📊 Interpreting Factor Models

13.10. 📊 Interpreting Factor Models#

Understanding factor models requires distinguishing between the academic and the practitioner perspective. Both use the same tools—but with very different motivations and interpretations.

13.10.1. 🎓 The Academic View#

Factor models are formal theories of how risk is priced in the economy.

  • When you run a time-series regression (e.g., test if a factor explains asset returns), you are jointly testing:

    1. Whether the model captures all relevant priced risks

    2. Whether markets are efficient

13.10.1.1. What Does It Mean to Find Alpha?#

If you find a statistically significant alpha, then at least one of the following must be false:

  1. ✅ The model is correct

  2. ✅ Markets are efficient

But both might be wrong. So when you reject the model (find alpha), you have to decide:

Are markets inefficient?
Or does your model miss an important risk factor?

13.10.2. 🧭 What Should You Do After Rejecting the Model?#

Your decision depends on your belief:

13.10.2.1. 🧠 Case (2): Markets Are Inefficient#

  • The alpha is a mispricing—not free money, but a persistent anomaly due to behavioral or structural frictions.

  • You should exploit the anomaly—build a portfolio around it.

  • The higher the Appraisal Ratio (α / σ), the more this looks like “$100 on the sidewalk” (though not risk-free).

  • Adjust exposure based on:

    • Volatility

    • Correlation with your existing portfolio

This is an active investment decision rooted in your belief that markets get things wrong.

13.10.2.2. 🤔 Case (1): The Model Is Misspecified#

  • The alpha might be due to missing risk exposures.

  • You’re not sure if this is a genuine opportunity—or if you’re just exposed to unrecognized risk.

  • What to do:

    • Investigate!

    • Consider adding the anomaly as a factor if you think it reflects a real dimension of risk.

Examples of hidden risk sources:

  • Untraded wealth (housing, human capital)

  • Systemic risk (e.g. banking crises)

  • Firms that don’t yet exist (innovation risk)

  • Long-run macro exposures

Ask yourself: Is this a risk I’m willing to bear at this price?

13.10.2.3. 🔬 Post-Rejection Checklist#

After rejecting a model, perform a post-mortem:

  • Are you in Case (1) or (2)?

  • If (1), what kind of risk are you seeing?

  • If (2), why does this mispricing persist?

13.10.3. 💼 The Practitioner View#

Factor models are used as investment benchmarks, not tests of asset pricing theory.

  • Factors = well-known strategies (e.g., value, momentum, quality)

  • The goal: show value-add over benchmark exposures

Example:
If a client holds MKT + Value exposure, you want to show that your strategy has alpha beyond those. CAPM alpha only tells you that you outperform someone holding the market—but multifactor alpha tells you whether you outperform an investor holding other strategies, too.

13.10.3.1. Factors as Communication Tools#

  • Factors summarize how return and risk behave across the market.

  • They serve as common language between investors, managers, and allocators.

  • Good practitioners use factors to show:

    • Why a strategy works

    • What risks it carries

    • How it fits into a broader portfolio

13.10.4. 🧠 Questions a Good Practitioner Should Ask#

  1. What are the investor biases or frictions my strategy exploits?

  2. Why hasn’t this alpha been arbitraged away?

  3. How will I know if this trade is getting crowded?

  4. Do I have a plausible theory that explains why this works?

  5. Does the theory help guard against overfitting and data mining?

A clear economic narrative helps justify strategy persistence and protects against model error.

13.11. 🧠 Investors, Alpha, and Equilibrium Thinking#

Ultimately, investors must take a stand:

If a practitioner shows you a strategy with high alpha, do you believe this is
(1) Compensation for risk?
(2) A genuine mispricing?

If you believe (2), you’re assuming markets are not efficient, and that you can do better than the average investor. That’s a strong claim—and one you must justify with logic and evidence.

13.11.1. ⚖️ Using Equilibrium Thinking to Guard Against Overfitting#

Estimation risk is real—alphas that look impressive in-sample often disappoint out-of-sample. Equilibrium logic can help protect against such errors.

