Alpha Architect ETFs: QVAL, QMOM, and the Case for Concentrated Factors

Most factor funds are built to be held. Alpha Architect’s flagships are built to be different, and the firm says so out loud: its own materials describe a willingness to accept “high active-share and concentration,” and one slide in its firm overview is titled “We are NOT for everyone.”¹ That is an unusual thing for an asset manager to put in a deck, and it is the right place to start.

QVAL, QMOM, IVAL and IMOM hold roughly 50 stocks each. A total-market index fund holds a few thousand. That single contrast drives everything else: the fees, the tracking error, the turnover, and the kind of investor who should own them. This guide works through what these funds do, how they compare with Avantis, Dimensional and Vanguard, and where I think each one belongs. It also measures the comparison rather than asserting it, using factor regressions you can reproduce.

The lineup solves several unrelated problems

The most common mistake in writing about this firm is treating “Alpha Architect ETFs” as one product category. As of August 19, 2026 the branded lineup held about $18.4 billion, and roughly $13.8 billion of that sat in a single fund, BOXX, which has nothing to do with value or momentum.²

A review that averaged across those rows would say nothing useful about any of them. This guide is about the four factor funds. BOXX has its own guide, including the unresolved tax question, and I have not repeated that analysis here.

One piece of housekeeping that is easy to get wrong: the adviser is Empowered Funds, LLC, doing business as EA Advisers, and Alpha Architect, LLC serves as sub-adviser. CAOS, BOXA and BOXX carry a second sub-adviser, Arin Risk Advisors. PINE Distributors replaced Quasar as distributor in November 2025.³

Worth noting for anyone tracking the firm: the fastest-growing part of the lineup is not the factor funds. AAUS, AAEQ, AAUA and AAUB launched between July 2025 and July 2026 and together hold roughly $2.24 billion, more than QVAL, QMOM, IVAL and IMOM combined. They exist largely to serve Section 351 conversions, which let an investor contribute an appreciated portfolio into a new ETF without triggering gain when the statutory tests are met.⁴

QVAL and IVAL buy cheapness on an enterprise-value basis

QVAL starts from the largest 1,500 US-listed stocks, screens out accounting red flags, and ranks the survivors on cheapness. The ranking centers on EBIT to total enterprise value; the prospectus adds that book-to-market, cash-flow-to-price and earnings-to-price also inform it. The cheapest 500 pass into a quality stage that looks at financial strength, and the resulting portfolio holds roughly 50 names at close to equal weight. The prospectus permits approximately 50 to 200 holdings; in practice the funds sit at the bottom of that range.³

Two details matter more than they look. The primary valuation metric is an enterprise multiple rather than price to book, so even with the secondary measures folded in, QVAL is buying a different definition of cheap than the one academic value research uses. And equal weighting describes what the portfolio looks like rather than a rule the prospectus commits to; observed position sizes run around 1.9% to 2.5%.

The quality stage is what separates this from a naive low-multiple screen. Cheap stocks are sometimes cheap because the business is deteriorating, and a value process with no defense against that ends up owning the ones that deserved to be cheap. The approach comes out of Quantitative Value, the 2012 book by Wesley Gray and Tobias Carlisle.⁵

QMOM and IMOM screen for how the gain arrived

The momentum funds start from the standard academic signal: the trailing twelve-month return excluding the most recent month. The convention grew out of Jegadeesh and Titman’s work documenting that past winners kept winning over intermediate horizons; their 1993 paper tested formation and holding periods from three to twelve months, and the twelve-minus-one form became the standard afterward.⁶ QMOM adds a pre-screen on six-month momentum, nine-month momentum and beta, then applies a quality stage that favors, in the prospectus’s words, “the most consistent positive returns, as opposed to short-lived success.”⁷

That screen has a specific academic motivation. Da, Gurun and Warachka found that momentum profits concentrate almost entirely in stocks whose past return arrived as a long series of small daily moves rather than a few large jumps. Holding the total formation-period return fixed, their continuous-information group returned 5.94% while the discrete-information group returned negative 2.07%, which they attribute to investors underreacting to information that arrives gradually.⁸ A stock that doubled on one acquisition headline and a stock that ground upward for a year look identical to a naive momentum sort and behave differently afterward.

The funds rebalance monthly. That is a change: through September 2023 they rebalanced quarterly, and in October 2023 Alpha Architect moved all four to monthly reconstitution and introduced a trade optimizer intended to weigh expected factor benefit against projected trading cost.⁹

There is an older change worth knowing too. From February 2017 through January 2022, all four funds were passively managed, tracking proprietary indexes the prospectus describes as constructed in a manner substantially similar to the current process.³ Any long-window statistic on these funds, including the regressions below, spans three implementation regimes.

