Running the Seven Gates: From the Russell 2000 to Fifty Nine Companies, and the Accrual Band I Published Backwards: A Sharia Analysis for Muslim Investors (2026)

Verdict: The previous article in this series criticised a method for naming no metrics and publishing no output. It then named seven metrics and published no output either. This article closes that gap. The seven gates have now been run on the whole small cap market, and the funnel is set out here in full: 7,196 US listings become a 2,000 name index reconstruction, 1,437 non financial candidates, 492 survivors of the momentum gate, 82 companies that clear the six quality gates, and 59 that clear the leverage gate as well and can be confirmed inside the index's own regulatory filing. A second list of 63 replaces MSCI's total assets divisor with AAOIFI's market capitalisation divisor, and the two disagree about twelve companies for reasons worth reading. One correction comes first, because it is mine.

Post 65 published this row in its table of gates: accrual ratio, cash flow based, 50 to 100 percent. The text glossed it as a demand that reported profit be substantially backed by operating cash. That row is wrong twice over, and a reader running the screen from it would have been quietly worse off.

The first error is that it names the wrong statistic. An accrual ratio is a term of art with a paper behind it. Richard Sloan, writing in The Accounting Review in July 1996, defined accruals operationally as "the change in non-cash current assets, less the change in current liabilities (exclusive of short-term debt and taxes payable) less depreciation expense, all divided by average total assets", and showed that this accrual component of current earnings is markedly less persistent than the cash component, with prices behaving as if investors fail to notice. A quantity expressed as a percentage that rises as quality rises is not that. It is a cash conversion ratio, operating cash flow divided by net income. The two statistics point in opposite directions, so publishing one under the other's name inverts the gate.

The second error is the shape of the band. On the accrual ratio proper, low is good and negative is better still. Sloan's own test portfolio, in his words, takes "a long position in the stock of firms in the lowest decile of accruals and an equal-sized short position in the stock of firms in the highest decile of accruals". Low accruals are the side you want to own. A band with a floor at 50 and a ceiling at 100 therefore rejects exactly the companies the research says to want. Even on the cash conversion reading the ceiling makes no sense, since a company turning 180 percent of its net income into operating cash is in better shape than one turning exactly 100 percent, not worse.

The replacement I now use, and the one this screen runs on:

Pass at or below 10 percent. Negative is better than positive. No floor.

Three deliberate choices sit inside that formula. It takes the cash flow statement version rather than the balance sheet version, because Paul Hribar and Daniel Collins showed in the Journal of Accounting Research in 2002 that measuring accruals as the change in successive balance sheet accounts, rather than reading them from the cash flow statement, contaminates the estimate with measurement error, acquisitions and divestitures being the obvious culprits. It scales by average total assets, which is Sloan's own denominator, rather than by average net operating assets, which is theoretically tighter but collapses for cash rich, asset light small companies: in this run the net operating assets denominator silently discarded 26 of 442 companies, and total assets discarded none. And it is one sided, because a two sided band throws out the best names.

That last point is not hypothetical. STRATTEC Security posts an accrual ratio of minus 12.1 percent, with operating cash flow at 3.8 times net income. It is the most cash backed company on the entire final list, and both readings of the gate I published throw it out: a two sided band of plus or minus 10 percent refuses it for being too far below zero, and a cash conversion ceiling at 100 percent refuses it for converting at 380. Under the corrected gate it ranks third by cheapness. One line of a published table, and a company that belongs near the top of the list is deleted from it.

Removing the floor does cost something, and the cost should be named rather than buried. Putting the whole investing line in the numerator means the measure picks up more than capital spending: purchases and maturities of marketable securities, acquisitions and disposals all land there, so a company shuffling money between deposits and short dated paper can post a large negative reading that reflects treasury management rather than cash backed earnings. In this screen only 11 of the 59 companies have an investing line consisting of capital expenditure and nothing else. The floor I removed was a crude guard against exactly that artefact. The replacement is not a lower bound but an inspection rule: any reading below about minus 10 percent should be checked against the cash flow statement before it is believed. STRATTEC survives that check, since its entire investing line for the year is 7.2 million dollars of capital expenditure against 71.7 million of operating cash flow, so the negative reading is a working capital release and not a securities transfer.

There is a Qur'anic frame for this that is not decorative:

The verse is usually read against the seller who shortchanges a customer. It reads equally against the analyst who publishes a scale that runs backwards. The duty attaches to the instrument, not only to the intention behind it, and a measure you got wrong is still a wrong measure when someone else uses it.

