Paper 02 · v1.0 · September 2026

After Halal screens, do high returns that get reinvested still compound?

After banned businesses and AAOIFI limits, keep names with ROIC above 15% that put a large share of that profit back into the business. Own them in proportion to size. This page is a reading of that backtest. It is an exploratory note, with no probability attached and no live-return target.

Read technical report Jump to findings

Olive loops are profit put back into the firm. Gray strokes are profit that leaves.

Scroll to read

Return on invested capital asks how hard each dollar of operating capital works. Profit after tax, divided by the capital already in the business. Reinvestment asks what share of that profit is put back in, instead of paid out. A high return compounds only when it is plowed back into a base that can earn it again.

Islamic equity screens add a second filter. AAOIFI caps interest-bearing debt, cash, and receivables against market value. A separate screen removes banks, conventional insurance, alcohol, tobacco, gambling, and weapons. The hope is that those caps already drop the highly indebted firms, so that among what remains, ROIC and reinvestment are cleaner signals.

That hope is not automatic. The screens change the sector mix. A backtest that looks strong after them may be riding the same mega-cap platform names that dominated 2023 and 2024. The original aim was compounding in capital-light sectors — SaaS, healthcare, and med-tech. Those sectors have to be checked.

How a return becomes a larger base

One dollar, followed through a cycle

The firm

Invested capital

growth ≈ ROIC × reinvestment

Invested capital
Return kept in the firm
Paid out

One dollar of capital

Invested capital is the money already at work in the business. ROIC asks how much operating profit that dollar earns.

One dollar of capital

Invested capital is the money already at work in the business. ROIC asks how much operating profit that dollar earns.

It earns a return

ROIC is NOPAT divided by invested capital. The live floor is 15%. A print above 500% is dropped as a broken denominator.

The return splits

Reinvestment is the share put back: capital spending above depreciation, the change in working capital, and R&D. The rest leaves.

What stays becomes next year’s base

The same ROIC on a larger base earns more dollars. Growth is roughly the return times the share put back.

A harvester does not compound

Apple earned a 57% ROIC and still ranked 58 of 99 on reinvestment. The rule leaves that cash-rich harvester unbought.

The reinvestment rate used here is capital spending minus depreciation, plus the change in net working capital, plus research and development, all divided by NOPAT. R&D is counted on purpose. A software firm or a med-tech firm should not look like a low-reinvestment business just because it owns no factories.

The tax rate is the effective rate when the filing supports it, and 21% otherwise. Names with ROIC above 500% are dropped. Fundamentals come from annual statements, lagged 90 days, so the figure is known before it is used.

Two gates, then size

The two gates

Two questions, one plane

Horizontal is how hard capital works. Vertical is how much of that profit goes back into the firm. The chips mark named outcomes. They are not a plot of the 99 filings.

Two questions, one plane

Horizontal is how hard capital works. Vertical is how much of that profit goes back into the firm. The chips mark named outcomes. They are not a plot of the 99 filings.

A hard floor at 15%

Left of the line, the name is out, even if it reinvests heavily. ISRG, SYK, TMO, INTU, NOW, and ORCL sat there on these filings.

Then the top half by reinvestment

Among names already above 15%, with positive profit and positive reinvestment, keep the higher half. About 50 names are bought in a typical month.

The kept corner

Nvidia, Microsoft, Alphabet, Adobe, and Lilly were in it at year-end 2024. Apple was to the right of 15% and below the reinvestment cut.

Held

NVDA

Held. 20.4% of the book at the last rebalance. It cleared the ROIC floor and sat in the top half by reinvestment.

Not bought

AAPL

ROIC of 57%, rank 58 of 99 on reinvestment. The rule treats that as a cash-rich harvester and does not buy it.

Not bought

ISRG

Below the 15% ROIC floor on the Yahoo filing, with a strong reinvestment rate. SYK, TMO, INTU, NOW, and ORCL sat with it.

Held

GOOGL

Held, and so is GOOG. The two share classes together were about 28.7% of the book. That double-counts Alphabet.

At a typical month the funnel is about 418 names with fundamentals, about 100 Halal names above the ROIC floor with positive reinvestment, and about 50 actually bought. Median book size is 51 names. If fewer than 20 survive, that month is skipped.

Weights are market cap. There is no equal-weight book, no volatility target, and no single-name cap in this test. At the last rebalance Nvidia was 20.4% and Microsoft 19.0%.

Same return, different loop

Illustration · ROIC 20% · fundamental capital, not price

Puts 80% back

$1.00

Puts 20% back

$1.00

Each column is the capital base after that year. The olive pile pulls away because the return is plowed back in.

Same return, year zero

Two firms. Both earn 20% on invested capital. Both start at one dollar. Only the share they put back will differ.

Same return, year zero

Two firms. Both earn 20% on invested capital. Both start at one dollar. Only the share they put back will differ.

After one year

The firm that puts 80% back grows about 16%. The firm that puts 20% back grows about 4%. The rate was the same.

