Value? What value?

There has long been an ambiguity at the heart of what we mean by value investing. Is it a method of evaluating companies? An investing factor? A style box?  

One thing current value investing is not is what Ben Graham outlined initially in the 1930s. When you go back and read the 1934 edition of Security Analysis, much of it, when translated into modern parlance, reads like a special situations handbook, with specific circumstances applying to each security. For some stated reason, say, the net asset value of Acme Widget is greater than its market value. The opportunity was linked to a stated line on the balance sheet or the income statement.

Graham’s approach was not a “style,” nor was it a factor. It was a company-specific circumstance, with an implied or articulated catalyst or resolution in the near or intermediate term. Graham did not even use the term value investing. For him it was just sensible security analysis versus outright speculation. The term was bestowed decades later and applied retroactively.

In 1984, the Frank Russell Company, an institutional and pension consulting firm, introduced the Russell 3000 and its large cap subset, the Russell 1000 benchmark. They did so because retirement plan and endowment sponsors were dissatisfied with the available benchmarks. Russell’s answer was a realistic, investable “whole market” universe of roughly 3,000 US stocks and a carve-out of the largest 1,000 as the natural arena for mainstream managers. In 1987, the Russell 1000 was sectioned into value and growth indices.

Splitting the main benchmark into two equal sleeves based on price to book—the cheaper half as value, the other half as growth—worked well enough at the time because a low price to book ratio was in the mid-1980s a tolerable shorthand for “cheap.” It offered a single metric, two sleeves, and a tidy 50/50 partition of the parent index by market cap. Importantly, at the time, the market was not 40% concentrated in a handful of individual stocks, so each of the two indices could have a roughly similar number of stocks to get to its respective 50% market cap of the overall Russell 1000 index. A 1995 methodological shift saw Russell add earnings and sales growth metrics to book value and also let part of a given company be assigned to one index and part to the other.

The Russell 1000 Value and Russell 1000 Growth (and their smaller-sized cousins) have become, in the subsequent decades, the leading measurement tools for institutional investors and their gatekeepers. Despite that great success, almost everything that has occurred since has lessened the utility of their definition of value and this mechanism of measuring it. In particular, the US economy’s migration from smokestacks to code necessarily diminished book value as a credible measure of growth, value, or much of anything else.

The mismatch between definition and calculation has become all too apparent in the last few years. In the summer 2025 reconstitution, three members of the Magnificent Seven entered the value index with a combined weight of more than five percent, while their combined weight in the growth sleeve declined only modestly. The rules were working as designed: according to the measurement metrics, these firms had slid away from their prior growth characterization, and the style score reclassified a slice of their capitalization as “value.” But one does not have to be a Graham disciple to see the disconnect between that particular logic and any intuitive notion of value investing, defining the value opportunity set, and measuring investment outcomes.

The just-completed 2026 reconstitution raised the stakes, with massive style shifts as mega caps ping ponged between the growth and value sleeves. Mega-cap tech stocks are now regularly pushed into the value bucket, not because a Graham style analyst would necessarily recognize them as bargains (though they might be!), but because the scoring algorithm must find enough “value” to fill a quota in a universe dominated (in terms of market cap) by a small number of market-leading tech stocks.

But while there remains some coherence in both the definition and popular understanding of the growth sleeve, the value half of the benchmark appears to be a mishmash of things that just don’t happen to make it into the other half. The result has been several decades of intellectual muddle and, not coincidentally, real-world investor frustration.

From my perspective, two changes above all are required. First, the styles deserve distinct treatment. A value benchmark should be structured around valuation, with a continuum running from cheap to expensive. A growth benchmark should be structured around business outlook, with a continuum running from stagnant to explosive. These are two different communities with unrelated goals. While it was once innovative to measure them with a single yardstick, that is no longer the case.

Second, the tidy solution of having 50% of the large cap universe be characterized as value and 50% as growth creates unnecessary chaos. Well-defined value and growth benchmarks shouldn’t have to split the market’s value evenly.

As we approach the centenary of Security Analysis, and forty years since professional investors were shoe-horned into a one-size-fits-all measurement system, it’s time to revisit how we define our goals and measure our acumen.