The Russell 2000 is not a list you can simply download. The fund that tracks it blocks scripted requests to its holdings file, so the universe had to be rebuilt and then verified against something the fund cannot refuse to publish.

Reconstruction first. FTSE Russell ranks every eligible US security by total market capitalisation on rank day, and describes the outcome plainly: "the largest 4,000 become the Russell 3000E Index and the Russell 1000 and Russell 2000 indexes are formed from the Russell 3000E: the largest 1000 become the Russell 1000 index and the next 2000 become the Russell 2000 index". So: 7,196 US listings from the Nasdaq screener, cut to 3,721 US incorporated common stocks after removing warrants, units, preferred lines and foreign domiciles, ranked by market capitalisation, and sliced at ranks 1,001 to 3,000. That produced a band running from 104 million to 5.35 billion dollars. FTSE Russell's own figures for the 30 April 2026 rank date give 1,996 constituents running from 146.4 million to 5.7 billion, averaging 1.1 billion. Close, and close is not the same as correct, which is why nothing was assumed.

Verification second. Every finalist was checked against the iShares Russell 2000 ETF's form N-PORT, series S000004344, for the period ending 31 March 2026, filed on 28 May 2026 under accession number 0002071691-26-012507, which lists 1,946 holdings lines. The check earned its place immediately: five companies that passed all seven gates were not in the index at all and were struck out, among them Seaboard and IPG Photonics. It also caught one false match of my own making, when a normalised name comparison paired RF Industries with UFP Industries, after which the matcher was tightened to require agreement on the first word.

The same filing settled a share class question that no screener would have flagged. Both Seneca Foods classes passed the gates. The fund holds one Seneca line, priced at 151.12 dollars per share, which is Class A's close that day and not Class B's 148.03. Class A stays and Class B goes. Sixty tickers matched a holding by name; once the share class was resolved, fifty nine are actual constituents, which is where the final count comes from.

Two honest limits, and the first is the more serious. The filing describes the index as it stood on 31 March 2026, and FTSE Russell reconstituted the Russell US indexes after the close on Friday 26 June 2026, on ranks taken on 30 April. From 2026 that rebalance happens twice a year rather than once, in June and December. Membership here is therefore verified as of March, before a reconstitution that has since happened, and a company sitting near the boundary between the large cap and small cap indexes may have crossed it in June. And banks, insurers and property trusts were removed before scoring, because Altman's Z score and Beneish's M score are undefined for balance sheets built from deposits and rent rolls. That removal is not the crude sector cut this series criticised last time. It is a statement about where two specific models apply.

Taken one at a time on the 442 fully scored companies, the gates refuse very different amounts. Altman's solvency test is the harshest, passing 189. Piotroski's nine point fundamental score passes 208. Greenblatt's earnings yield passes 279. The leverage gate passes 250. Beneish's manipulation model passes 363, and the corrected accrual gate passes 399, against 298 under the two sided version this screen ran before the correction. The gates are not redundant with each other, which is the point of running seven of them: only 82 companies satisfy all six of the non leverage tests at once.

The fifty nine, in order of cheapness: NATR, KE, STRT, EBF, BLBD, PNRG, SENEA, TTAM, SCSC, GIII, UTMD, RCKY, RCMT, IDT, DAKT, VMD, DORM, CIX, CMT, GIC, ESCA, TILE, SKY, ACIW, MCRI, BKTI, TDW, NTCT, CTS, HUBG, EML, MLR, RAMP, PGNY, HCSG, FSTR, FELE, PRSU, PSMT, NPKI, THRM, NSSC, WDFC, GRDN, NVEC, CXDO, LQDT, IRMD, DIOD, PKE, INOD, SLP, FIGS, RELY, RDVT, MRTN, VCEL, POWI, DCTH.

Ranked by earnings yield, since that is the gate where a higher number is a better one. Market capitalisation in billions of dollars, everything else in percent except the two scores and the Piotroski count.

Ten of the eleven were cross checked against the underlying regulatory filings, the exception being Titan America, which files as a foreign private issuer and tags no comparable data. The check found a systematic bias worth naming: the commercial data feed reports an operating profit a few percent above the figure the companies themselves tag as operating income, because it is built up from pretax income rather than lifted from the operating line. Nature's Sunshine is the widest gap, 29.9 million dollars against 24.7 million filed, which moves its earnings yield from 16.9 to 14.0 percent. STRATTEC moves from 10.0 to 8.9 percent, Blue Bird and ScanSource by less than a point, and Ennis and Utah Medical match the filed figure to the dollar. No gap moves any company across a gate, but a reader trading these numbers should take the filed figure. Screening tools that skip this step are not lying to you, they are just quoting a different definition than the one in the annual report. The same discipline is what the stock screener exists to enforce.

