The ABCs of Systematic and Factor Investing

Factor investing arrives with an intimidating vocabulary: alpha, beta, HML, SMB, RMW, CMA, momentum, quality, defensive, smart beta, and several hundred published anomalies behind them. The words do most of the gatekeeping. An alphabet seems like a fair way to take them apart.

Systematic investing means following explicit, repeatable rules instead of case-by-case manager judgment. Factor investing is one kind of systematic investing that deliberately overweights securities sharing a characteristic: low valuations, strong profitability, smaller size, positive price momentum. A factor strategy can be fully systematic and still be active. A fund that deviates from the market portfolio by formula is making choices about what to overweight, how often to trade, and which definition of a factor to use.

The conclusion here is deliberately unexciting. A low-cost, globally diversified, market-cap-weighted portfolio is a complete equity strategy and a sensible default. Factor tilts are optional active bets that have to survive costs, taxes, tracking error, and your own behavior before they help you.

Each letter below gives a definition, the caveat investors most often miss, and a link to the guide that treats the subject in depth.

A is for Alpha

Alpha is the return that survives after a model accounts for the risks and factor exposures it measures. It does not mean “the amount my fund beat the S&P 500 by.”

Suppose a small-value fund outruns the broad market. Some or all of that gap may be explained by its exposure to smaller and cheaper companies, which is factor beta available for a few basis points. Many products charge active prices for returns that a transparent factor model can reproduce. The reverse also happens: a well-built factor fund can show almost no measured alpha precisely because the model has correctly identified what it is doing, and that is a sign the fund works as advertised.

Factor investing tries to buy compensated exposures efficiently. Treat a marketing claim of persistent alpha with more suspicion than a claim of clean, cheap factor beta. Fama-French Factor Analysis walks through reading alpha, loadings, and t-statistics on a real portfolio.

B is for Beta

Beta measures how sensitive an investment is to a broader source of return. Market beta measures sensitivity to the market. Factor betas measure exposure to value, size, profitability, momentum, or another systematic pattern. A beta is a description of an exposure, and whether that exposure earns a premium is a separate question.

Market beta stays the dominant risk in almost every stock portfolio. Factor investing rearranges the stocks inside an equity allocation without removing the risk of owning equities. An investor who buys four factor funds usually still holds a portfolio whose month-to-month behavior is overwhelmingly explained by the market.

Beta also has its own anomaly. Frazzini and Pedersen argue that investors who cannot use leverage bid up high-beta assets to reach their return targets, which leaves low-beta assets with better risk-adjusted returns. Their betting-against-beta factor earns significant risk-adjusted returns across US equities, 20 global equity markets, Treasury bonds, corporate bonds, and futures, and it does badly precisely when funding constraints tighten.¹ A separate literature explains the same pattern through investor demand for lottery-like payoffs rather than leverage constraints.² The two mechanisms get conflated constantly, and they are different claims about why the effect exists.

C is for Costs

A premium you cannot keep is not a premium. The expense ratio is the visible cost. Bid-ask spreads, market impact, turnover, index-reconstitution trades, and taxes are the rest.

Novy-Marx and Velikov studied 23 well-known anomalies and found that transaction costs reduce both their profitability and their statistical significance. Most strategies with one-sided monthly turnover under 50% kept generating significant net spreads; few of the higher-turnover ones did.³ Their most effective single fix was a buy/hold spread, which sets a higher bar for buying a stock than for continuing to hold one already owned. Across those 23 anomalies it cut turnover by 41% and transaction costs by 42%.

So the number to work with is a plausible future premium minus fees, trading costs, taxes, implementation shortfall, and the behavioral mistakes you make along the way. The calculator below runs that subtraction on a tilt you are considering.

D is for Diversification

Diversification spreads risk across securities, countries, industries, and sources of return. Owning 500 companies is not automatically enough if those companies share valuations, economic sensitivities, and one country of domicile.

Factor diversification is regularly oversold. Factors are not independent engines that take turns outperforming on a schedule. Their correlations move, and several can disappoint at once. A century of data across six asset classes shows value, momentum, carry, and defensive premiums with meaningful time variation that does not line up neatly with macroeconomic risk.⁴

The workable version is a small number of economically distinct exposures, each broadly diversified internally. How Many Funds Do You Actually Need? and Conviction vs Diversification cover where the additions stop paying.

