Small-Cap Value Sleeve · Key Findings
The headline conclusions from screening the small-cap universe for halal-compliant, low-leverage value names — what the balance-sheet rule reveals, where the value premium actually lives, and what the live Fama-French regression confirmed about this sleeve's factor profile (a significant size tilt, a weaker-than-hoped value tilt, and a flat profitability loading).
Headline Conclusions
The single most consequential finding is how aggressively the balance-sheet rule prunes the field. Most small-cap companies carry meaningful leverage — they use debt to fund growth precisely because they lack the scale to self-finance. Requiring debt/equity below 30% removes the large majority of names before any other quality test is applied.
What survives is a specific kind of business: highly profitable, asset-light, or so cash-generative that it never needed to borrow. That is not a representative slice of small-cap value — it is a deliberately curated subset of financially fortress-like small companies.
~80% of universe excluded Survivors skew asset-lightConventional small-cap value indices are dominated by community banks, regional lenders, and thrifts — these trade at low price-to-book ratios and are the textbook "value" cohort. Every one is excluded by the lending screen.
This is the defining structural difference between this portfolio and a standard small-cap value index such as the Russell 2000 Value. The names that remain are concentrated in industrials, technology, healthcare, and consumer — and they are higher quality on average, but the sleeve will not track or behave like a conventional value index.
All financials excluded Industrials & tech dominate survivorsPowell Industries, NAPCO Security, and Exponent carry zero debt and net cash positions, yet post returns on equity near or above 28% — figures more typical of asset-light large caps than small-cap value names. They demonstrate that the balance-sheet rule, far from being a constraint that forces compromise, actually surfaces the highest-quality businesses in the universe.
Powell's 0% debt-to-equity with a 29.5% ROE and ~$389M net cash is the clearest example: a business that funds its own growth, can weather any downturn, and still trades at a value multiple.
POWL · 0% D/E · 29.5% ROE NSSC · $100M cash · 28% ROE EXPO · debt-free · ~28% ROEThe deepest-value small caps (lowest price-to-book) are overwhelmingly the financially weakest. The strongest balance sheets command higher multiples. The intersection — cheap AND debt-free — is narrow, which is exactly why this is a concentrated 12-name list rather than a broad index.
Where book value is low (asset-light consultancies like Exponent), the value characteristic shows up in earnings and cash-flow multiples rather than price-to-book. The screen accommodates both kinds of value, but each name has to demonstrate it somewhere.
Value + fortress balance sheet = rareApplying the Sloan earnings-quality test — comparing operating cash flow to reported net income — removes companies whose cheapness is built on low-quality, accrual-heavy earnings. A small cap trading at 8× earnings is not cheap if those earnings don't convert to cash.
NAPCO is the positive example: its earnings are flagged as high quality, backed by recurring SaaS cash flow with low accruals. The names that failed this test looked cheaper on a P/E basis but were excluded precisely because the cash-flow statement contradicted the income statement.
CFO / Net Income > 1.0 required Low accruals (Sloan factor)UFP Technologies sits at exactly 0.29 debt/equity — inside the ceiling, but only just, and after recent acquisitions. It is included with that flag fully visible. AAON, an excellent HVAC business, was excluded outright because its debt/equity rose to 0.42–0.88 after acquisition activity.
This is the rule working as intended: applied consistently even when it costs a genuinely good company. A screen that bends for attractive names is not a screen.
UFPT · 0.29 · included w/ flag AAON · 0.42–0.88 · excludedThe live Fama-French five-factor regression on the final 12 equal-weighted holdings (83 monthly observations) confirms the central thesis: SMB loads +0.43 with a t-statistic of 2.50 — strongly positive and statistically significant. The small-cap tilt this sleeve was built to capture is real and measurable.