The debt gate is where finance and fiqh touch, and it is the one gate where the two disciplines have not agreed on the arithmetic.

MSCI's Islamic methodology puts all three of its financial screens over total assets. Its own words: total debt over total assets, cash and interest bearing securities over total assets, receivables and cash over total assets, with the instruction that none of the ratios may exceed 33.33 percent. Read one sentence further, though, and the operative number for a stock that is not already in the index is tighter: "in order to reduce index turnover resulting from financial screening, a lower threshold of 30% will be used in determining new inclusions", and a non constituent qualifies "only if all three financial ratios do not exceed 30%". AAOIFI Shari'ah Standard No. 21, on financial papers, issued in 2004 and published in English in 2015, uses the same 30 percent figure but divides by market capitalisation rather than assets, and adds a separate ceiling on income from prohibited sources at 5 percent of total income.

A gate of 30 percent of total assets, which is what this screen uses, is therefore not a tightening of the index convention at all. It is exactly MSCI's own admission threshold for a stock entering the index, applied with MSCI's own divisor. That is a defensible place to stand, and it is worth knowing that it is the threshold for getting in rather than the looser 33.33 percent for staying in. It is still not AAOIFI's.

One difference remains between this screen and MSCI's. MSCI strips Shari'ah compliant debt and instruments out of the numerator before taking the ratio, which this screen does not. MSCI applies that carve out only to a named list of markets, the Gulf Cooperation Council countries excluding Saudi Arabia, along with Bangladesh, Egypt, Indonesia, Malaysia, Pakistan and Turkey. No US company falls inside it, so nothing in this universe is affected.

The difference is not cosmetic. The assets divisor measures the company. The market capitalisation divisor measures the company as priced today. One is stable through a mania and blind to price; the other tracks what you actually pay and drifts with sentiment. Neither is obviously the more conservative test, which is precisely why running both was worth the effort.

Rerun the identical six quality gates and swap only the leverage test, requiring debt below 30 percent of market capitalisation and cash plus interest bearing securities below 30 percent of market capitalisation, and the screen returns 70 tickers. Sixty four match a holding in the index filing, and the same Seneca share class resolution takes that to 63 companies.

One approximation in that second ratio has to be admitted, because it decides two of the twelve disagreements below. AAOIFI's test is on interest taking deposits. An annual report does not usually tell you which portion of a cash balance sits in an interest bearing account and which does not, so this screen uses the whole of cash and short term investments as the numerator. That is the conservative reading and it is what practitioner screeners do, but it is a proxy, and a company holding genuinely non interest bearing cash is being failed for something AAOIFI does not test. Anyone applying this to a specific company should read the cash note in the accounts rather than accept the proxy.

Fifty five companies appear on both lists. Twelve do not, and they split cleanly into two failure modes.

Utah Medical is the cleanest illustration of the first mode. It carries debt equal to two tenths of one percent of its assets, which is as close to debt free as a listed company gets, and it fails the AAOIFI screen anyway, on the cash ratio, because its cash pile is 37.7 percent of what the market says the whole company is worth. Nothing is wrong with the business. The market has simply priced the equity low enough that the cash looms large inside it.

Harrow is the cleanest illustration of the second. Its debt is 63.1 percent of its assets, twice any threshold in this article, and it passes AAOIFI comfortably at 16.0 percent of market capitalisation, because the market values the equity at several times the book value of the assets that debt sits against. An investor using the assets divisor sees a heavily borrowed company. An investor using the market capitalisation divisor sees a lightly borrowed one. Both are looking at the same balance sheet.

That is the honest state of the disagreement. Twelve companies out of sixty seven, roughly one in six, change status depending on which denominator a scholar's methodology chose decades ago.

Gharar in its classical form attaches to the contract and its object, to sale of what is not yet possessed or not yet known. Extending it to a screening process is my reasoning rather than a transmitted ruling, and I would rather say so than dress it up.

The extension is this. A screen is a claim about knowledge. When the metrics are unnamed, the buyer is accepting an outcome produced by a process he cannot inspect, which is closer to the structure gharar objects to than to an informed purchase. Publishing the seven formulas, the thresholds, the universe, the survivor count at every stage and the tickers that came out is the removal of that uncertainty, not a display of confidence. Everything in this article can be rebuilt by a reader who disagrees with any part of it, which is the only real test of whether a method exists.