E is for Evidence

A backtest starts the investigation. A factor worth acting on generally has an economic or behavioral reason to exist, evidence across multiple periods and markets, stability under reasonable alternative definitions, results not driven by tiny illiquid stocks, a return that survives realistic costs, and support from data the original researchers never saw.

That list eliminates most of what gets published. It does not reduce to “nothing works.” Evidence quality is a spectrum, and the useful question about any given factor is where on that spectrum it sits. Is Factor Investing Dead? works through the strongest cases on both sides.

F is for Factor

A factor is a shared characteristic associated with differences in assets’ risks or expected returns. Three explanations compete for why a premium exists, and they are not mutually exclusive.

• Risk. The premium compensates for bearing something genuinely unpleasant that shows up at the worst times.
• Behavior. Investors make persistent, repeated mistakes, such as extrapolating past growth too far.
• Constraints. Leverage limits, short-selling costs, benchmark mandates, and taxes stop investors from arbitraging the pattern away.

You do not need to settle the academic argument. You do need a reason to expect the premium to persist now that everyone knows about it, because the constraint and behavior stories make very different predictions about what happens after publication. Compensated vs Uncompensated Risk covers which risks have historically paid.

G is for Global

Evidence is stronger when it travels. A pattern found in one country, in one sample, using one database deserves more skepticism than one that appears in independent markets.

The Fama-French datasets now cover North America, Europe, Japan, Asia Pacific ex Japan, Developed, Developed ex US, and Emerging markets, with each region’s factors built from that region’s own size, value, profitability, and investment sorts.⁵ Global replication does not prove a factor will keep working, since countries share investor behavior and common economic shocks. It does make it much harder to dismiss a result as an accident of one sample.

For portfolios, “global” also means not concentrating a factor tilt in a single country. A US value tilt and a global value tilt are different bets. The Case for Global Diversification has the evidence.

H is for HML

HML means high minus low. It is the Fama-French value factor: the average return on the two value portfolios minus the average return on the two growth portfolios, where portfolios are sorted on book-to-market equity.⁵

HML is an academic long-short portfolio built to isolate a source of return, which makes it a poor guide to what a long-only value ETF will earn. That gap runs through all of factor research: published factor returns assume both a long and a short leg, skip real-world frictions, and use construction rules no retail product follows exactly. Research factors are evidence about the world, and products are what you can actually buy. Fama-French HML Explained goes deeper.

I is for Investment

The investment factor is about how aggressively companies grow their assets. In the Fama-French five-factor model, CMA means conservative minus aggressive, comparing firms with modest asset growth against firms expanding quickly.⁵

The logic comes from the valuation equation. Fama and French showed that given a firm’s book-to-market ratio and expected profitability, higher expected investment implies lower expected returns.⁶ This says nothing about whether capital spending is good management. A company should invest when its projects create value. The factor asks whether the price already reflects an optimistic view of that growth.

Watch for double counting. Value and investment are related, and different multifactor models slice the same underlying return patterns differently. Fama and French themselves noted that once profitability and investment enter the model, HML becomes redundant for describing average returns in their sample.⁷

J is for Junk

A stock can be cheap for a good reason. Quality-minus-junk strategies favor profitable, financially sound, stable companies and avoid weak profitability, aggressive growth, and fragile balance sheets. Screening for quality keeps a value strategy from mechanically buying businesses whose fundamentals are collapsing.

The best-known illustration is Berkshire Hathaway. Frazzini, Kabiller, and Pedersen found Berkshire’s Sharpe ratio of 0.76 was higher than any other stock or mutual fund with a history longer than 30 years, and that its alpha becomes statistically insignificant once you control for exposure to betting-against-beta and quality-minus-junk. They estimate Buffett’s leverage at about 1.6-to-1 on average.⁸ Their reading is that the record reflects cheap, safe, quality stocks bought with patient, cheap leverage rather than luck or magic.

