The cards separate the fund's live record from the longer factor history used for the backtest, then show model fit, cost, valuation, and risk-adjusted results.
Age
11.9 years
2014-Oct → 2026-Sep
Long live record
Backtest Basis
63.2 years
1963-Jul → 2026-Aug
Full long-run factor record
Adj. R²
0.841
Good model fit
AUM
US$0.5B
Mid-sized live fund
MER
0.29%
Low-cost active tilt
Dividend valuation
138
Rich
138
80 90 100 110 120+
1.08% Trailing 12-month yield
100 = median · below 100 inexpensive · above 100 rich Dividend valuation scale
Backtested Rel. Sharpe
0.68×
Weak risk-adjusted edge
R
0.5 1.0 1.3 1.8
Relative to a plain market benchmark. 1.0 = market-like risk-adjusted return; above 1.0 clears the market hurdle.
Smart-β Net
-0.81%
Negative value-add
S
-3% 0% 2.5% 4.5%
Annualized non-market factor contribution plus alpha. Above 0% means the smart-beta sleeve added value over plain market exposure.
Implementation: Core equity. No deterministic enhanced-income or derivative-income mandate detected.

What QVAL Is

QVAL (Alpha Architect US Quantitative Value ETF) from Alpha Architect is a US equity ETF with $0.5B in assets, regressed on the FF6 US over 142 months. The fund tilts toward size (smb) (β=0.40); tilts toward value (hml) (β=0.33); tilts toward profitability (rmw) (β=0.40). Backtested-simulation excess return is +5.73% annualized; the in-sample realized excess was +7.15%.

Why QVAL Ranks This Way

  • Backtested relative Sharpe is 0.68, below the market's 1.00 benchmark.
  • Realized relative Sharpe is 0.45 over the fund's 142-month live sample.
  • Strongest non-market tilts point toward size (smb) (β=0.40) and toward profitability (rmw) (β=0.40).
  • Dividend valuation is 38% richer than its historical median (138 on the 100 scale).
Best-fit classification
US Equity FF6 US Regression basis: USD Adj. R² 0.841 Fit window: 142M Backtest basis: 758M

This fund is classified as US Equity because that factor set produced the strongest accepted fit. Backtested figures use the winning model's full factor history (1963-Jul → 2026-Aug, 758 months), while the betas themselves are estimated over the fund-overlap window (142 months).

Factor return decomposition
Each bar = βi × factor return, geometrically annualized. Summary rows are computed from the combined monthly series they name.
Model: FF6 US
Sharpe — Backtested Sim.
0.326
Rel. to market: 0.677
Sharpe — Realized
0.338
Rel. to market: 0.452
How are these computed?

Currency basis: returns and factor contributions on this page are in USD. Fama-French US 6-factor set (Ken French data library, monthly, USD basis).

Classification winner: FF6 US with Adj. R² 0.841.

Factor context panel: each non-market row compares the current fund with stronger-fit public peers using FF6 US. Beta mode shows raw loadings, percentile mode ranks the fund within the positive-exposure subset, and the linked ticker points to the strongest peer in that factor.

Backtested Sim. applies QVAL's regression betas to the full factor history (1963-Jul → 2026-Aug, 758 months). Individual factor bars show separately annualized sleeves, while the subtotal rows are computed from the combined monthly non-market and smart-beta series directly. The Sharpe uses QVAL's realized in-sample volatility as the denominator — a deliberately conservative choice, since fixed-loading factor returns are smoother than what any real fund actually delivers.

Realized applies the same betas to the factor returns observed during QVAL's lifetime only (2014-Nov → 2026-Aug). The Sharpe is the fund's actual excess return over its actual volatility. Full methodology →