The value tilt is directionally correct but weaker than hoped: HML +0.17 (t = 1.17) — positive, but not statistically significant. This is exactly the cost the screen warned about: excluding the entire halal-incompatible banking universe, where the value premium most strongly lives, dilutes the value loading. And RMW came in at +0.05 (t = 0.28) — essentially flat. The quality screens did not translate into a portfolio-level profitability tilt, because the per-stock RMW loadings cancel out (Powell, NAPCO, FormFactor and Celestica load strongly negative on profitability). The model explains 70% of returns (R² = 0.70), with a statistically significant annualised alpha of +12.1% (t = 2.32).
SMB: +0.43 (t=2.50) ✓ significant HML: +0.17 (t=1.17) directional RMW: +0.05 (t=0.28) flatFactor Profile · Actual Regression
These are the actual factor loadings from the five-factor regression on the final 12 equal-weighted holdings (83 monthly observations, January 2019 – December 2025). The contrast with the large-cap quality book is the reason this sleeve exists — and the regression confirms it on the factor that matters most for a small-cap sleeve.
"The regression delivered the size premium cleanly and significantly. The value tilt came through directionally but weak — the price of excluding the banking sector. And the quality screens, which work beautifully on the large-cap book, did not survive aggregation here. An honest result beats a convenient one."
The Evidence
Balance-sheet and profitability data verified from SEC filings and financial data providers. Sorted by balance-sheet strength. This is the data behind the findings above.
| Ticker | Company | Debt/Equity | ROE | Op Margin | Balance-Sheet Verdict |
|---|---|---|---|---|---|
| POWL | Powell Industries | 0% | 29.5% | ~22% | Debt-free · $389M net cash |
| NSSC | NAPCO Security | 0% | 28.0% | ~28% | Debt-free · $100M cash |
| EXPO | Exponent | ~0% | ~28% | ~24% | Debt-free · asset-light |
| IOSP | Innospec | ~3% | ~12% | ~9% | Near debt-free · net cash |
| FORM | FormFactor | ~4% | ~9% | ~10% | Minimal debt · cyclical |
| EXPD | Expeditors Intl | ~5% | ~35% | ~10% | Asset-light · net cash |
| PLAB | Photronics | ~7% | ~13% | ~22% | Low debt · net cash |
| SMPL | Simply Good Foods | ~25% | ~12% | ~17% | Verify · acquisition debt |
| MGRC | McGrath RentCorp | ~28% | ~17% | ~25% | Verify · fleet leverage |
| UFPT | UFP Technologies | 29% | 16.1% | ~15% | Borderline · just inside |
| CLS | Celestica (CA) | verify | ~25% | ~6% | SEDAR+ verification needed |
| CGY.TO | Calian Group (CA) | verify | ~12% | ~8% | SEDAR+ verification needed |
All figures approximate, drawn from most recent fiscal-year filings and data providers; re-verify against current filings before acting. Canadian names require SEDAR+ confirmation.
Strongest signal
The top seven names — Powell, NAPCO, Exponent, Innospec, FormFactor, Expeditors, Photronics — all carry less than 10% debt to equity, with several entirely debt-free. This is an unusually strong balance-sheet profile for a small-cap value cohort and the direct result of the binding rule.
Where to be cautious
UFPT (29%, borderline), SMPL and MGRC (mid-20s, acquisition/fleet debt), and the two Canadian names (CLS, CGY) require fresh filing verification. Small-cap leverage can change quickly between filings — the rule must be re-checked, not assumed.
Per-Stock Regression
Each holding regressed individually against the five factors. This is what reveals why the portfolio-level RMW washes out: the profitability loadings are genuinely split, with the debt-free names (POWL, NSSC, FORM, CLS) loading negative on RMW despite their strong balance sheets. SMB is positive for nearly every name — the source of the significant portfolio tilt.