The argument with Dimensional is about implementation

It would be convenient to say Alpha Architect thinks momentum works and Dimensional thinks it does not. That is not the disagreement. Dimensional publishes on momentum and uses it; its position is that momentum decays too quickly to anchor a long-term allocation, so the premium is better captured inside the trading process than inside a separate fund. Wes Crill’s framing: “While both simulated and real-world data suggest momentum may not be suitable as a driver of long-term asset allocations, we believe momentum considerations can be integrated in a cost-effective way to help inform daily portfolio management decisions.”¹⁰ Their prospectus says the same thing in registered language: the advisor may adjust a position “based on shorter-term considerations, such as a company’s price momentum and short-run reversals.”¹¹

Avantis takes a similar line. Its funds delay buying stocks with large negative recent returns and delay selling stocks with large positive ones, which harvests some of the effect without paying to chase it.

I checked whether either firm offers a dedicated momentum product. Across roughly 40 Avantis funds and all six Dimensional US registrants, there is not one momentum series on file with the SEC.¹² That absence is a verified fact. The reason for it is my inference from their published horizon argument: neither firm has ever published a document saying “here is why we do not run a momentum fund.”

So the live question is whether a high-intensity momentum signal can be harvested cleanly enough, after turnover and trading costs, to justify a dedicated fund. That question has a real literature and it does not point one way. Lesmond, Schill and Zhou argued the profits are “an illusion of profit opportunity when, in fact, none exists.”¹³ Korajczyk and Sadka rebutted that directly, estimating a break-even fund size above $5 billion and concluding that “transaction costs do not appear to fully explain the return persistence.”¹⁴ Frazzini, Israel and Moskowitz put the break-even far higher still, over $50 billion, using live institutional trade data rather than modeled spreads; all three authors are affiliated with AQR, which runs these strategies commercially.¹⁵ Novy-Marx and Velikov land in between: momentum survives costs at 0.68% per month net, but the costs eat roughly half the gross spread.¹⁶

Alpha Architect is taking one side of a genuine academic dispute, and the October 2023 move to monthly rebalancing with cost-aware optimization is the firm acting on exactly the criticism that matters most to its case.

What the regressions show

Claims about factor exposure are testable, so rather than take anyone at their word I ran six-factor regressions on each fund’s monthly excess returns. The model is Fama-French five-factor plus momentum; standard errors are heteroskedasticity-robust. US funds are measured against the US research factors and the international funds against the developed ex-US factors, so IVAL and IMOM are never scored against a US market they do not invest in.

The chart below covers October 2019 through June 2026, the window in which all seven US funds existed, so no fund is flattered by a luckier start date. That window spans the passive-index years, the active-quarterly stretch and the current monthly process, so it describes the funds’ realized history rather than a clean read on the current implementation.

Two things stand out, and only one of them is the story Alpha Architect’s marketing tells.

The momentum result is decisive. QMOM loads +0.48 on momentum and VFMO, Vanguard’s momentum fund, loads +0.37. Every Avantis and Dimensional fund in the sample sits between negative 0.01 and +0.02, and those near-zero estimates are precise. Measured over its own full history since 2016, QMOM loads +0.51 and IMOM +0.57. One caution on reading the table: the QMOM-versus-VFMO gap sits within sampling error at 81 months, so treat the two as different intensities of the same exposure rather than a ranking. The gap between either fund and zero does not. If you want a momentum tilt in your portfolio, the Avantis and Dimensional lineups do not contain one, and this is what that looks like in the data rather than in a brochure.

The value result cuts against the simple story. Over the common window QVAL loads +0.39 on HML while AVUV loads +0.53 and VFVA +0.52. A concentrated deep-value fund registering less academic value exposure than a diversified one looks wrong until you remember that HML is defined on book-to-market and QVAL’s ranking centers on an enterprise multiple. The two definitions disagree, most of all for companies whose value sits in intangibles that never reach the balance sheet. The definitional gap is probably the largest cause, though not the only one: AVUV works a small-cap universe where value sorts are sharper, and quality screens plus a 50-name portfolio pull loadings around too. QVAL is delivering a different exposure from AVUV’s, and the standard academic yardstick is a poor instrument for measuring it.

The column that does confirm the concentration claim is R-squared, the share of each fund’s return variance the six factors explain.

A six-factor model explains 99.9% of VTI’s variance and 99.4% of AVUS’s. It explains 85.6% of QVAL’s and 79.8% of QMOM’s, meaning roughly a fifth of QMOM’s month-to-month movement is something the six factors do not explain. R-squared cannot say what that residual is made of: it bundles stock-specific movement together with sector weights, any signal the model omits, and gaps between the fund’s definitions and the academic ones. For a portfolio of roughly 50 names, concentration is likely the largest contributor, and the reading agrees with the low benchmark overlap Alpha Architect advertises from holdings.

