Factors — value, momentum, quality, size, low-vol

Factor investing organises the cross-section of equity returns around systematic exposures that have historically delivered premia. Rather than picking individual stocks, factor investors construct portfolios that are tilted toward stocks with the desired factor characteristics. The approach is a middle ground between pure passive (cap-weighted index) and active stock picking, and has become a substantial share of institutional equity allocation through factor ETFs and smart-beta strategies.

The Fama-French 3-factor model — the regression

Eugene Fama and Kenneth French's 1992-1993 papers introduced the three-factor model that became the academic standard for explaining the cross-section of stock returns. The model says that the excess return on any stock or portfolio over the risk-free rate can be decomposed into exposures to three systematic factors, each carrying its own market-priced risk premium, plus an idiosyncratic residual:

  R_i,t  −  R_f,t   =   α_i  +  β_M · ( R_M,t  −  R_f,t )
                                +  β_S ·  SMB_t
                                +  β_V ·  HML_t
                                +  ε_i,t

where:
  R_i,t          =  the return on stock or portfolio i in period t
  R_f,t          =  the risk-free rate in period t (1-month T-bill conventionally)
  R_i,t − R_f,t  =  the excess return on i — what we are trying to explain
  R_M,t − R_f,t  =  the market risk premium (the CAPM market factor)
                    — daily/monthly return on a value-weighted broad market
                       portfolio in excess of T-bills
  SMB_t          =  Small Minus Big — the return on a portfolio long small-cap
                    stocks and short large-cap stocks in period t
                    — captures the size premium
  HML_t          =  High Minus Low — the return on a portfolio long high-book-to-
                    market (value) stocks and short low-book-to-market (growth) stocks
                    — captures the value premium
  β_M, β_S, β_V  =  the stock or portfolio's loadings on each factor — estimated
                    as the OLS slope coefficients in a time-series regression
                    of excess returns on the three factor returns
                    — β_M of 1 means the stock has full market exposure;
                      β_S of +0.5 means a small-cap tilt; β_V of +0.3 means
                      a value tilt
  α_i            =  the regression intercept — the average excess return
                    NOT explained by the three factors; significant non-zero
                    α is what active managers claim and what factor models
                    seek to explain away
  ε_i,t          =  the residual return in period t — idiosyncratic noise
                    not captured by the factors

Why each factor matters in practice

• Market factor (R_M − R_f): the broad equity-market risk premium that the CAPM had already identified. Carries a historical premium of ~5-7% annualised in developed markets. Loadings (β_M) above 1 indicate aggressive market exposure; below 1 indicates defensive.
• SMB (size factor): historically small-cap stocks have outperformed large-cap stocks by ~2-3% annualised in the US, less reliably in international samples. The premium has shrunk since the 1990s. β_S > 0 indicates a small-cap tilt.
• HML (value factor): high book-to-market (cheap) stocks have outperformed low book-to-market (expensive) stocks by ~3-4% annualised long-run, with severe drawdowns in 2018-2020 followed by recovery in 2021-22. β_V > 0 indicates a value tilt.
• α (alpha): the part of average return that the three factors do not explain. A significant positive α is the academic test of 'genuine' active-manager skill. Most fund managers' α evaporates once exposures to size and value are controlled for — which is the entire point of factor models and the reason factor ETFs have grown so much.

The 3-factor model explained cross-sectional returns far better than the single-factor CAPM, and Fama won the 2013 Nobel Prize in Economics in part for this work. The model re-shaped both academic finance and the asset-management industry — it gave investors a tool to decompose any manager's track record into factor exposures and 'true' alpha, which dramatically reduced the bargaining power of any active manager whose returns turned out to be explained by simple factor tilts.

The 5-factor model and beyond

Fama and French expanded to a 5-factor model in 2015, adding profitability (RMW, robust-minus-weak operating profitability) and investment (CMA, conservative-minus-aggressive asset growth). Beyond Fama-French, academic researchers have proposed hundreds of factors. The 2018 paper 'Replicating Anomalies' (Hou, Xue, Zhang) tested 452 published factors and found that only about half replicated convincingly. The 'factor zoo' became a polite term for the proliferation of data-mined results.