K is for Kenneth French

Nearly every factor claim you will read traces back to one free website. The Kenneth R. French Data Library at Dartmouth publishes the monthly and daily factor returns that academics and fund companies use: the US three-factor and five-factor series back to July 1926, momentum, international and emerging-market versions, and hundreds of sorted portfolios. The monthly US series currently runs through mid-2026.⁵

Knowing this is useful for two reasons. It means you can check a claim yourself rather than trusting a fund’s marketing chart. And it explains why so much factor research shares the same blind spots: a large share of the literature runs on one data library, with common construction choices, common breakpoints, and a common start date. Independent evidence is rarer than the volume of papers suggests.

L is for Low volatility

More excitement has not historically meant more reward. Low-beta and low-volatility strategies favor stocks that move less than the market, and the pattern shows up across markets and asset classes.¹

Novy-Marx offers a deflating explanation. High-volatility and high-beta stocks tilt strongly toward small, unprofitable, growth firms, and those tilts explain the poor performance of the most aggressive stocks and therefore the apparent abnormal performance of defensive strategies. He finds defensive performance is explained by controlling for size, profitability, and relative valuations, while the reverse is not true: value and profitability strategies cannot be explained by defensive equity.⁹

One popular criticism does not hold up. Asness, Frazzini, and Pedersen tested whether low-risk investing is really an industry bet and found an industry-neutral version performs better than the across-industry one.¹⁰ The premium is not an artifact of overweighting utilities. Individual low-volatility products are a separate matter, and unconstrained ones do run large sector concentrations and interest-rate sensitivity that you should look at before buying.

M is for Momentum

Stocks that have recently outperformed have tended, for a while, to keep outperforming. Jegadeesh and Titman documented that buying past winners and selling past losers generated significant positive returns over 3- to 12-month holding periods, with the six-month formation and six-month holding version earning about 12% a year on NYSE and AMEX stocks from 1965 to 1989.¹¹

Momentum is also the most temperamental of the major factors. It trades far more than slow-moving characteristics like value, and it breaks down violently when market leadership turns. Daniel and Moskowitz show that momentum crashes occur in panic states, following market declines and when volatility is high, and that they coincide with market rebounds.¹² The crashes are partly forecastable, which is what makes a dynamic version work better than a static one in their data.

Momentum often earns its place as an ingredient rather than a standalone sleeve: delaying the purchase of a cheap stock whose price is still falling, or delaying the sale of a value stock that has started working. Growth vs Momentum and Is the S&P 500 a Momentum Trade? cover the signal and the exposure you may already own.

N is for Net of tax

Turnover is a poor proxy for tax burden. This is the place where intuition most often fails, and the direction of the error surprises people.

Israel and Moskowitz examined after-tax returns for size, value, growth, and momentum. Despite momentum having roughly five times the turnover of value, the two faced similar tax rates. Momentum realizes substantial short-term losses that offset many of its capital gains and pushes the remainder into long-term gains, while value’s tax burden comes mostly from dividend income, which offers no such relief. They also found that trading to manage capital gains costs less tracking error than avoiding dividend income, so capital-gain-heavy styles like momentum benefit most from tax-aware implementation.¹³ (This is a working paper, not a peer-reviewed article.)

The practical consequence is that account location deserves as much thought as fund selection, and it does not always point where you expect. Tax-Aware Long-Short covers the aggressive end of this.

O is for Out-of-sample

The data that tell you most are the data the strategy’s designer never saw. A strategy discovered on US stocks from 1963 to 2000 should be tested after 2000, in other countries, and under alternative construction rules.

McLean and Pontiff studied 97 published return predictors and found portfolio returns 26% lower out of sample and 58% lower after publication. They treat the 26% out-of-sample decline as an upper bound on data-mining effects, which leaves about 32 percentage points attributable to investors trading on newly public information.¹⁴

Some decay is normal and does not invalidate a factor. A strategy that exists only inside its original backtest is a different case. Ask when, where, how, and after whose discovery a result held up.

P is for Profitability

Profitable companies have earned higher average returns than unprofitable ones. Novy-Marx found that gross profits-to-assets has roughly the same power as book-to-market for predicting the cross-section of average returns, and that profitable firms deliver those returns despite having lower book-to-markets and higher market capitalizations. Controlling for profitability also improved value strategies substantially, especially among the largest and most liquid stocks.¹⁵

Because profitable firms tend to be larger and more expensive than classic value stocks, profitability pairs naturally with value rather than duplicating it. The Fama-French version is RMW, robust minus weak operating profitability.⁵ Definitions vary more than the label suggests: gross profitability, operating profitability, cash-based operating profitability, and return on equity are related but distinct signals. RMW Explained and Cash-Based Operating Profitability cover which definition has held up.