A note on alpha, because the number is easy to misuse

Regressions like these also produce an alpha, and for these funds it is negative. I am not going to lead with that number, because it is a property of the model at least as much as of the fund.

Consider IMOM. Against the six-factor model its alpha is negative 5.97% per year with a t-statistic of negative 3.0, which sounds damning. Drop the momentum factor from the model and the same fund over the same months produces a positive 1.17% alpha. Nothing about the fund changed. What changed is the question. A model containing the momentum factor asks whether IMOM added anything beyond a costless, long-short, zero-spread paper portfolio of the very thing IMOM is built to hold. That is a high bar rather than an impossible one; in this same dataset VFMO clears it, with a six-factor alpha of +1.0% per year over its full sample.

The negative number measures underperformance conditional on this particular model. Fees, spreads and trading costs plausibly contribute, and those costs are what the Lesmond and Korajczyk debate above is fought over. But the intercept also absorbs everything else that separates IMOM from the paper factor: a 50-stock long-only portfolio, quality screens, and a signal that is not the academic twelve-minus-one. The regression cannot split the cost share from the construction share, so the number is a flag worth understanding rather than a bill for implementation. The dataset behind this guide reports alpha under four nested models for every fund so you can see how much the choice of model is doing.

Four houses, four answers

Vanguard belongs in that table as more than the cheap-beta column. It runs VFMO, VFVA and VFMF as genuinely active factor funds, and its own prospectus language is blunt about what that entails: “there may be periods when momentum investing is out of favor, and during such periods, the performance of the Fund may suffer.”¹⁷

Purity and integration are closer than they look

The natural way to frame this comparison is that Alpha Architect gives you concentrated single-factor sleeves while Avantis and Dimensional blend signals together, and that you must pick a philosophy. That framing is mostly right, but it oversells the distinction in one specific way worth knowing.

Novy-Marx showed that for linearly signal-weighted portfolios, blending two signals inside one portfolio and holding two separate single-signal portfolios produce the same thing: the active returns of a 50/50 mix of value and momentum z-scores are “completely indistinguishable” from a 50/50 mix of the two standalone strategies.¹⁸ If that were the whole story, owning QVAL alongside QMOM would be equivalent to owning one integrated value-and-momentum fund, and the philosophical argument would dissolve.

It is not the whole story, because the equivalence holds for linear construction and these are not linear portfolios. Quantile sorts, long-only constraints, quality screens and a 50-name cap all break it, and that is exactly where the real disagreement lives. An integrated fund can hold a stock that is moderately cheap and moderately trending; two separate funds cannot express that, and must instead own the deepest value names and the strongest trending names separately. Those are different portfolios with different risks.

There is a genuine portfolio argument for pairing the two. Asness, Moskowitz and Pedersen found value and momentum returns strongly negatively correlated, averaging about negative 0.60 for their signal-weighted equity factors.¹⁹ That figure is specific to the construction and the asset class: their tertile spreads run closer to negative 0.45, and in fixed income the relationship weakens and under some definitions turns positive. As a statement about global equities it holds up, and it is a better reason to add QMOM to a value-tilted portfolio than to add a third value fund.

Costs are reasonable, and turnover deserves a closer look

QVAL and QMOM charge 0.28%; IVAL and IMOM charge 0.38%. Those fees have come down twice, from 0.39% and 0.49% at the start of 2024, then another basis point in February 2026.²⁰ For portfolios this differentiated, that is a fair price, and the direction of travel deserves credit.

Comparing 0.28% against VTI’s 0.03% and declaring VTI the winner misses the point. You are not buying the same exposure more expensively; you are buying a different exposure. The question is whether that exposure is one you want, and the 25 basis points is the price of asking.

Turnover is the more interesting number. For the fiscal year ended September 30, 2025, QVAL turned over 332% of its portfolio, IVAL 267%, QMOM 399% and IMOM 411%.³ Those are high, and Alpha Architect discloses that higher turnover can raise trading costs and realize gains. Three things temper it. The ETF structure’s in-kind creation and redemption mechanism handles a great deal of what would otherwise be a taxable event in a mutual fund, and so far it has: neither QVAL nor QMOM has paid a capital gains distribution in its published history through mid-2026, across 47 and 15 distributions respectively; every one was income.² The 2023 optimizer exists specifically to weigh factor benefit against trading cost. And for a signal that decays as fast as momentum, refusing to trade is not free either: holding a stock after the reason you bought it has gone is its own cost. High turnover is a consequence of the strategy rather than a defect in it, though it does mean implementation quality matters more here than in a fund that rebalances annually.