The loop widens

Year two starts from a larger base on the left. The same ROIC now produces more dollars there, and those dollars are put back again.

After three years

The gap is already wide. Nothing else changed. The olive pile is larger only because last year’s return stayed inside the firm.

After four years

About $1.81 versus about $1.17. Same 20% ROIC. Only the share put back changed. This is fundamental capital, not a stock price.

Try the identity

Set the return and the share put back

This is the textbook loop, five years of fundamental capital. It is not the backtest, and it is not a price forecast. A high ROIC with nothing put back leaves the base where it started.

ROIC25%Reinvestment80%

25% × 80% = 20.0% a year

$2.49

One dollar of capital, five years later, before any change in what the market will pay.

The picture above is the identity: growth is about ROIC times the share put back. The backtest is a different object. It is a cap-weighted stock portfolio, from 31 October 2022 through 30 December 2024, scored against SPY and against SPUS.

Inside the real high-ROIC pool, the high-reinvestment half did compound faster in price: 43.2% a year against 28.7%. That rhyme is the useful evidence. It is still one window.

Growth of $1, from calendar-year returns

Compounders
SPUS
SPY

Growth of $1 from calendar-year returns in the live window
Book	Ending value	CAGR
Compounders	$2.05	39.57%
SPUS	$1.69	27.35%
SPY	$1.56	22.94%

Three published years

Growth of $1 from 31 October 2022 to 30 December 2024. The line joins the paper’s calendar-year returns. It is not the daily path. 2022 is only the stub.

Three published years

Growth of $1 from 31 October 2022 to 30 December 2024. The line joins the paper’s calendar-year returns. It is not the daily path. 2022 is only the stub.

Without the ROIC cut

SPY finishes near $1.56. That is 22.9% compound annual growth.

The Halal index

SPUS finishes near $1.69. That is 27.4% a year. Beating SPUS is not the same as beating every Sharia book.

High ROIC, high reinvestment

The live book finishes near $2.05. That is 39.6% a year. It also took a deeper hole: −15.3%, against −10.0% for SPY and −11.5% for SPUS.

What the window shows

Finding 1: Total return

The compounder book led both benchmarks, and it fell harder on the way.

Compounders

39.6%

CAGR · max DD -15.33%

SPUS

27.4%

CAGR · max DD -11.46%

SPY

22.9%

CAGR · max DD -9.97%

Calendar-year total returns. 2022 starts 31 October.
Year	Compounders	S&P 500	SPUS
2022from 31 Oct	-2.3%	-1.2%	-1.6%
2023full year	56.7%	26.2%	34.2%
2024full year	34.2%	25.3%	27.7%

39.6% compound annual growth versus 22.9% for SPY and 27.4% for SPUS, from 31 October 2022 through 30 December 2024. Sharpe was 1.71 versus 1.41 and 1.45. Maximum drawdown was −15.3%, deeper than SPY (−10.0%) and SPUS (−11.5%). Calmar still led, 2.58 versus 2.30 and 2.39, because the return gap was large.

These are annualized figures from a mega-cap bull market, with only a stub of 2022. They are not a forecast of 40% a year.

Same ROIC floor. Split only on reinvestment. CAGR, live window.

Puts it back0.0%

Max drawdown -15.33% · Calmar 2.82

Pays it out0.0%

Max drawdown -11.45% · Calmar 2.51

+14.5 points

High reinvestment also took the deeper drawdown. The gap is the return, not a free reduction in risk.

ROIC quintiles inside the Halal universe, CAGR

Q1 highest

27.1%

33.8%

16.9%

26.2%

Q5 lowest

18.9%

Cap-weighted ROIC quintile CAGRs
Q1 highest	27.09%
Q2	33.81%
Q3	16.85%
Q4	26.23%
Q5 lowest	18.9%

Five buckets, ROIC only

Same Halal universe, cap-weighted, split by ROIC. Q1 is the highest return on capital. Q5 is the lowest. Reinvestment is not in this sort.

Five buckets, ROIC only

Same Halal universe, cap-weighted, split by ROIC. Q1 is the highest return on capital. Q5 is the lowest. Reinvestment is not in this sort.

The spread you would hope for

Q1 returned 27.1%. Q5 returned 18.9%. Q1 was ahead on 52% of days. A 6.9 point gap, and a coin-flip on the daily lead.

Q2 is the tall bar

Q2 returned 33.8%, ahead of Q1. Q4, at 26.2%, nearly matches Q1. Q3 is last, at 16.9%. Higher ROIC did not line up as a staircase.

Finding 2: Reinvestment, not ROIC alone

Inside the high-ROIC pool, putting capital back in beat paying it out. Sorting on ROIC alone did not line up.

High reinvestment returned 43.2% a year. Low reinvestment returned 28.7%. That is a 14.5 point gap. The high-reinvestment sleeve also drew down more, −15.3% versus −11.5%. Across ROIC quintiles, Q2 led at 33.8% and Q1 was 27.1%. Q1 beat Q5 on 52% of days. The spread was 6.9 points, and it was noisy.

