Your Alpha Might Be Beta in Disguise

This is the uncomfortable truth that Fama-French regression reveals. An RIA posting 12% annual returns over five years sounds impressive -- until the factor decomposition shows that 10% of that came from a persistent small-cap tilt and value exposure that any passive factor ETF could have delivered for 15 basis points annually.

The Fama-French factor model was designed exactly for this purpose: to separate genuine skill-based alpha from systematic factor exposure. For RIAs managing $50M-$500M, applying this framework is no longer optional -- institutional clients expect it, and it is the most credible way to justify active management fees.

What the Fama-French Model Actually Measures

Eugene Fama and Kenneth French published their three-factor model in 1992. It extended the Capital Asset Pricing Model (CAPM) by adding two factors that explained return patterns CAPM could not account for. Their five-factor extension (2015) added two more. Each factor represents a systematic risk premium historically rewarded by markets:

The Five Factors

• Market (Mkt-RF) -- Excess return of the broad market over the risk-free rate. Every diversified portfolio has significant exposure here. This is the primary driver of most portfolio returns.
• Size (SMB -- Small Minus Big) -- Small-cap stocks have historically outperformed large-cap stocks. Positive SMB loading means your portfolio tilts smaller. Negative means large-cap concentration.
• Value (HML -- High Minus Low) -- Value stocks (high book-to-market ratio) have historically outperformed growth stocks. Positive HML loading means a value tilt. Negative indicates growth exposure.
• Profitability (RMW -- Robust Minus Weak) -- Companies with robust operating profitability outperform those with weak profitability. This factor captures the quality premium.
• Investment (CMA -- Conservative Minus Aggressive) -- Companies investing conservatively (low asset growth) outperform those investing aggressively. This captures the anti-growth-capex premium.

The intercept of the regression -- alpha -- is what remains after all five systematic factors are accounted for. If your alpha is positive and statistically significant (t-statistic above 2.0), you have genuine skill-based outperformance. If not, your fees need rethinking.

How to Run a Fama-French Regression

The factor data is publicly available, maintained by Kenneth French at Dartmouth, and updated monthly:

Data source: mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html

The regression equation for the five-factor model:

R_portfolio - R_f = alpha + B1(Mkt-RF) + B2(SMB) + B3(HML) + B4(RMW) + B5(CMA) + e

Where:

• R_portfolio = Your portfolio monthly return
• R_f = Risk-free rate (1-month T-bill)
• alpha = Intercept (Jensen alpha -- true skill)
• B1-B5 = Factor loadings (exposure to each systematic factor)

Run this regression on 36-60 months of monthly returns. Fewer than 36 observations produces unreliable t-statistics.

Interpreting Factor Loadings for Client Reporting

The factor loadings translate directly into plain-language client communication:

Three Practical Use Cases for RIAs

1. Fee Justification

When a client questions your 75bps annual fee, a Fama-French attribution report provides the answer. If your regression alpha (after accounting for all five factors) is +2.3% annualized with a t-statistic of 2.4, you have a defensible, quantitative argument that you are generating returns beyond what systematic exposure alone would predict.

2. Portfolio Construction Audits

Run Fama-French attribution quarterly. If your HML loading drifted from +0.2 to -0.4 over six months, you have unknowingly shifted from a value tilt to a growth tilt -- likely because recent growth outperformance caused drift in your holdings. Attribution makes this visible before it becomes a mandate violation.

3. Peer Comparison

Comparing your Sharpe ratio to a competitor is misleading if they run different factor exposures. A proper apples-to-apples comparison controls for factor loadings. Two RIAs with identical 5-year Sharpe ratios but different factor models are not equally skilled -- the one with lower factor beta and higher alpha is demonstrably more skilled.

Common Mistakes When Applying Fama-French

• Using daily instead of monthly returns -- Daily regressions introduce microstructure noise. The Fama-French factors are constructed using monthly data. Match the frequency.
• Ignoring the R-squared -- An R-squared of 0.95 means 95% of your variance is explained by these five factors. An R-squared of 0.40 means significant unexplained variance -- which may be genuine skill or exposure to unlisted factors (momentum, quality, etc.)
• Treating loadings as static -- Factor exposures drift with portfolio changes and market movements. Rolling 12-month windows reveal how your factor profile evolves over time.
• Skipping the benchmark -- Always run the same regression on your benchmark. If your benchmark has an HML loading of +0.3 and you have +0.35, you are not differentiated on value -- you are just tracking the benchmark with slightly more value exposure.

Connecting Fama-French to Daily Operations

The challenge for most RIAs is that Fama-French attribution is a backward-looking monthly exercise. What matters operationally is knowing your current factor exposures in real-time -- and flagging when they drift outside your mandate.

Daily factor model monitoring solves this. By running rolling Fama-French regressions on your live portfolio each morning, you catch factor drift before it compounds. This is the same workflow institutional quant funds use -- the difference is access. For RIAs building this capability, real-time portfolio risk attribution is now accessible without Bloomberg infrastructure.

Related Research

• Portfolio Risk Attribution: How AI Decomposes Factor Exposures in Real-Time
• Factor Model Monitoring: Daily Regime Detection for Small Funds
• Backtesting Factor Strategies: A Step-by-Step Guide for Independent RIAs
• How Autonomous AI Is Changing Quantitative Research in 2026

Want daily Fama-French factor attribution for your portfolio -- without the Python overhead? Get your free Quantscope brief