One practical note for anyone buying. Thirty-day median bid-ask spreads run about 0.08% on QVAL and 0.10% on QMOM, but 0.23% on IVAL and 0.22% on IMOM.² A spread is a one-time cost rather than an annual one, so for a long-horizon holder it is nothing like a 20-basis-point fee. It is still a reason to use limit orders on the international funds and avoid trading in the first and last minutes of the session.

Active share measures how different a portfolio is

Alpha Architect reports active share between 93.85% and 97.25% across the lineup, with QVAL at 97.13% against the S&P 500.² The firm uses the statistic to make a fair point: paying an active fee for a portfolio that mostly replicates an index is a bad deal, and these funds are unambiguously not that.

The statistic will not carry more weight than that. Frazzini, Friedman and Pomorski re-examined the evidence that active share predicts returns and found that once you control for benchmark, it does not; within individual benchmarks it is about as likely to correlate positively with performance as negatively.²¹ Differentiation is a precondition for outperformance rather than evidence of it. A 97% active share tells you a fund can beat its benchmark, not that it will, and Alpha Architect’s own disclosure says as much: “Active share is not a performance measurement.”² The reason to own QVAL is that you want its specific exposure, not that its active share is high.

Tracking error is a benchmark statement, and it still hurts

A fund that looks nothing like the S&P 500 is not automatically riskier in any economic sense. Tracking error measures deviation from a benchmark, and if your view is that the benchmark is not the best portfolio for capturing the premium you are after, then deviation is the entire point. Judging a concentrated value fund by how closely it resembles a cap-weighted index is judging it by the standard it was built to reject.

That is the theory. The practice is that tracking error is the thing that makes people sell at the bottom, and Alpha Architect has published the best illustration of this that I know of. In “Even God Would Get Fired as an Active Investor,” Wes Gray builds portfolios with perfect five-year foresight, buying the stocks he already knows will perform best, from 1927 to 2016. That portfolio compounds at 29.37% a year against the S&P 500’s 9.87%. It also endures a 75.94% drawdown, and on a rolling one-year basis the perfect long-short version is “often getting beaten by 50 percentage points or more.” Gray’s conclusion: “even God would get fired multiple times over.”²²

The severity of that drawdown is the less interesting half, since the market itself fell 84.59% over the same stretch. The relative shortfall is what would have driven an investor out. A strategy can be working perfectly, with literal foresight, and still look broken for years at a time. If a merely good strategy has to survive that kind of stretch in the hands of an investor who cannot see the future, the binding constraint is not the model. It is the person holding it.

When not to buy these funds

Alpha Architect has been unusually direct about this, which makes the section easy to write honestly. Gray, in a 2021 interview: “we don’t sell to everybody. Like we know we’re a boutique.” And more pointedly: “if you just do this other thing? Well, I’m like, well, then just go buy the Vanguard fund.” On the five-year relative performance chart, he notes it has “happened 10 times where you got fired.”²³

Skip these funds if any of the following is true.

• You check your portfolio against the S&P 500. If a three-year stretch of trailing the index by 20 points a year would make you sell, high tracking error will eventually take the decision out of your hands at the worst moment.
• This would be your core holding. A 50-stock portfolio is a sleeve. Building a retirement plan on one manager’s implementation of one signal concentrates risks that have nothing to do with whether the premium is real.
• You are not prepared to wait a decade or more. Value spent 2007 to mid-2020 in the deepest drawdown since 1963 on the standard academic measure.²⁴ The calendar length of your horizon matters less than whether you can watch a strategy trail for years without folding; a thirty-year horizon does not help if you capitulate in year three.
• You cannot say what job the fund is doing. If the answer is that the backtest looked good, that is not a portfolio decision.

There is also an honest scale consideration. IVAL and IMOM hold $223 million and $156 million respectively.² Small funds are likelier to be closed or reorganized than large ones, and in a taxable account a closure forces a realization on someone else’s schedule. That is a general observation about small funds rather than a prediction about these ones, which have run for over a decade. It argues for holding small, specialized positions in tax-advantaged accounts where you can.

The rest of the lineup, briefly

CAOS sells options and buys protection with the proceeds

CAOS is often described as a tail hedge, which undersells the mechanism. The fund sells S&P 500 index options to generate income and gain index exposure, allocating roughly 20% of capital to that exposure and 1% to 10% to protective option structures, with the remainder held in a collateral portfolio that may sit entirely in BOXX. Net exposure ranges from +120% to negative 40%. The manager targets a defined protection ratio and treats a tail event as a decline of more than 25% within a few months with implied volatility sustained above 50.²⁵

A history note that matters for anyone backtesting it: CAOS carries an August 2013 inception, but it existed as the Arin Large Cap Theta Fund, a mutual fund, until it was reorganized into an ETF in March 2023 and reverse split one-for-eight. A chart that runs the whole series without flagging that is showing you two different vehicles.

