the investment lead can open, review, and act on without rebuilding the supporting materials from scratch. Previously, that assembly work fell to the analyst before any thesis consideration could begin. With the pack generated automatically, the analyst's time shifts from document construction to investment judgment.

Results

Systematic coverage of a five-hundred-name universe, with fundamental context on every qualifying result

• Full-universe screens across S&P 500 and TSX completed within a single session, replacing a truncated weekend review of a fraction of the watchlist.
• Nine concurrent technical pattern scans with scored matches, annotated price charts, and configurable detection parameters.
• Twelve fundamental ratio charts per name, generated both from scan results and direct ticker entry, covering valuation, returns, earnings quality, and capital structure.
• Automated rationale packs assembling FactSet and Bloomberg metric workbooks, pattern charts, and a pre-structured thesis scaffold for each qualifying name.
• All allocation decisions remain with the investment lead; the system produces the materials for review, not the decision itself.

The data foundation built at the outset is now the substrate for capabilities beyond the initial scope of the engagement. Custom fundamental gates, alert-driven re-scans triggered by universe composition changes, and deeper sector-level models are all viable extensions on the same data layer. Whatever the office chooses to build next begins from a structured foundation rather than from disparate terminal outputs assembled by hand.