Q is for Quality

Quality is a useful idea with a flexible definition. A high-quality company might show strong profitability, manageable leverage, stable earnings, conservative accounting, durable margins, or low share issuance. That flexibility helps portfolio construction and also makes backtests easy to overfit, because a provider can pick whichever metrics worked and keep a reassuring label.

Recent work suggests most of the label is redundant. Novy-Marx and Medhat argue that profitability subsumes quality investing, explaining both the strategies the industry markets and the factors academics use, and that it also prices defensive-equity strategies and much of the abnormal performance of alternative-value approaches.¹⁶ (A working paper; one author works at Dimensional and the other has consulted for them.)

Three questions separate a real quality strategy from a label: which exact variables define quality, does the fund buy quality at any price, and is the apparent premium just profitability, value, or low volatility wearing a different name?

R is for Replication

Two careful teams looked at the same literature and disagreed. Their disagreement is the current state of the field.

Hou, Xue, and Zhang rebuilt 452 anomalies with stricter standards, using NYSE breakpoints and value-weighted returns to limit the influence of microcaps. About 65% failed to clear a t-statistic of 1.96, and 82% failed under a multiple-testing hurdle of 2.78. Their conclusion was that capital markets are more efficient than previously recognized.¹⁷

Jensen, Kelly, and Pedersen built a Bayesian model of factor replication and reached a different verdict: the majority of factors replicate, they cluster into 13 themes, most of which matter in the tangency portfolio, and they work out of sample in a new dataset covering 93 countries.¹⁸ Their framework treats the sheer number of observed factors as evidence that strengthens rather than weakens the case.

Both papers are in top journals and neither is a fringe position. A reader who has only encountered one of them has an incomplete picture. Is Factor Investing Dead? works through what to do given the disagreement.

S is for Smart beta

Smart beta is a marketing term for rules-based portfolios that weight stocks by something other than market capitalization. It covers equal weight, fundamental weighting, dividend screens, single-factor tilts, and multifactor blends, sold at a fee between an index fund and an active manager.

The term is imprecise by design, and that is the problem with it. Two funds both described as smart beta can hold entirely different portfolios with opposite factor exposures. The phrase also implies the plain version is dumb, which prejudges the question of whether the deviation is worth paying for.

Useful questions to ask of anything wearing the label: what rule determines the weights, what factor exposure does that rule actually produce, what is the turnover, and what is the fee relative to a plain index fund holding the same asset class? The Passive Investing Myth covers how much active choice sits inside any index.

T is for Tracking error

Tracking error is the volatility of your return relative to a benchmark. Factor tilts create it by design. A value portfolio cannot deviate from the market only during the years value wins.

This makes tracking error a behavioral risk as much as a statistical one. Plenty of investors accept a factor allocation in theory and abandon it after watching a plain index fund beat them for five years, which converts a long-horizon strategy into a realized loss. The bigger the tilt, the stronger the pull.

Answer this before you buy: how much relative underperformance can you tolerate, for how long, before you change course? The best factor allocation is the one you can finance, explain, and keep, which is usually smaller than the one with the best backtest. RMW Explained includes a tracking-error calculator for sizing a tilt.

U is for Universe

The stocks a study includes can matter more than the signal it tests. Microcaps are roughly 60% of listed US companies by count and a few percent by value. Equal-weighting them, or using all-stock breakpoints instead of NYSE ones, hands enormous influence to companies that are expensive to trade and impossible to own at scale.

This single methodological choice drove one of the largest results in modern finance. When Hou, Xue, and Zhang controlled for microcaps using NYSE breakpoints and value-weighted returns, 65% of 452 anomalies stopped clearing conventional significance.¹⁷ The signals did not change. The universe did.

When you read that a strategy earned some impressive number, find out which stocks it held and how they were weighted before you believe it. The Case for Small-Cap Value covers where implementation quality matters most.