For most people accumulating assets, my answer is that more equities plus enough safe assets is easier to understand and easier to hold than a convex payoff structure. CAOS is a credible instrument for someone deliberately building a multi-strategy portfolio who can articulate why they need convexity rather than duration. It is not a better bond fund. Our tail-risk hedging guide works through that decision, including what CAOS did in 2020 versus 2022 and how it compares with trend following.

HIDE is a trend fund aimed at both tails of inflation

The name is the strategy: the Alpha Architect High Inflation and Deflation ETF runs a quantitative absolute-momentum and trend-following model, updated at least monthly, across three target asset classes: intermediate-term US Treasuries, real estate and commodities. It leans toward commodities and REITs when it reads inflation and toward Treasuries when it reads deflation, and it can sit in cash and cash equivalents when no trend is attractive.²⁶

At 0.29% after a fee waiver running to February 2027, it is far cheaper than most managed futures funds, and correspondingly narrower: three asset classes held long or not at all, rather than dozens of markets held long and short. Neither dominates the other. They are different instruments, and HIDE only becomes interesting once the equity and bond architecture underneath it is already sound.

AAVM is younger than its inception date suggests

AAVM holds the four flagships in one ticker, currently about 30% QVAL, 27% IVAL, 23% IMOM and 20% QMOM, and can shift those weights on relative momentum. Its 0.38% cost is almost entirely the fees of the funds it holds: the management fee is 0.05% and acquired fund fees are 0.33%.

The important caveat is the track record. Until January 31, 2025 this fund was VMOT, with a different objective that included a drawdown-minimization mandate. The supplement effecting the change removed the hedging, derivatives and short-sale risk disclosures entirely.²⁷ Any performance series running from the 2017 inception describes a strategy that no longer exists. Funds evolve and there is nothing improper here, but a backtest has to respect the discontinuity.

What I would do

If you do not specifically want factor risk, a broad index portfolio remains the right default. There is no obligation to harvest every documented premium, and VTI at 0.03% plus VXUS at 0.05% minimizes manager risk, factor risk and the behavioral risk that ends most factor strategies early. Note that VTI was renamed in July 2026 and now tracks a Morningstar total-market index, a rebranding of the CRSP index it already tracked; Vanguard says the change affects neither the objective nor how the fund is managed. VXUS was unaffected and still tracks FTSE.¹⁷

If you want factor exposure in the core, I prefer Avantis or Dimensional, and the regressions above are part of why. AVUV delivered a higher HML loading than QVAL with an R-squared of 0.977, so very little of its month-to-month movement was left for anything outside the six factors. When a strategy has to be held through a decade of underperformance, I would rather the thing being tested is the premium than one manager’s stock selection. That is a judgment about portfolio role rather than about signal quality.

If you want a deliberate standalone sleeve, these funds are good at the job they were built for. QMOM and IMOM are the strongest case in the lineup, because a long-only momentum tilt is genuinely unavailable from Avantis and Dimensional, and the measured loadings show the funds deliver it.

They are not the only dedicated momentum funds, though, so the choice is about intensity. In the US, VFMO charges 0.13% against QMOM’s 0.28% and holds hundreds of names against QMOM’s 50; the loadings above, +0.37 and +0.48, are those two intensities, priced.¹⁷ Internationally, iShares IMTM tracks the MSCI World ex USA Momentum Index with 301 holdings, a 0.30% fee, about $4.3 billion in assets and a 0.06% median spread, against IMOM’s roughly 50 stocks, 0.38% fee and 0.22% spread.²⁸ The reason to pick the Alpha Architect version is wanting the concentrated, quality-screened implementation and accepting the active risk that comes with it. Sizing is a personal call; 10% to 15% of equities is one sensible example of a sleeve, and the principle behind it is that the position should be small enough that its worst stretch cannot wreck the plan.

QVAL and IVAL are credible for an investor who wants enterprise-multiple deep value specifically. If what you want is maximum exposure to academic value, AVUV and VFVA delivered more of it over the same window at lower cost. Buy QVAL because you prefer how it defines cheap, not because you assume concentration mechanically produces more of the factor.

How Summitward helps

Every claim in this guide came from a regression you can run against your own holdings. If you want to know what your portfolio is actually exposed to, rather than what its fund names imply, the portfolio tools run the same style of factor decomposition on your positions.

Frequently asked questions

Is QVAL better than AVUV?

They are not substitutes. AVUV is a small-cap value fund holding 792 stocks that loaded +0.53 on HML and +0.90 on size over the window measured here. QVAL is a mid- and large-cap deep value fund holding roughly 50 stocks that loaded +0.39 on HML. If you want a core value tilt you can hold for decades, AVUV is the easier fund to own. If you want concentrated enterprise-multiple value as a deliberate sleeve, QVAL does something AVUV does not. For a like-for-like capitalization comparison, Avantis also runs AVLV, a US large-cap value fund at 0.15%; it launched in September 2021, too late to join the common window measured here.²⁹

Do Avantis or Dimensional offer a momentum ETF?