Momentum

Jegadeesh and Titman's 1993 paper documented that stocks that have outperformed over the past 6-12 months tend to continue outperforming over the next 3-12 months. Momentum is the most persistent and largest single 'anomaly' in equity returns documented across decades and across global markets. It is also expensive to implement (high turnover) and prone to occasional severe drawdowns ('momentum crashes' like 2009). AQR Capital was built largely on factor investing with momentum as a centrepiece.

Sharpe, Treynor, and Information Ratio — the risk-adjusted return metrics

Comparing portfolio managers on raw returns alone is meaningless — a manager who delivered 20% by taking on twice the market's volatility has not added the same value as a manager who delivered 12% with half the volatility. Risk-adjusted return metrics normalise returns by some measure of risk, allowing apples-to-apples comparison. Three are most-used: Sharpe, Treynor, and Information Ratio.

                          E[ R_p ]  −  R_f
  Sharpe ratio       =  ───────────────────
                              σ_p

                          E[ R_p ]  −  R_f
  Treynor ratio      =  ───────────────────
                              β_p

                          E[ R_p ]  −  E[ R_b ]
  Information ratio  =  ──────────────────────────
                              σ( R_p  −  R_b )

where:
  E[R_p]  =  expected (or realised average) return on portfolio p
  R_f     =  the risk-free rate
  σ_p     =  the standard deviation of portfolio returns (total volatility,
              capturing both systematic and idiosyncratic risk)
  β_p     =  the portfolio's beta to the market (systematic-only risk
              measure; ignores idiosyncratic risk that diversification
              should have removed)
  E[R_b]  =  expected return on the benchmark (typically the manager's
              stated benchmark, e.g. S&P 500, MSCI ACWI)
  σ(R_p − R_b)  =  the standard deviation of the active return (portfolio
                   minus benchmark) — the 'tracking error'
  Numerator in each case  =  excess return over some reference
                              (risk-free in Sharpe and Treynor;
                              benchmark in IR)
  Denominator in each case  =  the risk measure normalising the excess
                                return — different choice = different question

• Sharpe ratio — the most-used metric. Asks 'how much excess return per unit of total volatility?' Higher is better. A long-run Sharpe above 1.0 is excellent, 0.5-1.0 is solid, below 0.5 is mediocre. The S&P 500's long-run Sharpe is around 0.4-0.6 depending on the window. Hedge-fund track records are routinely cherry-picked to inflate reported Sharpe.
• Treynor ratio — same numerator as Sharpe, but divides by beta instead of total volatility. Asks 'how much excess return per unit of market exposure?' Useful when comparing well-diversified portfolios where idiosyncratic risk has been diversified away. Less applicable to concentrated portfolios.
• Information Ratio — for active managers benchmarked against an index. Asks 'how much excess return over the benchmark per unit of tracking error?' A skilled active manager should have IR > 0.5; very few sustain IR > 1.0 over a decade.

Quality and low-volatility factors

Quality factors capture profitable, low-debt, stable companies. They produce higher Sharpe ratios than the market but lower absolute returns — the trade-off appeals to risk-averse institutions. Low-volatility (Frazzini-Pedersen 'betting against beta') captures the empirical regularity that low-beta stocks deliver better risk-adjusted returns than CAPM would predict. Both are now broadly available through factor ETFs from BlackRock, Vanguard, Invesco, and others.

The replication crisis in factor research

Beyond the headline factors (size, value, momentum, quality, low-vol), the academic literature has produced an enormous catalogue of supposedly profitable factors. Critics argue many results are data-mined: when researchers run thousands of strategies on the same historical data, some will look profitable by chance. The replication research of the late 2010s found many published factor returns shrink dramatically out-of-sample. The takeaway: economic mechanism matters; a factor without theoretical foundation should be treated sceptically.

Smart beta and factor ETFs

Factor strategies have been productised into ETFs marketed as 'smart beta' — between passive cap-weighted indices and discretionary active management. By the mid-2020s, smart-beta ETFs hold hundreds of billions of dollars across providers. The challenge: the marketing has often outrun the substance. Some ETFs marketed as factor exposure have weak factor purity, high turnover, or both. Selection matters.

Factor investing in African equities

Factor investing in frontier and small EM markets is constrained by data quality, liquidity, and the small universe of investable names. Some studies have documented value and momentum premia on African equities; others find the signals are weak or unstable. For most African allocators, factor investing remains a developed-market and large EM tool rather than a daily practice.