Screening Methodology

Version 0.3 · August 2026 · changes are versioned and announced

No scholar has reviewed this methodology. Daleel has no Sharia advisory board and holds no certification from any standards body. What we implement are the published thresholds of AAOIFI, Dow Jones, MSCI, S&P and FTSE, applied to audited SEC filings and shown with the arithmetic exposed so a qualified scholar — or you — can check our work rather than trust it.

That is deliberate, but it is a limitation, not a feature. A verdict here is a calculation, never a fatwa. Commissioning independent scholarly review is on our roadmap; until it lands, this page is the whole of our claim to authority.

Data sources, in order of trust

1. SEC EDGAR company facts (us-gaap XBRL) — balance sheet and income statement values, preferring the most recent 10-K. Every value in a response carries its XBRL tag, form, period, accession number, and a link to the filing.
2. Market data — market capitalization and industry/sector classification, labeled source: market_data.
3. Disclosed estimates — where filings do not break out a figure (e.g. alcohol share of restaurant sales), we apply an industry-level estimate and say so in the response notes.

The five methodologies

Each company is checked independently against the published thresholds of AAOIFI, the Dow Jones Islamic Market index, MSCI Islamic, the S&P Shariah index, and the FTSE Russell Shariah screen. Every standard caps non-permissible revenue at 5%; the balance-sheet screens differ. The response shows every check: value, threshold, pass/fail.

AAOIFI screens debt and cash & interest-bearing securities, each < 30% of market capitalization. MSCI and FTSE Russell screen debt, cash & interest-bearing securities, and receivables + cash against total assets (FTSE: debt < 33.333%, cash < 33.333%, receivables + cash < 50%; MSCI's receivables + cash ceiling is 70%). Because they are asset-based, both stay computable even when market cap is unavailable. DJIM and S&P Shariah retired their cash and receivables screens in September 2023 and now apply a single leverage screen — debt < 33% of trailing-average market capitalization (24 months for DJIM, 36 for S&P; we approximate with point-in-time market cap and disclose it) — plus the 5% revenue ceiling.

The 0–100 score

Six components: business activity (25), interest income (25), debt (20), cash & securities (10), receivables (5), and cross-methodology agreement (15 — 3 points per methodology passed). A verdict of compliant requires a score of 65+ and methodology support — a verdict never contradicts all five methodologies.

Hard rules

• Primary haram products (alcohol, tobacco, gambling, adult content) → non-compliant regardless of financials.
• Interest income ≥ 20% of revenue → non-compliant.
• Financial-sector companies failing all five methodologies → non-compliant.
• Missing data → not screenable. Never assumed compliant.
• Missing market cap → market-cap ratios treated as failing (fail-safe), asset-based checks still computed.

Non-permissible revenue

Interest income (from filings) plus an industry-level estimate for categories with known haram revenue share (restaurants ~4% alcohol, hotels ~8%, supermarkets ~3%, etc.). The purification percentage equals this combined share — the portion of dividends and gains to donate.

ETFs

Screened by look-through of publicly disclosed top holdings, weighted by fund weight. Coverage is disclosed on every response; below 45% coverage we cap the verdict at questionable. More than 5% of screened weight in non-compliant holdings fails the fund.

Values overlays

Optional, advisor-oriented flags layered on top of the halal screen (?overlays=all on the screening API, any keyed tier). Overlays are informational only — they never change the halal verdict or the 0–100 score, which are computed solely from the methodologies above.

Official government lists

Each bundled list snapshot carries its official source URL and publication date, cited on every response:

• UFLPA Entity List — U.S. Department of Homeland Security, forced-labor enforcement. dhs.gov/uflpa-entity-list, list as updated July 31, 2026 (43 additions effective August 3, 2026; 187 entities per DHS).
• DoD Section 1260H Chinese Military Companies List — U.S. Department of Defense. Federal Register, June 10, 2026 (91 FR 35189), 188 entities including named subsidiaries.
• Florida SBA scrutinized companies (Sudan / Iran) — Protecting Florida's Investments Act prohibited-investments list, February 24, 2026 quarterly update (97 companies). A June 9, 2026 update exists; our snapshot may lag it by one quarter, and the response citation says exactly which report it came from.

No fuzzy matching. Official-list flags trigger only on a known ticker or an exact normalized-name match (case, punctuation, and Ltd/Inc/Corp/Co suffix variants) against the listed entity names, their published aliases, and explicitly named subsidiaries. A fuzzy false positive is worse than a miss, so anything short of an exact match is reported as not flagged — which means supply-chain exposure, unlisted subsidiaries, and renamed entities are not detected. Most listed entities are not US filers at all, so a clean overlay is evidence of absence from the list snapshot, not evidence of no exposure.

