Summary
Over the past week we turned two disparate data sources—Stock Rover’s CSV screeners and the Zen Research Terminal XLSX—into a single, automatically validated fuel supply for the Vulcan-mk5 research engine. The work produced a 4 800-stock cache with 35 clean, model-ready metrics per company, giving Vulcan-mk5 instant, offline access to ~170 000 individual data points. Below is the inside story (no proprietary code revealed) of how we built it, debugged it, and pressure-tested it with real tickers like UBER, CTRE, STX, and ADDYY.
Why We Needed a Dual-Source Feed
- Coverage vs. Depth – Stock Rover offers broad market reach and granular factor data (profitability, volatility, dividends).
- Forward-Looking Valuation – The Zen sheet supplies consensus fair-value estimates, buy-range tags, and long-term return forecasts that Stock Rover doesn’t track.
- Zero-Lag Model Runs – Caching both files locally lets Vulcan-mk5 run full Monte Carlo and Bayesian simulations without live pulls, avoiding rate limits and stale quotes.
The 56-Metric Rule
Stock Rover caps custom exports at 56 data columns (plus the ticker). After several iterations we settled on a “golden set” that satisfies every Vulcan-mk5 pillar:
| Pillar | Key Metrics Selected* |
|---|---|
| Value | Fair Value, Margin of Safety, Forward P/E, PEG Forward |
| Quality | ROIC, Gross/Operating/Net Margins, Piotroski F, Altman Z |
| Safety | Debt/Equity, Interest Coverage, Current Ratio, Volatility 1-Yr, Beta 3-Yr |
| Growth | 5-Yr EPS CAGR, Sales 5-Yr Avg %, Cap-Ex % Sales |
| Income | TTM Yield, Fwd Yield, Payout Ratio, Dividend 5-Yr Avg %, Forward Div Growth |
| Cash-Flow / DCF | Free Cash Flow / Share, FCF % Sales, Model FCF Y5 |
| Sentiment / Target | Mean Consensus Target Price, EPS Est. Next Year |
| Context | Sector, Industry, Shares Outstanding, Margin of Safety |
*We also import the parallel Zen fields—Quality Score, Safety Score, Fair-Value Bands—for cross-check and Bayesian priors.
How We Validated the Feed
- Double Parse – Every file is read twice with checksum comparison; row/column counts must match.
- Schema Guardrails – A YAML map ties each column header to its Vulcan-mk5 field; if a header is missing the loader halts.
- Out-of-Range Flags – Basic sanity checks (e.g., margins > 100 %, negative Z-Scores) quarantine bad rows.
- Hash Checksum – The post-processed cache writes a SHA-256 hash to the Audit Table so every report can self-verify its numbers.
Debugging in the Wild
- Missing Metrics – Stock Rover lacked S&P credit ratings, so Vulcan-mk5 now maps the on-board “Safety Score” to its bankruptcy-risk overlay, defaulting to BBB– if absent.
- Column Trade-Offs – We dropped intra-quarter EPS estimates to make room for higher-impact fields like Free Cash Flow as % Sales.
- Monte Carlo Smoke Tests – Each time we tweaked the column set, we ran a one-ticker simulation; when ADDYY’s median path aligned with its consensus target, we knew drift and σ were mapping correctly.
Real-World Trial Runs
| Ticker | Data Source Used | Outcome |
|---|---|---|
| UBER | Stock Rover + Zen | Full 2-chart Vulcan run; flagged 57 % discount and 18 % downside tail. |
| CTRE & STX | Stock Rover only (first versions) | originally failed—missing FCF / Sales and Sales CAGR; fixed after metric swap. |
| ADDYY | Final 35-metric set | Delivered a complete 1 800-word mini-report with Monte Carlo & Bayesian fan chart. |
Lessons Learned
- Start Wide, Trim Later – We began with 40 + metrics, then pruned until every column mattered.
- Map Synonyms Early – Stock Rover uses “Sales 5-Year Avg (%)”, Zen uses “Revenue CAGR 5Y”; a small YAML alias table saved hours.
- Automate Sanity Checks – A routine that yells when Fair Value < 0 or Yield > 25 % caught bad parses instantly.
- Visual Proof Beats Numbers Alone – Stakeholders “get it” when they see the Bayesian fan chart sitting beside the Audit Table hash.
Where We Go Next
- Macro Switchboard – Hook CPI, Fed-Funds path, and FX indices into the Bayesian node for macro-conditioned probabilities.
- ESG Overlay – Integrate Sustainalytics scores as a 5 % weight in the Quality pillar.
- Live Delta Mode – Hourly differential ingest so Vulcan-mk5 highlights which tickers slipped into “Ultra-Value” since last close.
Closing Thoughts
Merging Stock Rover’s breadth with Zen Research’s depth turned a pile of CSVs and an XLSX into a 170 000-cell data engine—and Vulcan-mk5 now digests the lot in seconds. The payoff: full-fidelity, visual-rich stock reports that need zero live queries, cut research latency, and keep your secret sauce private.
Stay tuned; the next iteration will surface fresh “Strong Buy” opportunities the moment they cross your margin-of-safety threshold—no spreadsheet wrangling required.
Vulcan Stock Research 
Leave a comment