V is for Value

Value means paying a lower price relative to some measure of fundamentals. The denominator can be book value, earnings, cash flow, sales, dividends, or an estimated intrinsic value, and no single ratio captures cheapness cleanly.

Classic HML uses book-to-market equity, a choice that has drawn steady criticism as intangible assets have grown as a share of corporate value. Modern value strategies typically blend several metrics and compare companies within industries to avoid unintended sector bets.

Keep two things separate: a company with a low price, and a company whose price is low relative to plausible future cash flows. Value investing means the second one. Cheap securities can stay cheap for many years, and the ones that stay cheap forever are the reason quality screens exist. Fama-French HML Explained covers the measurement debate.

W is for Weighting

Choosing securities is half a strategy. Deciding how much of each to hold is the other half. A portfolio can weight by market capitalization, equally, by factor score, by fundamental measures, by inverse volatility, by risk contribution, or by an optimizer under constraints.

Each choice embeds assumptions. Cap weighting requires almost no trading and gives the largest positions to the companies the market values most. Equal weighting creates a permanent smaller-company tilt and demands constant rebalancing. Score weighting produces stronger factor exposure and more concentration. Optimization balances several objectives and becomes sensitive to estimation error in its inputs.

This is why two funds both targeting value can behave very differently. The weighting rule, sector constraints, turnover controls, and treatment of negative earnings are part of the investment thesis, not implementation trivia. Modern Portfolio Theory covers the optimizer case and its failure modes.

X is for eXposure

A fund’s label and its measured factor exposure are two different things. The gap has been quantified twice, and the results are unflattering.

Blitz ran multifactor regressions on 415 US equity ETFs with at least three years of history and found that for every factor there were funds with large positive and large negative exposure, so that in aggregate the factor exposures came out close to zero. What remained was plain market exposure.¹⁹ Lettau, Ludvigson, and Manoel found the same thing in holdings rather than returns: there are almost no high book-to-market funds, growth funds concentrate in low book-to-market stocks, and most funds labeled value hold a higher proportion of low book-to-market stocks than high ones.²⁰

Judge a fund by its holdings, its valuation and profitability characteristics, a factor regression on its returns, its sector and country weights, its turnover, and its published construction rules. Regressions are imperfect when exposures shift over time, and they still beat trusting a category name. AVUV vs BSVO vs DFSV compares three funds with similar labels and different portfolios.

Y is for Years

A factor premium is a long-run average that arrives on no schedule. Genuine premiums go through long stretches of weak or negative relative returns. If they paid off reliably every year, they would be arbitraged away and emotionally easy to hold, and they are neither.

The Fama-French factors had negative average returns across the 2010s, much as they did through the 1990s. Negative ten-year premium periods are common enough that a plan should assume one rather than treat it as evidence the strategy broke.

A long horizon gives a premium more chances to show up without guaranteeing it, and you get one lifetime rather than a thousand simulated ones. The horizon that governs a tilt is how long you can leave that exposure alone without being forced, financially or emotionally, to sell it, which is usually shorter than the years left until retirement.

Z is for Zoo

The factor zoo is the accumulated menagerie of published return predictors. John Cochrane coined the phrase in his 2011 presidential address to the American Finance Association, observing that the field now had a “zoo of new factors.”²¹

Harvey, Liu, and Zhu made the statistical case for skepticism. Given how many factors have been tested, the usual significance threshold no longer makes sense, and they argue a new factor should clear a t-statistic above 3.0. Their conclusion was that most claimed research findings in financial economics are likely false.²²

The zoo is an argument for selectivity rather than cynicism. The goal is not the largest number of statistically significant factors but a small set of exposures with strong and varied evidence, a sensible economic story, tolerable implementation costs, some complementarity, a fit with your tax situation, and a decent chance you will still hold them after a bad decade.

What should a DIY investor actually do?

For most people the sensible default stays a low-cost, globally diversified, market-cap-weighted portfolio. Factor investing is optional, and the test for whether you are ready to add one is whether you can finish this sentence:

I am deviating from the market by owning more __________ because the evidence for it is __________, the strategy costs roughly __________ a year, and I am prepared for it to trail my benchmark for __________.