Industry-derived flags

Computed from the same market-data industry/sector classification shown in every response: weapons_defense, fossil_fuels, tobacco, alcohol, gambling, adult_entertainment, plus two curated ticker sets for categories no classifier exposes — private_prisons (GEO, CXW) and civilian_firearms (SWBI, RGR, VSTO, POWW, OLN — Olin via its Winchester ammunition business).

The daily value screen

The value screen is a separate pipeline that runs once a day over the same bundled universe and publishes at most 25 names. It is a screen of candidates for research — identical for every reader, never personalized, never a recommendation to buy, and not a fatwa. Four gates run in order, and each one only ever removes names:

1. Cheapness. Seven valuation multiples (trailing and forward P/E, P/B, P/S, EV/EBITDA, PEG, free-cash-flow yield) are percentile-ranked inside the company's own industry cohort, widening to the sector only when the industry bucket has fewer than 8 reporting peers. A multiple that is negative or zero is excluded rather than ranked "cheapest". A composite resting on fewer than 3 metrics is not treated as a cheapness score at all. The top 150 go on.
2. Quality. Seven Buffett-style components (return on equity, approximate ROIC, margins, free cash flow, leverage and interest coverage, growth, share-count trend) each score 0–100 from a documented linear ramp; the headline is the weighted mean over the components that could actually be computed. Debt prefers the cited SEC EDGAR figure. Anything below 45 is dropped — cheap against weak fundamentals is the trap this gate exists to remove.
3. Why it is cheap. Filed annual series (revenue, operating margin, free cash flow, debt, interest coverage) and an observed price panel classify the discount. Only sector wide and fundamentals intact are published. Fundamentals deteriorating and leverage stress are dropped by name, and so is insufficient evidence: if no classification can be asserted, the reason for the discount is unestablished, and an unexplained discount is not published.
4. Halal. The full five-methodology screen above. Only compliant and questionable survive, and questionable is flagged on its row rather than silently mixed in. Not screenable is dropped: missing data is never treated as passing.

Survivors are ranked by one disclosed blend — cheapness 40, quality 25, classification confidence 15, and distance to the central peer-relative fair-value estimate 20 (0% scores 0 points, 50% or more scores 100). Weights are renormalized across whichever components a name can actually supply, and every row shows its own arithmetic. The rank is a sort order for a research list, not a forecast of return.

Fair-value ranges

Each range is three transparent arithmetic exercises: what the share price would be at the cohort's median trailing P/E, EV/EBITDA and price-to-book, holding the company's own reported earnings, EBITDA and book value fixed. These are not discounted cash flows, not intrinsic value, and not price targets. A whole cohort can be mispriced together, and a median says nothing about which side of it any one company belongs on. Any method whose inputs are missing or non-positive is excluded with a reason rather than run on an assumed value — net debt is never assumed to be zero.

Published performance

Every list is snapshotted the day it is published, with the entry price of every name and of both benchmarks (SPY and SPUS). Past lists are then scored from their own publication date to the latest close in the same price panel that builds the screen: equal weighted, no rebalancing, and every name that was published stays in the arithmetic — the losers are named individually alongside the winners. A name whose current price is unavailable is excluded and counted, never carried at its entry price.

A list published fewer than 30 days ago gets no headline number. Over a window that short the figure is market noise, and a noisy number printed beside a benchmark reads as a claim; those rows say "too early" and how many days remain. Because each list is measured from its own publication date to today, the windows overlap — the across-list average is an average of overlapping single-list returns and is not a portfolio return.

Limits of this screen

• Cheapness is relative. A whole industry can be expensive, and the cheapest name in it still ranks well here.
• Filings are backward-looking and published with a lag. The market may be pricing in information that has not reached a filing yet, and "fundamentals appear intact" never establishes that a decline will reverse.
• Coverage is the bundled US universe only, and a name with thin market data or too few industry peers is excluded rather than estimated — absence from the list is not a verdict on the company.
• Confidence is capped at 85 for every classification, whatever the evidence, and at 40 when fewer than two comparable filed periods exist.
• Sector and industry classification comes from market data, so a misclassified company gets a misclassified peer cohort.

Known approximations

• DJIM and S&P specify trailing-average market-cap denominators; we use point-in-time market cap.
• Bank XBRL interest tags can understate interest income — hence the financial-sector methodology rule above.
• US-listed securities only; international filings have no free auditable EDGAR equivalent yet.

Disagree with a threshold or an estimate? Email daleel@o11r.com — the methodology is versioned precisely so it can improve in public.