The live book is the high-reinvestment half by construction. The 43% sleeve is the cleaner split of the idea. It is not an independent test of the live rule.

Where a compounded dollar ends up

Platforms · 82.4%
Healthcare
The rest

Other

-4.0

SaaS

+7.5

Healthcare

+3.7

Med-tech

-5.5

Percentage points versus an equal-weight Halal universe. “Other” stays large because the mega-caps are not tagged SaaS.

The same blocks

Start with the capital from the engine. Twenty-four units, still uncolored. This is the book before we ask what it actually owns.

The same blocks

Start with the capital from the engine. Twenty-four units, still uncolored. This is the book before we ask what it actually owns.

They sit in platforms

Technology is 44.6% and communication services 37.8%. Together that is 82.4% of average weight. The olive return has become a platform book.

Med-tech did not take the pile

SaaS is overweight by 7.5 points versus an equal-weight Halal universe. Med-tech is underweight by 5.5. The capital-light sleeve returned 27.6%, in line with SPUS.

Finding 3: Where the dollars sit

The book is a Halal mega-cap technology and platform portfolio.

Alpha vs SPY

10.0%

β 1.18 · R² 0.72

Alpha vs SPUS

8.1%

β 1.07 · R² 0.8

Technology plus communication services averaged 82.4% of weight. At the last rebalance Nvidia was 20.4%, Microsoft 19.0%, and the two Alphabet lines together about 28.7%. Versus SPUS, beta is 1.07 and R² is 0.80, with 8.1% annualized alpha left over. Versus SPY, beta is 1.18 and alpha is 10.0%.

Without a sector-neutral twin, that leftover cannot be split into the reinvestment sort versus even more Nvidia, Microsoft, Alphabet, and Meta than SPUS already owned.

Finding 4: Friction

A 10 bp cost stub is small next to the headline. Turnover and concentration are the real frictions.

Turnover, annualized

200%

One-way. The early ramp is in this number.

Cost drag at 10 bp

0.20%

Net CAGR 39.24%

Purification drag

0.00%

213 dividends while held. Tags can be missing.

Rebalances with holdings

Average names 47.6

One-way turnover annualizes to about 200%. The book ramps from a thin start and rebalances monthly across about 50 names. Costs cut CAGR from 39.57% to 39.24%. Purification on 213 dividends while held is about zero on the tags we have. There is no position cap, so three names can be half the book.

Slippage on smaller reinvestment picks could be wider than 10 bp. Missing interest-income tags are a gap, not evidence that purification is zero.

Robustness

The same rule, kept only inside SaaS, healthcare, and med-tech, returned 27.6% a year. SPUS returned 27.4%. The unrestricted compounder book returned 39.6%. Drawdown on the sleeve was -15.07%, close to the main book’s -15.33%.

SaaS was overweight. Med-tech was underweight, and many med-tech names never cleared 15% ROIC on these filings. Restricting the book to the capital-light tags does not improve on SPUS. If a later paper freezes the rule, uses a longer filing history, and a sector-capped twin still beats SPUS, the compounding claim has more to stand on. If it does not, the 40% CAGR in this sample is mostly the 2023–2024 platform tape.

Levers

The rule runs on a handful of named choices. Each row is marked data or assumption.

Lever	Value	Kind	Basis
ROIC	NOPAT / invested capital	data	EBIT after tax. Rate above 500% dropped
ROIC floor	> 15%	assumption	Fixed before the run. Not fit on a hold-out
Reinvestment	(CapEx − D&A + ΔNWC + R&D) / NOPAT	data	R&D counts, so software is not punished
Keep quantile	Top half of the high-ROIC pool	assumption	Positive reinvestment required first
Debt cap	< 30% of 24-month market cap	data	AAOIFI, plus cash and receivables screens
Weighting	Market cap	assumption	No name cap, no optimizer
Rebalance	Month-end	assumption	No intra-month stop or compliance check

What this does not capture

No full calendar year of 2022, no 2008, no rate-hike recession, and no hold-out after the rule was frozen.
Yahoo annual statements are short. The first metrics date is 30 November 2021. The first invested date is 31 October 2022.
Sector exclusions use the current S&P list, not the 2019 membership file.
GOOG and GOOGL are both weighted, so Alphabet is heavy versus a single-listing index.
No intra-month compliance monitor. A name that fails AAOIFI mid-month stays until the next rebalance.
ROIC quintiles are not monotonic. The capital-light sleeve does not beat SPUS. The original sector question is still open.

A cap-weighted book of AAOIFI-screened S&P 500 names with ROIC above 15% and high reinvestment beat SPY and SPUS from late 2022 through 2024. The reinvestment split is the part that matches the compounding idea. The book is still a concentrated Halal mega-cap platform portfolio, and the capital-light sleeve does not beat SPUS. The shape is worth another paper. It is not ready to promote to a live mandate.