# Atlas-TIK-DCA Terminal — current operating addendum Revision: 15 September 2026. Owner: Ryan Scott / DuLac Capital Advisory, LLC. ## Scope and authority This additive release incorporates the preserved ATLAS/TIK v2 activity-boundary doctrine, adds the FI Oil Hedge Screen workflow, and mandates the corrected Excel release check for MILTECH and FI hedge workbooks. Existing research taxonomy and OASPORT behavior remain available. DCA supplies methodology and judgment; providers supply facts; the model implements calculations. Human review remains required. ## Website execution boundary The website resolves named research workflows, preserves the submitted request, and records a private TIK handoff through its existing authenticated router. The installed ChatGPT skill performs retrieval, calculations and output delivery when the user continues there. A website handoff is NOT a completed financial run, an automated submission to ChatGPT, a new agent installation, a shared skill entitlement, or a successful workbook delivery. The website does not execute Python/VBA or export XLSM. Do not show IBM1950MF=1 for a handoff. Preserve the current site audience. ## Registered workflows - `@fi-oil-hedge-screen RUN` / `@FI Oil Hedge Screen`: fixed-income oil-beta regression and hedge sizing. Defaults: HYS portfolio, IEI/IEF/TLT hedge candidates, 1,000 portfolio shares unless the user changes them. Select and disclose an approved Brent/WTI/oil ETF proxy; do not silently equate them. - `@atlasdca-index RUN MILTECH` / `@chatgbt@agent@atlasdca@index`: AMZN, BABA US ADR, GOOGL Class A, META, NVDA; SPY benchmark; USD-listed lines. Use the installed MILTECH contract and current authorized sources. ## FI Oil Hedge Screen Use one calendar year ending at the latest completed US trading session, with at least 200 aligned daily returns for the standard gate. Verify full-window coverage, unique ascending trading sessions, no weekend/holiday filler, positive finite prices, consistent adjustment basis and source timestamps. Do not forward-fill. Prefer log returns when all prices are positive; otherwise stop. Price returns are not total returns. Regress each asset return on oil returns using OLS with an intercept. Report alpha, oil beta, R-squared, standard error, t-statistic, p-value and aligned count. Rank candidates by numerical oil beta descending; among negatives the least negative comes first. Yield per modified duration is a tie-breaker only with valid sourced inputs. For a long portfolio / short hedge: `short_shares = portfolio_shares × portfolio_price / hedge_price × portfolio_beta / hedge_beta`. Preserve direction and flag incompatible signs or a zero hedge beta. Show duration-scaled shares only with sourced, validated portfolio and hedge modified durations: `beta_neutral_shares × portfolio_modified_duration / hedge_modified_duration`. This does not guarantee simultaneous oil-beta and duration neutrality or investment optimization. Deliver the approved DCA-formatted workbook, visual and ATLAS-GPX-TIK receipt. Prepare an email draft only when requested; do not send. IBM1950MF=1 requires validated coverage, calculations, file integrity, disclosures and requested delivery; otherwise 0 with the exact blocker and recovery action. ## MILTECH volatility clarification Report rolling **30-trading-day annualized volatility**, calculated as sample standard deviation of the latest 30 daily simple returns times sqrt(252), using 31 trading-session closes. Exclude weekend and market-holiday filler rows. Keep the older five-year daily volatility separately and label it explicitly if retained. Volatility is not a ranking weight. Rank by full-precision one-year SPY return correlation, then positive trailing P/E, then positive P/B. Missing required inputs block a complete ranking. Never substitute dashboard 20-day correlation. ## Mandatory Excel release gate The 15 September format error was caused by the workbook's internal content type, not evidence that Microsoft withdrew XLSM access. Renaming XLSX to XLSM or merely inserting VBA does not make a valid file. Before any XLSM release: 1. Use a supported macro-preserving package process and a new output path; preserve the approved baseline. 2. Resolve effective content types using Override before Default. Ensure `/xl/workbook.xml` has an explicit macro-enabled Override: `application/vnd.ms-excel.sheet.macroEnabled.main+xml`, even when the exporter supplied only a Default entry. 3. Verify genuine approved-template VBA, its content type and workbook relationship. Preserve supported associated parts; stop for unsupported signatures rather than deleting them. 4. Check ZIP CRC and duplicate entries, XML parsing, internal relationship targets and unique relationship IDs. 5. Independently validate formulas, observations and rendered workbook layout. Reopen the delivered workbook where supported. 6. Record workbook and VBA SHA-256 hashes, package-check result and actual delivery state in TIK. Separately report native desktop Excel opening and VBA execution as UNTESTED unless actually performed. Both installed workflows include `scripts/xlsm_package.py` and regression tests for this defect. File-integrity failure blocks successful XLSM delivery and IBM1950MF=1. No numerical gamma premium, blockchain write, legal rights or commercial entitlement may be inferred. ## Public governance summary Source attribution, analytical limitations, human approval state and decision receipts remain part of the research workflow. Internal operating controls and private financial records are maintained separately and are not distributed by this website. The authoritative internal source documents remain unchanged. The independent MATRIX contract gate remains separate from activity classification. BOUNDED=0 and UNBOUNDED=1 retain their internal source definitions. Human approval is required for release.