| Ticker | Mkt | SMB | HML | RMW | CMA | Alpha % | R² |
|---|---|---|---|---|---|---|---|
| POWL | 0.63 | +0.39 | +0.60 | −0.33 | −0.91 | +44.6 | 0.09 |
| NSSC | 1.41 | +0.75 | −0.13 | −0.97 | −0.77 | +24.3 | 0.33 |
| EXPO | 0.50 | +0.73 | −0.34 | +0.45 | −0.12 | +0.4 | 0.24 |
| UFPT | 0.83 | +0.91 | −0.15 | +0.78 | +0.07 | +20.5 | 0.27 |
| EXPD | 0.87 | +0.13 | +0.14 | +0.91 | +0.08 | −2.6 | 0.50 |
| MGRC | 0.72 | +0.54 | +0.41 | +0.35 | −0.05 | +4.1 | 0.47 |
| IOSP | 0.86 | +0.97 | +0.49 | +0.44 | +0.07 | −5.9 | 0.65 |
| FORM | 1.15 | +0.24 | −0.01 | −0.65 | −0.69 | +13.6 | 0.28 |
| SMPL | 0.60 | +0.52 | −0.35 | +0.66 | +0.86 | −5.1 | 0.14 |
| PLAB | 0.99 | +0.80 | +0.73 | +0.38 | −0.54 | +12.9 | 0.24 |
| CLS | 2.48 | −0.88 | +0.80 | −1.32 | −0.26 | +34.6 | 0.44 |
| CGY.TO | 0.69 | +0.04 | −0.19 | −0.05 | +0.03 | +3.9 | 0.16 |
Equal-weighted portfolio, 83 monthly observations (Jan 2019 – Dec 2025). Alpha annualised. Loadings from individual OLS regressions against the Kenneth French five-factor series; t-statistics not shown per stock but most individual loadings are noisy given the short window and low single-name R². Celestica (CLS) behaves like a high-beta large cap, not a small-cap value name, and is the largest single drag on the portfolio's SMB and RMW.
On the Regression
The Fama-French five-factor regression was run on the final 12 equal-weighted holdings over 83 monthly observations (January 2019 – December 2025), using the Kenneth French factor series — the same method applied to the large-cap book. Three results matter, and only one of them is the clean confirmation the thesis hoped for.
The size tilt is real and significant. SMB loaded +0.43 with a t-statistic of 2.50, comfortably past the conventional significance threshold. For a sleeve whose entire reason to exist is to capture the small-cap premium the large-cap book gives up, this is the result that justifies the construction.
The value tilt is directional but unproven. HML loaded +0.17 (t = 1.17) — the right sign, but not statistically significant. This is the predicted cost of the halal screen made visible: by excluding the entire banking and financials universe, where the value premium has historically been strongest, the sleeve gives up much of its potential value loading. The page warned about this before the regression was run; the data confirmed it.
The quality tilt did not survive aggregation. RMW came in at +0.05 (t = 0.28) — effectively zero at the portfolio level. The per-stock loadings explain why: while Exponent, UFPT and Expeditors load positively on profitability, Powell, NAPCO, FormFactor and especially Celestica load strongly negative, and the equal-weighted average cancels out. A low debt/equity balance sheet does not guarantee a high RMW loading — leverage and the profitability factor are distinct properties, and this regression is a useful reminder of that. The model explained 70% of returns with a statistically significant +12.1% annualised alpha (t = 2.32), meaning a meaningful slice of the sleeve's return is not captured by the five factors at all.
The honest conclusion: this sleeve delivers the size premium as designed and complements the large-cap book on that dimension, but it is a small-cap sleeve more than a clean small-cap-value sleeve, and it should not be relied upon to add a profitability tilt. Finalised weights (rather than equal weighting) and the SEDAR+ verification of the Canadian names could shift these loadings, and the regression should be re-run whenever holdings change.
Not investment advice. This page summarises research findings for informational and educational purposes only. Factor loadings are from a regression on equal-weighted holdings over a specific window and will change with weights, holdings, and time period. Past factor performance does not guarantee future results. All financial data must be re-verified against primary filings before any investment decision. Small-cap stocks carry elevated liquidity and volatility risk. Consult a licensed financial advisor.