Even if your data says “this works,” equilibrium forces might say “this can’t scale.”

Why?

  • If everyone tried to invest in high-alpha strategies:

    • Prices would adjust: assets get more expensive

    • Expected returns would fall

    • The anomaly would shrink or vanish

So even if you find a “$100 bill,” ask yourself:

Can the world support everyone picking it up?

13.11.2. 🌍 What Is Market Equilibrium?#

At equilibrium:

  • All assets must be held.

  • All investors are optimizing given their preferences and beliefs.

  • Expected returns must adjust so portfolios clear.

This is the foundation of rational asset pricing:

  • Investors with different background risks and preferences allocate capital

  • Asset prices adjust so that every security is held and no investor wants to deviate

13.11.3. ❓ A Fundamental Asset Pricing Question#

If everyone wants higher returns, why do only some assets actually earn higher returns?

Why does the tangency portfolio put larger weight on some assets and smaller (or negative) weight on others?

13.11.4. 📉 The General Answer: Risk in Bad Times#

Assets earn higher expected returns not because they’re riskier in general, but because they:

  • Perform poorly in bad times

  • Are hard to hold when investors are most stressed

That’s what investors must be compensated for.

  • Assets that tank in recessions, financial crises, or other aggregate shocks require a risk premium

  • Assets that hedge those risks are in high demand and command lower expected returns

13.11.5. 🧩 The Role of Investor Preferences#

The definition of “bad times” depends on:

  • Consumption drops

  • Income volatility

  • Market drawdowns

  • Credit frictions

  • Labor market instability

  • Crashes in correlated wealth (e.g., housing, private businesses)

Different investors have different sensitivities to these conditions. Together, these preferences and exposures define the equilibrium.

13.11.6. 📐 CAPM as a Special Case#

The Capital Asset Pricing Model (CAPM) assumes:

  • “Bad times” = when the market portfolio performs poorly

  • Investors care only about mean and variance

  • Expected excess return is proportional to beta

But this is just one narrow version of a more general truth:

In any rational model, high returns must reflect exposure to risk that investors fear.

Let me know if you want a figure illustrating tangency portfolios or equilibrium clearing, or code for simulating Sharpe-optimal portfolios under different risk preferences.

13.12. 🧮 Portfolio Implications of the Equilibrium View#

Equilibrium thinking doesn’t just help explain expected returns—it also guides portfolio construction.

13.12.1. 🧑‍🤝‍🧑 If You’re Like the Average Investor…#

📌 Hold the market portfolio.

That’s the essence of the CAPM:
Everyone with identical risk aversion and no special circumstances should simply scale their position in the market up or down.

13.12.2. ⚖️ Deviate From the Market Only if You’re Different#

13.12.2.1. 1. Different Preferences#

  • More risk-averse than average?
    → Tilt toward safer assets (e.g., bonds)

  • More risk-tolerant than average?
    → Leverage the market (e.g., borrow and invest more)

13.12.2.2. 2. Background Risk#

Background risks = exposures that come from your life, job, or wealth outside the financial market.

Example:

  • You work in the automobile industry, with highly specialized human capital.

  • Your personal income already moves with that sector.

✅ Optimal response:

  • Avoid auto stocks in your portfolio

  • Still invest in the market—but hedge your background risk

In practice: Hold the market minus auto sector (or highly correlated names)

13.12.3. 🔁 This Logic Extends to Multifactor Models#

Even if CAPM fails, and multiple risk factors are needed to explain returns:

🧠 The market portfolio is still optimal for the average investor.

You should only tilt toward high-return factors if:

  • You’re less exposed to those risks than the average investor

  • Or, you have unique preferences that change how much risk you want to bear

13.12.4. 🎯 Intuition Behind High Factor Returns#

  • A factor (like value, momentum, or size) earns a premium because it does poorly in times that matter to the average investor.

  • If you’re less exposed to those risks, you can harvest that premium.

  • If you’re more exposed, you should underweight it.

✅ This links alpha to who you are, not just what you believe.

Contents