Where uncertainty genuinely remains, it should be stated rather than smoothed. Titan America listed in early 2025 and has a short filing history, so its Beneish and Piotroski inputs rest on fewer years than the others. PrimeEnergy and Seneca have stopped tagging operating income in their structured filings, so their operating profit was derived from revenue less costs. Seneca's most recent year was unusually strong, with net income of 114.7 million dollars, and its earnings yield falls to roughly 5 percent if computed on the prior year instead. None of that is a reason to drop them. All of it is a reason to look before buying.

Maysir is the transfer of wealth on an outcome that is neither earned nor analysed. The seven gates are not maysir. They are a refusal to gamble, expressed in arithmetic: a solvency test, a fraud test, a leverage test, a fundamentals test, an accruals test, a trend test and a price test, each with a published number and each declining a specific way of losing money.

The momentum gate is the one that needs its fiqh stated rather than assumed, because none of the scholars in scope for this blog has written on price trend bands, so what follows is reasoning from principle on my own account. The floor at 10 percent declines to buy a business the market is actively repricing downwards while you are still forming a view, which is a discipline about timing rather than a claim about value. The ceiling at 100 percent declines to buy after a doubling, which is where analysis quietly hands over to crowd following. In this run both ends did real work. The floor refused 648 companies and the ceiling refused 220, and the most extreme name the ceiling caught had risen more than twentyfold over the measurement window. Keeping that kind of stock out of a value list is exactly the failure the previous article documented in a portfolio assembled during the meme stock peak.

The same reasoning does not transfer to digital assets, where there is no EBIT line, no accrual to measure and no balance sheet to score. That is a different question with a different answer, and it belongs to the crypto screener.

The divisor is the live disagreement, and it is not going to be settled here. One caution about the shape of it first: only one side of this is a scholarly position. AAOIFI's market capitalisation divisor comes from a Shari'ah standard. MSCI's total assets divisor comes from an index methodology document written by a commercial index provider whose stated reason for its own 30 percent admission threshold is "to reduce index turnover". That is an operational concern, not a fiqh one, and a reader should weigh the two accordingly rather than treating them as two schools of thought.

Mufti Faraz Adam sets out the AAOIFI criteria directly in "Making Sense of the 30% Rule in Islamic Finance", published by Amanah Advisors on 14 December 2020, quoting all three tests: interest based debt not exceeding 30 percent of market capitalisation, interest taking deposits not exceeding 30 percent, and income from prohibited components not exceeding 5 percent of total income. His argument is not that the numbers are wrong but that they were justified by necessity, and he asks whether percentages adopted on that basis should be "elastic to contextual changes". That question sits underneath this entire article: both lists here are built on thresholds nobody claims were revealed.

Sheikh Joe Bradford is named as Zoya's Shari'ah advisor on his own site at joebradford.net, and Zoya's published help centre article on how it screens applies the AAOIFI style thresholds with market capitalisation in the denominator. The rationale I would offer for that divisor, that a shareholder buys a claim at market prices so his exposure should be measured against what he actually pays, is my inference from the practice rather than a quotation from either source.

Mufti Taqi Usmani, in his paper Principles of Shari'ah Governing Islamic Investment Funds, does not adjudicate between denominators, and it would be an overreach to enlist him for one. What his framework does add is that admission by a ratio is not the end of the shareholder's duty: disapproval of the impermissible portion, and purification of income traceable to it, remain owed whichever screen let the stock through.

Where that leaves a reader is with a choice to make consciously. Both lists are published above. An investor who follows AAOIFI should take the second, and should know that it admits Harrow at 63 percent debt to assets. An investor who follows the index convention should take the first, and should know that it admits Utah Medical while AAOIFI would not. What no one should do is take whichever list happens to contain the stock he already wanted, which is how a screen becomes a decoration.

Monarch Casino and Resort passes all seven gates. It passes on both divisors. It sits on both published lists, and no arithmetic in this article removes it, because it operates casinos.

That deserves to be stated as plainly as possible, because the previous article in this series criticised another writer for treating a single sector exclusion as a Shari'ah screen. The seven gates are a quality and valuation screen. They are not a compliance screen. Activity screening comes first and separately. MSCI's methodology excludes companies "directly active in, or derive more than 5% of their revenue (cumulatively) from" a named list, and the list names gambling and casinos, alcohol, tobacco, pork, conventional financial services, adult entertainment, hotels, cinema and music, each defined in its own sentence. AAOIFI applies its own 5 percent ceiling on impermissible income. Neither test can be computed from a balance sheet. Both require reading what the company actually sells, which is work no screen does for you.