No. Neither firm has a momentum fund registered with the SEC. Both use momentum inside their trading process, delaying purchases of falling stocks and sales of rising ones, but neither sells a standalone momentum product. This is the clearest gap that QMOM and IMOM fill. Dedicated momentum funds do exist elsewhere, from Vanguard’s VFMO to the iShares MSCI momentum trackers MTUM and IMTM; QMOM and IMOM are the concentrated, high-intensity implementations of the idea.

Why is QVAL’s value loading lower than a diversified value fund’s?

Largely because HML is defined on book-to-market while QVAL’s ranking centers on EBIT to enterprise value, with book-to-market, cash-flow-to-price and earnings-to-price as secondary inputs. The two measures disagree, particularly for companies whose assets are intangible. Universe and construction contribute too: AVUV sorts a small-cap universe where value spreads are widest, and a 50-name portfolio moves loadings around. QVAL is buying a different definition of cheap rather than a weaker dose of the same one, and the academic factor is an imperfect instrument for measuring it.

Are these funds tax-efficient given the turnover?

The record so far says yes. Despite turnover above 300%, neither QVAL nor QMOM has paid a capital gains distribution in its published history through mid-2026; every distribution has been income.² The ETF creation and redemption mechanism, which lets a fund move low-basis shares out in kind rather than selling them, has absorbed the realizations. Past distributions do not guarantee future ones, the prospectus warns that high turnover can realize short-term gains, and trading costs are real regardless of the wrapper. For small, specialized holdings I would still lean toward tax-advantaged accounts.

What happened to VMOT?

VMOT became AAVM on January 31, 2025, and the change was substantive. The old fund had a drawdown-minimization mandate with a hedging overlay; the current fund is a long-only holder of the four flagship factor ETFs. Performance data spanning the 2017 inception mixes two different strategies.

Should I buy AAUS instead of VTI?

For fresh cash, I would take VTI: 0.03%, enormous scale, and no explanation required. AAUS exists mainly to serve Section 351 conversions, where an investor with a large appreciated portfolio contributes it into a new ETF without triggering gain if the statutory diversification tests are met. That is a genuinely different reason to exist, and judging AAUS purely on expense ratio misses it.

Key takeaways

• The lineup is not one thing. QVAL and BOXX share a brand and nothing else. Evaluate each fund against the problem it was built to solve.
• Momentum is the clearest gap Alpha Architect fills. QMOM loaded +0.48 on momentum over the common window and IMOM +0.57 over its own full history, while every Avantis and Dimensional fund tested sat near zero. Neither firm registers a momentum fund.
• Concentration shows up in R-squared. A six-factor model explained 99.9% of VTI’s variance and 79.8% of QMOM’s over the windows measured here. That residual is consistent with what buying 50 stocks instead of thousands looks like, and it is a better signature of concentration than the HML loading is.
• Fees are fair and falling. 0.28% and 0.38% after two cuts is reasonable for portfolios this differentiated. Turnover of 332% and 399% is the number that deserves more attention than the fee.
• The binding constraint is behavioral. Alpha Architect’s own research shows a perfect-foresight portfolio trailing the market by 50 points or more in individual years. If that would make you sell, the strategy is wrong for you no matter how good the evidence is.

Related guides

• The HML Value Factor: how academic value differs from what value funds buy, which is why QVAL’s loading looks the way it does
• Growth vs. Momentum: why a valuation bet and a price-trend bet are different things, and how the main momentum ETFs compare
• The Rise of Avantis: how the other side of this comparison went from launch to $150 billion in seven years
• Is Factor Investing Dead?: how much of the premium survives costs, and how to size a tilt you can actually hold
• BOXX ETF Explained: the largest fund in the lineup, and the unresolved tax question around it