If the blanks are hard to fill, the tilt is premature. For investors who do go ahead, a workable order of operations:

1. Protect the core. Start from broad global diversification instead of assembling a portfolio entirely out of narrow factor funds.
2. Prefer the well-evidenced themes. Value, profitability, and momentum carry stronger and more varied evidence than most of the zoo, with no guarantees attached.
3. Handle size carefully. Smaller companies can strengthen other characteristics, and indiscriminate microcap exposure is expensive to own.
4. Check the implementation. Read the holdings, methodology, turnover, costs, and measured factor exposure rather than the fund name.
5. Diversify the tilt. Avoid depending on one factor, one metric, one country, or one provider.
6. Size it for your behavior. A small allocation you keep beats a large one you abandon in year six.
7. Write the policy first. Set rebalancing rules and the conditions for changing strategy before returns start influencing the decision.

Systematic investing replaces forecasts with rules, which cuts down on impulsive decisions. Rules do not guarantee good outcomes, and badly chosen ones simply automate bad decisions at low cost.

Frequently Asked Questions

What is the difference between systematic and factor investing?

Systematic investing means following explicit, repeatable rules rather than case-by-case judgment. Factor investing is one type of systematic investing that deliberately tilts toward securities sharing a characteristic such as low valuation, high profitability, small size, or positive momentum. All factor investing is systematic; plenty of systematic strategies, such as a rules-based rebalancing policy on a plain index portfolio, involve no factor tilts at all.

What do SMB, HML, RMW, CMA, and MOM stand for?

SMB is small minus big (size), HML is high minus low book-to-market (value), RMW is robust minus weak operating profitability, CMA is conservative minus aggressive investment, and MOM is momentum. Each is an academic long-short portfolio published in the Kenneth French Data Library, not a fund you can buy.

Is smart beta the same as factor investing?

Smart beta is a marketing term for any rules-based portfolio weighted by something other than market capitalization, which includes factor funds but also equal-weight funds, dividend screens, and fundamental indexing. Two funds both labeled smart beta can have opposite factor exposures, so the label tells you little about what a fund holds.

Do factor premiums still exist?

The evidence is genuinely mixed. Hou, Xue, and Zhang found that 65% of 452 anomalies failed conventional significance once microcaps were controlled for, while Jensen, Kelly, and Pedersen found that most factors replicate and work across 93 countries. Both papers are in top journals. A reasonable reading is that a small number of well-evidenced factors probably retain a premium smaller than backtests suggest, and that most of the zoo was never real.

How long can a factor underperform?

A decade or longer. The Fama-French factors had negative average returns through the 2010s and again through much of the 1990s. Any plan built on a factor tilt should still work if the premium is zero for ten years.

Is a total market index fund enough?

Yes. A globally diversified, low-cost, market-cap-weighted portfolio is a complete equity strategy on its own. A factor tilt is an optional evidence-based addition for investors who can hold it through long stretches of underperformance.

Key Takeaways

• The market portfolio is the default. Factor tilts are optional active bets that must survive costs, taxes, tracking error, and your behavior.
• Research factors are not products. HML, RMW, and CMA are academic long-short portfolios with no fees, no taxes, and construction rules no ETF follows exactly.
• The label tells you almost nothing. Across 415 US equity ETFs, aggregate factor exposure came out near zero, and most funds labeled value hold more growth stocks than value stocks.
• Turnover is a poor proxy for tax cost. Momentum and value faced similar tax rates despite momentum trading five times as much.
• Serious researchers disagree. Two top-journal replication studies reached opposite conclusions, which argues for a small, cheap, diversified tilt rather than confidence in either direction.

Related Guides

• Is Factor Investing Dead?: the full evidence on both sides, and what to do about the disagreement.
• Fama-French Factor Analysis: how to read loadings, t-statistics, alpha, and R-squared on your own portfolio.
• Systematic Investing: the ladder from no process to a full rules-based portfolio policy.
• How Many Funds Do You Actually Need?: where added complexity stops paying for itself.
• Compensated vs Uncompensated Risk: which risks have historically paid you for taking them.

