Imagine you own a used car lot, and you need to fill it. You have 500 spaces, and you are headed to an auction where 2,000 cars will cross the block. What is your buying strategy?

If you wanted to borrow a strategy from the investment world, you might “index,” buying a group of 500 cars based on how common each model is on the road. Lots of Camrys are on the road, so you buy a lot of Camrys. Fewer Volvo wagons, so you buy fewer of those. You do not skip anything, you do not haggle over any individual car, and you do not spend a minute forming an opinion about which car is a bargain. You just apply the rule.

Before dismissing that, notice it is not a terrible idea. If you bought 500 of the same car — 500 Priuses — you would be exposed to a single shock. A problem at Toyota. A shift in EV demand. A tax credit that disappears. Spreading your purchases across makes, models, and years protects you from any one of those. Diversification can be as useful a strategy on a car lot as it is in a stock portfolio.

Full disclosure: the S&P 500 Index is not purely composed of firms based on total market capitalization. Companies are selected by the S&P Dow Jones Indices committee based on additional strict rules around profitability, sector balance, plus committee discretion. In the car world, think of 500 of the most popular cars that also have a clean title, match the broader car market by car type (sedan, SUV, etc.) and are recommended by Consumer Reports. The popularity rule would be the optimal strategy under one specific condition: that no additional observable characteristics of a car are related to the return you will earn when you resell it. To be precise about terms, by expected return we mean this. You buy a car at auction for $10,000. You put it on your lot and sell it for $12,000. Your return on that purchase is 20%. If nothing else you can observe at the auction — color, options, mileage, ownership history, accident record, service records — tells you anything about that number, then there is no information to act on.

First, Give the Rule Its Due

The rule above is transparent. Anyone can audit it. It is objective — no judgment enters, so there is nothing to second-guess and nobody to blame. It is cheap to run, because you are not paying anyone to evaluate cars one at a time. It is broadly diversified and passes some quality and market-alignment checks, so no single model, manufacturer, or recall can take down your lot. And it barely turns over: as the mix of cars on the road drifts, your purchases drift with it, automatically.

That is similar to what a (predominantly) capitalization-weighted index does, and it is why indexing has been a very valuable innovation in the history of asset management. An index fund is not a random draw. It is a published, rules-based, objective procedure that holds essentially the whole market and weights each position by market capitalization. Low cost, high transparency, low turnover, tax efficiency, and real diversification are not small things. Any strategy that asks an investor to do something else has to clear that bar, and we hold ourselves to it. So the case for factor investing cannot be that indexing is careless. It plainly is not. The case has to be something narrower and more precise.

The Condition That Doesn’t Hold

Now change the assumption. Suppose you know, from years of watching what sells and at what price, that some characteristics are related to resale return. Low mileage. A single owner. A clean title. Complete service history.

Under this assumption, you might consider whether incorporating these additional characteristics could improve your return — tilting your lot toward the cars with characteristics that have historically been associated with stronger resale values, and away from those without them. You are not abandoning diversification — you still want 500 different cars. You are simply incorporating additional information that has historically been associated with differences in resale outcomes.

The analogy to equity investing is direct. Index investing is the first approach: buy the most popular group of 500 that also controls for some useful metrics, but not all. Think of the S&P 500, or any capitalization-weighted index. You get diversification, and you get it at a relatively low cost, but you are selecting based on a limited set of characteristics such as capitalization, profitability, sector representation — potentially resulting in different risk and return outcomes than approaches that incorporate a broader range of characteristics.

Factor investing is a different approach. It is tilting your portfolio toward stocks that have been ranked across 40+ characteristics that, across large swaths of historical data, have historically been associated with differences in subsequent returns. That reflects the quantitative investment approach behind CPIEX and CPAI.

And Cap-Weighting Is Itself a Factor

Here is the part that is easy to miss, and it matters more than the rest.

It is tempting to describe indexing as the neutral option — the portfolio with no tilts, against which every other portfolio is a deviation. But weighting by market capitalization is not the absence of a factor investment approach. It is a specific approach that emphasizes a relatively limited set of characteristics including size, profitability, sector and quality as determined by a committee. Among the companies that meet the index’s eligibility requirements, the largest companies by market capitalization receive the largest weights, and in a broad U.S. index the top handful of names can account for a substantial share of the portfolio.

So the framing is not “characteristics versus no characteristics.” Every portfolio is built on characteristics. The index has simply committed to a handful including — size, profitability, sector representation and committee-determined eligibility criteria, rather than explicitly incorporating a broader range of characteristics.