Sources and method

1. Alpha Architect, Firm Overview. Source for the “We are NOT for everyone” slide and the “willing to be different: high active-share and concentration” language. Alpha Architect
2. Alpha Architect fund pages, QVAL, QMOM, IVAL, IMOM, CAOS, HIDE, AAVM, BOXA. Source for expense ratios, assets under management, 30-day median bid-ask spreads, holdings counts, active share, standardized returns and the QVAL and QMOM distribution histories (income only, no capital gains rows, from inception through the June 2026 distributions), all retrieved August 19, 2026. Fund data changes daily; re-check before relying on any figure. funds.alphaarchitect.com
3. Alpha Architect, Prospectus, dated April 21, 2026 (the combined statutory prospectus covering the lineup). Source for the universe screen, the EBIT/TEV metric, the approximately 50 to 200 holdings range, fiscal-2025 portfolio turnover of 332% (QVAL), 267% (IVAL), 399% (QMOM) and 411% (IMOM), and the adviser and sub-adviser structure. The document names the adviser as “Empowered Funds, LLC dba EA Advisers” throughout; older secondary sources give a different trade name. ETF Architect
4. Alpha Architect, 351 Education Center, and the AAUS, AAEQ, AAUA and AAUB fund pages. Source for the Section 351 conversion business and the combined assets of the US Equity series. Summitward’s guide to capital gain planning covers the statute itself. Alpha Architect
5. Gray, Wesley R., and Tobias E. Carlisle, Quantitative Value, Wiley, 2012; and Gray, Wesley R., and Jack R. Vogel, Quantitative Momentum, Wiley, 2016. The research the QVAL and QMOM processes implement.
6. Jegadeesh, Narasimhan, and Sheridan Titman, “Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency,” Journal of Finance 48, no. 1 (1993): 65–91. The foundational momentum result; the six-month strategy realized a compounded excess return of 12.01% per year over 1965–1989. DOI
7. Alpha Architect, Prospectus, dated April 21, 2026, QMOM and IMOM sections. Source for the twelve-month-minus-one-month signal, the six-month, nine-month and beta pre-screens, and the “consistent positive returns, as opposed to short-lived success” language quoted above. ETF Architect
8. Da, Zhi, Umit G. Gurun, and Mitch Warachka, “Frog in the Pan: Continuous Information and Momentum,” Review of Financial Studies 27, no. 7 (2014): 2171–2218. Source for the 5.94% versus negative 2.07% split between continuous and discrete information portfolios. DOI
9. Alpha Architect, Investment Process Change, press release effective October 1, 2023. Source for the move from quarterly to monthly rebalancing and the upgraded trade optimizer across QVAL, QMOM, IVAL and IMOM. Alpha Architect
10. Crill, Wes, “Myth-Busting with Momentum: How to Pursue the Premium,” Dimensional Fund Advisors, November 2, 2021. Source for Dimensional’s published position on using momentum at the trade rather than as an allocation. Dimensional
11. Dimensional ETF Trust, Form 485BPOS, filed May 27, 2026. Source for the registered language on adjusting exposure based on price momentum and short-run reversals. SEC EDGAR
12. SEC EDGAR series listings, American Century ETF Trust (CIK 0001710607) and the six Dimensional US registrants, retrieved August 19, 2026. Basis for the statement that neither firm has a momentum series on file. Avantis fund data, including the 792 and 1,911 holdings counts, is from the June 30, 2026 quarterly fact sheets.
13. Lesmond, David A., Michael J. Schill, and Chunsheng Zhou, “The illusory nature of momentum profits,” Journal of Financial Economics 71, no. 2 (2004): 349–380. DOI
14. Korajczyk, Robert A., and Ronnie Sadka, “Are Momentum Profits Robust to Trading Costs?” Journal of Finance 59, no. 3 (2004): 1039–1082. The direct rebuttal to Lesmond et al., estimating break-even fund size above $5 billion. DOI
15. Frazzini, Andrea, Ronen Israel, and Tobias J. Moskowitz, “Trading Costs of Asset Pricing Anomalies,” working paper, 2015 draft (SSRN 2294498). Estimates a break-even fund size for momentum above $50 billion using live institutional trade data. All three authors are affiliated with AQR, which manages momentum strategies commercially; the paper has not been published in a peer-reviewed journal.
16. Novy-Marx, Robert, and Mihail Velikov, “A Taxonomy of Anomalies and Their Trading Costs,” Review of Financial Studies 29, no. 1 (2016): 104–147. Momentum nets 0.68% per month after costs, with costs consuming roughly half the gross spread. DOI
17. Vanguard, fund fact sheets for VTI, VXUS, VFMO, VFVA and VFMF as of June 30, 2026; the VFMO prospectus filed March 27, 2026 (total annual operating expenses of 0.13%); and the July 29, 2026 index and name changes for CRSP-tracking US equity funds. VTI now tracks a Morningstar total market index; Vanguard’s April 29, 2026 press release states the name changes “will not affect the funds’ investment objectives or how they are managed.” VXUS was not affected and still tracks FTSE Global All Cap ex US. Vanguard