Sources

1. Andrea Frazzini and Lasse Heje Pedersen, “Betting Against Beta,” Journal of Financial Economics 111(1), 2014. doi.org.
2. Turan G. Bali, Nusret Cakici, and Robert F. Whitelaw, “Maxing Out: Stocks as Lotteries and the Cross-Section of Expected Returns,” Journal of Financial Economics 99(2), 2011 (the lottery-demand explanation, distinct from the leverage-constraint mechanism). doi.org.
3. Robert Novy-Marx and Mihail Velikov, “A Taxonomy of Anomalies and Their Trading Costs,” Review of Financial Studies 29(1), 2016 (23 anomalies; buy/hold spread cuts turnover 41% and costs 42%). doi.org.
4. Antti Ilmanen, Ronen Israel, Rachel Lee, Tobias J. Moskowitz, and Ashwin Thapar, “How Do Factor Premia Vary Over Time? A Century of Evidence,” Journal of Investment Management 19(4), 2021. ssrn.com.
5. Kenneth R. French, Data Library, Tuck School of Business at Dartmouth (factor definitions, regional coverage, and monthly series). dartmouth.edu.
6. Eugene F. Fama and Kenneth R. French, “Profitability, Investment and Average Returns,” Journal of Financial Economics 82(3), 2006. doi.org.
7. Eugene F. Fama and Kenneth R. French, “A Five-Factor Asset Pricing Model,” Journal of Financial Economics 116(1), 2015. doi.org.
8. Andrea Frazzini, David Kabiller, and Lasse Heje Pedersen, “Buffett’s Alpha,” Financial Analysts Journal 74(4), 2018. doi.org.
9. Robert Novy-Marx, “Understanding Defensive Equity,” NBER Working Paper 20591, 2014 (working paper; no peer-reviewed version). nber.org.
10. Clifford S. Asness, Andrea Frazzini, and Lasse Heje Pedersen, “Low-Risk Investing Without Industry Bets,” Financial Analysts Journal 70(4), 2014. doi.org.
11. Narasimhan Jegadeesh and Sheridan Titman, “Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency,” Journal of Finance 48(1), 1993. doi.org.
12. Kent Daniel and Tobias J. Moskowitz, “Momentum Crashes,” Journal of Financial Economics 122(2), 2016. doi.org.
13. Ronen Israel and Tobias J. Moskowitz, “How Tax Efficient Are Equity Styles?” working paper, 2012 (revised November 2020); no peer-reviewed version. ssrn.com.
14. R. David McLean and Jeffrey Pontiff, “Does Academic Research Destroy Stock Return Predictability?” Journal of Finance 71(1), 2016. doi.org.
15. Robert Novy-Marx, “The Other Side of Value: The Gross Profitability Premium,” Journal of Financial Economics 108(1), 2013. doi.org.
16. Robert Novy-Marx and Mamdouh Medhat, “Profitability Retrospective: What Have We Learned?” NBER Working Paper 33601, 2025 (working paper; Medhat is employed by Dimensional Fund Advisors). nber.org.
17. Kewei Hou, Chen Xue, and Lu Zhang, “Replicating Anomalies,” Review of Financial Studies 33(5), 2020. doi.org.
18. Theis Ingerslev Jensen, Bryan T. Kelly, and Lasse Heje Pedersen, “Is There a Replication Crisis in Finance?” Journal of Finance 78(5), 2023. doi.org.
19. David Blitz, “Are Exchange-Traded Funds Harvesting Factor Premiums?” Journal of Index Investing, 2017 (415 US equity ETFs, 2011–2015). ssrn.com.
20. Martin Lettau, Sydney C. Ludvigson, and Paulo Manoel, “Characteristics of Mutual Fund Portfolios: Where Are the Value Funds?” NBER Working Paper 25381, 2018 (revised 2021); working paper only. nber.org.
21. John H. Cochrane, “Presidential Address: Discount Rates,” Journal of Finance 66(4), 2011 (origin of the “zoo of new factors” phrase). doi.org.
22. Campbell R. Harvey, Yan Liu, and Heqing Zhu, “…and the Cross-Section of Expected Returns,” Review of Financial Studies 29(1), 2016. doi.org.

Disclosure: the author owns Avantis and DFA factor funds and treats the premium as uncertain possible upside rather than a planning assumption. Educational content, not individualized advice.