Several other names on these lists raise activity or income questions that the ratios do not answer, and a reader should not assume otherwise. Neither list is a halal list. Both are the arithmetic that comes after the halal question has already been answered somewhere else. What that prior work looks like on a single company, activity by activity and revenue line by revenue line, is set out in the Apple screening piece, and reading the business itself the way the DuPont analysis piece sets out is where it begins.

Run the activity screen first, and run it yourself. A list of fifty nine companies that clear seven quantitative gates is a research queue, not a portfolio, and the first thing to do with it is delete the ones whose business a Muslim cannot own.

Take the filed numbers rather than the convenient ones. Where a data feed and an annual report disagree about operating income, the annual report is the company speaking under legal liability and the feed is a convention. On these eleven the gap reached almost three percentage points of earnings yield.

Decide your divisor once, in advance, and write it down. Choosing the assets convention or the market capitalisation convention after seeing which list contains your preferred stock is not screening.

Re verify membership rather than assuming it. Five companies here passed every gate and were not in the index. If a screen's universe is wrong, nothing downstream of it can be right.

Remember what comes after the screen. These seven gates say a company is solvent, honestly reported, lightly borrowed, improving, cash backed, not falling and not expensive. They do not say what it is worth. That is a valuation question, and the reverse discounted cash flow approach in the NVIDIA piece is where it starts. Zakat on whatever survives is a further and separate calculation, set out in the zakat on stocks article.

The distance between criticising a method and having one is the distance between naming seven numbers and running them. Running them produced two lists rather than one, because the fiqh has not settled which denominator belongs under the leverage ratio. Fifty nine companies clear the seven gates on the index convention. Sixty three clear them on the AAOIFI convention. Twelve companies sit in the gap between the two, and one casino operator sits on both lists as a permanent reminder that arithmetic screens for quality and never for permissibility. A method that publishes its errors alongside its output is not weaker than one that publishes neither. It is the only kind that can be checked.

This analysis is educational and is not a fatwa, a price target, or financial advice. The fifty nine and sixty three companies named above are the output of a quantitative screen, not recommendations: they have had no activity screening, no reading of the business and no valuation applied to them. Figures come from the filings and sources cited and change with each reporting period. Verify before acting, and consult a qualified scholar where your circumstances require it.

What exactly was wrong with the accrual band in the previous article? Two things. It named an accrual ratio but described a cash conversion ratio, which is a different statistic pointing the opposite way. And it set a band with a floor and a ceiling, when the research it rests on says lower is better without limit. The corrected gate is net income minus operating cash flow minus investing cash flow, divided by average total assets, passing at or below 10 percent.

Does the correction change the recommended companies? Yes, by one. Rocky Brands leaves the top eleven and STRATTEC Security enters it, in third place. STRATTEC generates operating cash flow at 3.8 times its net income, which is what the old two sided band had been rejecting it for.

Why remove banks and property companies before screening? Because Altman's Z score and Beneish's M score were estimated on industrial balance sheets and are not defined for institutions whose liabilities are customer deposits or whose assets are rent producing property. Removing them is a statement about model applicability, not a Shari'ah ruling, though most conventional banks fail the Shari'ah test on separate and more fundamental grounds.

Is the AAOIFI list halal and the other one not? No. Both lists are financial ratio screens. The AAOIFI list applies AAOIFI's leverage and liquidity ratios, which is one component of a compliance screen, but neither list applies the activity test or the 5 percent impermissible income test. Monarch Casino and Resort appears on both.

Which divisor should I use for the debt ratio? AAOIFI Shari'ah Standard No. 21 divides by market capitalisation. MSCI's Islamic methodology divides by total assets, admitting new constituents at 30 percent and tolerating existing ones to 33.33 percent. Both are used by serious practitioners, and this article publishes a list under each rather than declaring a winner. What matters is choosing before you look at the results.

How current are these numbers? Prices run through 20 August 2026 and fundamentals are the most recent annual statements as filed. Index membership is verified against a fund filing for the period ending 31 March 2026, which predates the reconstitution that took effect after the close on 26 June 2026, so a small number of these companies may have entered or left the index since. From this year the rebalance happens twice, in June and December, so the same caveat will recur.

Can I just buy the eleven? Not from this article. They are the cheapest eleven on one of two lists, ranked on a single metric, before any activity screening, any reading of the business and any valuation work. A screen narrows the field to what deserves study. Nothing in it substitutes for the study.