Which brings the whole debate down to a single empirical question: do stock characteristics predict returns?

Fifty Years of Evidence That Characteristics Matter

A large body of academic literature suggests certain stock characteristics have historically been associated with differences in long-term returns. However, researchers continue to debate the persistence, economic significance, and inevitability of these relationships.

The literature is deep, and it is not the product of one or two studies. A partial map:

• Momentum: Recent winners have tended to keep winning and recent losers to keep losing over intermediate horizons. Jegadeesh and Titman documented this in the Journal of Finance in 1993, and it has been replicated across markets, asset classes, and decades since.
  
• Value: Cheap stocks, measured on ratios like earnings-to-price and sales-to-price, have historically outperformed expensive ones. Basu published on the earnings-to-price effect in 1977. Fama and French formalized the cross-sectional evidence in 1992. Lakonishok, Shleifer, and Vishny took up the question of why in 1994, arguing the premium reflects investor overextrapolation rather than compensation for risk.
  
• Quality and profitability: Novy-Marx showed in 2013 that gross profitability predicts returns about as well as book-to-market does, and that the two are complementary rather than redundant. Fama and French folded profitability and investment into their five-factor model in 2015.
  
• Sentiment: Chan, Jegadeesh, and Lakonishok showed in 1996 that earnings-momentum signals — past earnings surprises and revisions to analyst estimates among them — predict subsequent returns beyond what past price momentum explains. The share lending market carries information too: Asquith, Pathak, and Ritter found in 2005 that heavily shorted stocks with constrained lendable supply subsequently underperformed, most pronounced among smaller names. Low short interest and low demand to borrow point in the other direction.
  
• Stability: Ang, Hodrick, Xing, and Zhang documented in 2006 that high-idiosyncratic-volatility stocks earned lower returns, not higher ones. Frazzini and Pedersen’s 2014 work on betting against beta reached a related conclusion.
  
• Long-term reversal: De Bondt and Thaler’s 1985 paper on market overreaction found that extreme long-horizon losers subsequently outperformed extreme long-horizon winners.

Six families of characteristics constructed from almost fifty years of data. Thousands of pages of peer review. An index fund draws on only a fraction of this evidence.

Not Every Characteristic Deserves a Tilt

We should be careful here, because the sheer volume of published research is itself a hazard. Harvey, Liu, and Zhu catalogued more than 300 factor candidates in the Review of Financial Studies in 2016 and argued that most of them fail once you adjust for how many hypotheses have been tested against the same data. Data mining is real.

Publication bias is real. That is an argument for discipline in choosing which characteristics to trust, not an argument for ignoring all of them. It is the same principle we apply to our machine learning inputs: a model is only as good as the variables you feed it. The characteristics worth tilting toward are the ones with a strong economic rationale, out-of-sample persistence, and evidence across regions and time periods — not the ones that happen to have fit well in a single backtest.

Back to the Car Lot

The dealer who consults the auction sheet is not abandoning discipline. He is not walking the rows on instinct, falling in love with a particular car, or trying to guess which model will be fashionable next spring. He is running a rule, exactly like the popularity rule — published, objective, repeatable, applied without exception across the whole auction.

His rule is simply built out of more inputs. Popularity still matters for diversification and liquidity. But so do mileage, title history, and prior owners, because those are the things that have historically been related to what he earns on resale. He buys across the whole auction, stays diversified, and tilts toward the cars whose characteristics point to better margins and away from the ones that do not.

Those are the basics of factor investing, and they are the foundation of CPIEX and CPAI. Not stock picking. Not forecasting the news. A rules-based portfolio, held broadly and systematically, weighted by the characteristics that decades of research have tied to expected returns.

Conclusion

Index investing is rules-based, transparent, and disciplined, and we would never argue otherwise. Our argument is narrower: a rule can be perfectly disciplined and still be limited in its inputs, because discipline describes how consistently you apply your inputs, not how much the inputs tell you.

Cap-weighting commits to a handful of characteristics — size, profitability, sector and committee-determined eligibility criteria — while leaving many other characteristics on the auction sheet. Factor investing seeks to retain many of the attributes associated with indexing, including breadth, rules-based implementation, relatively low turnover, and reduced reliance on subjective judgment, while incorporating additional characteristics that academic research has associated with differences in long-term risk and return expectations.

You still buy the whole auction. You just read the complete auction sheet first.