18. Novy-Marx, Robert, “Testing strategies based on multiple signals,” NBER Working Paper 21329. Shows that for linearly signal-weighted portfolios, mixing signals within a portfolio and mixing single-signal portfolios produce indistinguishable active returns. The equivalence does not extend to quantile sorts or long-only constrained portfolios.
19. Asness, Clifford S., Tobias J. Moskowitz, and Lasse Heje Pedersen, “Value and Momentum Everywhere,” Journal of Finance 68, no. 3 (2013): 929–985. The negative 0.60 correlation figure is for signal-weighted equity factors; tertile spreads average closer to negative 0.45 and the relationship is weaker, and sometimes positive, in fixed income. DOI
20. Alpha Architect, fee reduction press release dated November 8, 2023, effective January 31, 2024, taking QVAL and QMOM from 0.39% to 0.29% and IVAL and IMOM from 0.49% to 0.39%. A further one-basis-point reduction took effect February 1, 2026. Alpha Architect
21. Frazzini, Andrea, Jacques Friedman, and Lukasz Pomorski, “Deactivating Active Share,” Financial Analysts Journal 72, no. 2 (2016). Finds that after controlling for benchmark, active share does not predict fund returns. AQR
22. Gray, Wesley R., “Even God Would Get Fired as an Active Investor,” Alpha Architect, results updated June 14, 2017. Perfect-foresight decile portfolios, 1927 to 2016, value-weighted and gross of costs. The best portfolio compounds at 29.37% against the S&P 500’s 9.87% and still suffers a 75.94% drawdown; note that the index itself fell 84.59% over the same window, so the relevant result is the relative shortfall rather than the absolute one. Alpha Architect
23. Gray, Wesley R., interviewed in “Concentrated Factor Investing,” AdvisorAnalyst, April 6, 2021. Quotations are from the published transcript. AdvisorAnalyst
24. Arnott, Robert D., Campbell R. Harvey, Vitali Kalesnik, and Juhani T. Linnainmaa, “Reports of Value’s Death May Be Greatly Exaggerated,” Financial Analysts Journal 77, no. 1 (2021): 44–67. Documents the 2007 to mid-2020 value drawdown as the deepest in the series since 1963. DOI
25. Alpha Architect Tail Risk ETF, Summary Prospectus. Source for the option-selling mechanism, the roughly 20% index exposure and 1% to 10% protective allocation, the +120% to negative 40% net exposure range, the protection ratio target, and the March 2023 reorganization from the Arin Large Cap Theta Fund with a one-for-eight reverse split. ETF Architect
26. Alpha Architect High Inflation and Deflation ETF (HIDE), Prospectus, dated April 21, 2026. Source for the absolute-momentum and trend-following model, the three target asset classes, the ability to hold cash equivalents, and the 0.29% net expense ratio after a fee waiver in effect until February 1, 2027. ETF Architect
27. EA Series Trust, Form 497 filed January 13, 2025, effective January 31, 2025. Changes the fund’s name from VMOT to AAVM, removes “while attempting to minimize market drawdowns” from the objective, and deletes the hedging, hedging model, derivatives and short sale risk disclosures. SEC EDGAR
28. iShares, MSCI Intl Momentum Factor ETF (IMTM) fund page. Source for the 0.30% expense ratio, $4.28 billion in net assets (August 19, 2026), 301 holdings and 0.06% 30-day median bid-ask spread (August 18, 2026), and the MSCI World ex USA Momentum Index. iShares
29. Avantis, US Large Cap Value ETF (AVLV) fund page. Source for the 0.15% expense ratio (as of January 1, 2026) and the September 21, 2021 inception date, retrieved August 19, 2026. Avantis
30. Method. Factor loadings are from six-factor regressions (Mkt-RF, SMB, HML, RMW, CMA and momentum) of each fund’s monthly excess return, using dividend-adjusted monthly closes from Yahoo Finance and the Kenneth French Data Library. US funds use the US research factors; IVAL, IMOM and AVDV use the developed ex-US factors. Standard errors are HC1 heteroskedasticity-robust and annualized alpha is twelve times the monthly intercept. The chart and the loadings table cover October 2019 through June 2026, the window in which all seven US funds existed; full-sample figures quoted in the text carry their own windows. Loadings on portfolios holding roughly 50 names have wide standard errors, so nothing here rests on a borderline significance call. Re-running the same regressions with Newey-West standard errors at 6 and 12 lags changes no conclusion. Comparisons between funds are point estimates: the gap between QMOM’s +0.48 and VFMO’s +0.37 on momentum, like the gap between QVAL’s +0.39 and AVUV’s +0.53 on HML, sits within sampling error at 81 monthly observations, while the gap between the momentum funds and zero does not. The window also spans three implementation regimes (passive index tracking through January 2022 per the prospectus, quarterly rebalancing through September 2023, monthly with a trade optimizer since), so the estimates describe each fund’s realized history rather than its current process alone. These describe a past window and are not forecasts. Reproducible with scripts/extract_alpha_architect_factors.py, which also reports alpha under four nested models for every fund. Fund-level figures were retrieved August 19, 2026 and change daily. Nothing here is investment, tax or legal advice.