Factor exposure context
Each non-market row compares this fund with stronger-fit public peers using FF6 US. Toggle between raw factor beta and percentile rank within the same positive-exposure peer set.
Compared against 334 funds using FF6 US
Peer filter: Equity · AUM floor · 24M+ history · Adj. R² > 0.80 · |βmkt−1| < 0.25 · Non-Mkt > 0
Default view is percentile rank. Switch to β scale when you want the raw loading magnitude.
SMB
-0.30
0
0.24
0.47
0.71
0.95
0
20
40
60
80
90
100
0
2
4
6
8
10
12
β +0.40
51st pct
t +4.5
Current fund: 51st percentile of positive peers · Top decile begins at β +0.88 · Leader: AVSC
HML
-0.30
0
0.24
0.47
0.71
0.95
0
20
40
60
80
90
100
0
2
4
6
8
10
12
β +0.33
81st pct
t +4.1
Current fund: 81st percentile of positive peers · Top decile begins at β +0.41 · Leader: KBWB
RMW
-0.30
0
0.24
0.47
0.71
0.95
0
20
40
60
80
90
100
0
2
4
6
8
10
12
β +0.40
100th pct
t +4.2
Current fund: 100th percentile of positive peers · Top decile begins at β +0.27 · Leader: QVAL
CMA
-0.30
0
0.24
0.47
0.71
0.95
0
20
40
60
80
90
100
0
2
4
6
8
10
12
β +0.10
51st pct
t +0.8
Current fund: 51st percentile of positive peers · Top decile begins at β +0.23 · Leader: RWJ
UMD
-0.30
0
0.24
0.47
0.71
0.95
0
20
40
60
80
90
100
0
2
4
6
8
10
12
β -0.09
rank n/a
t -1.5
Current fund: non-positive loading · Top decile begins at β +0.19 · Leader: MTUM
Factor exposures
Factor β t-stat Backtested % Realized %
Size (SMB) 0.402 4.51 +0.81% -0.79%
Value (HML) 0.331 4.05 +1.11% -0.30%
Profitability (RMW) 0.397 4.22 +1.15% +0.87%
Investment (CMA) 0.096 0.80 +0.28% -0.08%
Momentum (UMD) -0.088 -1.50 -0.64% -0.24%
Non-Market Factor Contribution · combined non-market series +2.75% -0.66%
Alpha (intercept) — -1.38 -3.47%
Smart-Beta Net · combined non-market series + alpha -0.81% -4.11%
+ Mkt-RF contribution · market 1.077 20.93 +6.56% +12.33%
= Total excess return +5.73% +7.15%
+ Rf base (%) 4.44% → 10.17% total 1.98% → 9.13% total
Fund snapshot
Issuer
Alpha Architect
Asset class
Equity
Inception
2014-10-20
Trailing 12-month yield
1.08%
Div CAGR
9.19%
Dividend valuation
138 (100=median)
Embed this chart

Use this iframe URL on Discourse forums (e.g. Rational Reminder Community) or any site that allows iframe embeds. For X, Reddit, or Slack, paste the fund page URL directly — the social card image is auto-generated for this fund.

<iframe src="https://verifiedbeta.com/embed/us/QVAL/waterfall/" width="720" height="540" style="border:0;" loading="lazy"></iframe>

Matches share at least one verified factor exposure with QVAL. They are ranked by modeled factor tracking error: the annualized volatility of their factor-return difference, not a return gap or forecast. Both funds use the same 36-month fitting window; a longer factor-volatility check also limits each match to 3% annualized. How matching works.

FLQM
RMW Modeled factor tracking error 2.01%/yr · longer factor history 1.86%/yr · 36 months through 2026-08
Compare QVAL vs FLQM
87
Factor-similar
OUSM
RMW Modeled factor tracking error 2.70%/yr · longer factor history 2.66%/yr · 36 months through 2026-08
Compare QVAL vs OUSM
82
Factor-similar
DSTL
RMW Modeled factor tracking error 2.81%/yr · longer factor history 2.87%/yr · 36 months through 2026-08
Compare QVAL vs DSTL
81
Factor-similar
IMCV
HML Modeled factor tracking error 2.17%/yr · longer factor history 2.89%/yr · 36 months through 2026-08
Compare QVAL vs IMCV
81
Factor-similar
View full QVAL similarity cluster
All US Leaders US Universe Methodology

FAQ

What does this QVAL ETF page measure?

This page evaluates QVAL using factor regressions, relative Sharpe, valuation, and peer context. It is designed for ETF due diligence rather than quote tracking.

Why can Backtested Sim. and Realized differ?

Both views use the same estimated betas. The difference is the return window: Backtested Sim. applies them to the full factor history, while Realized applies them only during the fund's live sample.

Why might a high-return fund still look weak here?

A fund can post strong raw returns and still look unattractive on this screen if its non-market factor capture is weak, its implementation drag is high, or its volatility-adjusted return profile is weaker than the market.

How should US investors use this page?

Use it as a starting point for ETF selection: check whether the fund's live behavior, valuation, and peer context match its stated strategy.

Research disclaimer

Factor decomposition shows what QVAL's historical betas would have earned under each scenario — not a forecast, not investment advice. Backtested simulation assumes loadings stay fixed across the long sample, which is unrealistic for any real-world fund. Use this as one input to ETF due diligence, not a substitute for it.

Important context

VerifiedBeta publishes educational ETF research, not personalized investment advice, portfolio management, or security recommendations. Funds that screen well here can still be unsuitable for your objectives, taxes, liquidity needs, or constraints. Review fund documents, methodology assumptions, and your own circumstances before acting. See the full disclaimer.