Separate signal from price: trading factors and fundamental growth

1. A cleaner way to time factors: subtract revaluation

TLDR

• Most of a factor’s long-run return comes from its structural premium, not from its valuation getting richer.

• The “revaluation” part is noisy and averages near 0 over long spans. 

• Sorting factors by their structural premium (past return net of revaluation) improves timing, especially over 12-24m horizons.

What this means: A factor is a shared stock trait tied to returns, and its lasting gains come from the trait itself, not from investors just paying higher and higher prices for it. (BlackRock)

Some background to explain Factors and Revaluation

What is a factor

A “factor” is a measurable trait that many stocks share that has been linked to different average returns. (BlackRock)

Common examples are: value, size, momentum, quality, and low volatility. In practice, if you wanted to trade a factor, you would take a universe of stocks, rank them by that factor (e.g., by high volatility to low volatility, or high valuation multiple to low multiple), buy one end of the range and short the other end.

Over long periods, the payoff you get from holding a factor mostly comes from its underlying economic edge (the “structural premium”), not from investors bidding up that factor so it gets more expensive (the “revaluation” part).

In the research that coined this wording, “revaluation” is the return due to changes in the factor’s valuation multiple, while the “structural” piece is what is left after stripping that out. The authors find that revaluation averages near 0 over time, so you shouldn’t count on it.

Take the value factor. Your long-run return mostly comes from owning higher cash flow yields and from fundamentals improving relative to price. That is the structural premium. You might also get short bursts when value becomes fashionable and valuation gaps close, but that “multiple expansion” tends to mean-revert and cannot power long-run results by itself. The same logic applies to other factors.

What the authors did

They decomposed each factor’s historical return into: 

• Revaluation, driven by changes in the factor’s valuation ratio, and 

• Structural, the residual that should proxy for the true long-run premium

They applied this to 14 well-known long-short factors from July 1973 to December 2022. In their sample, the “average factor” earns 4.0% per year, of which -0.4% per year is attributed to revaluation and 4.4% per year is structural.

What they showed

1. The revaluation piece is small on average and not statistically different from 0 across individual factors, which matches the idea that valuations mean-revert over time.
   

2. In cross-sectional forecasts, structural premium predicts future factor returns out to 24 months and outperforms momentum and relative valuation signals as the horizon gets longer.
   

3. Takeaway: a factor’s strong backtest that came from getting “more expensive” should not anchor forward expectations; the structural piece travels better into the future.

Why it matters & how to use it

When you implement a factor portfolio or assess a factor fund, ask how much of the backtest came from true structural payoff versus from getting re-priced richer. Expect the first to be more durable and the second to wash out.

If you build a factor-timing overlay, include a structural-premium signal alongside your standard momentum and valuation blocks. This helps on 6-24 month horizons, where many allocators actually rebalance.

Practical steps:

1. Compute each factor’s valuation ratio through time.

2. Attribute historical returns into revaluation vs structural.

3. Form a structural-premium score and rank factors monthly on an expanding window.

4. Tilt toward high-structural factors; keep momentum for short horizons and use valuation as a check on extremes.

2. Redefining “growth” so you’ll own fast improvers, not just expensive stocks

TLDR

• Replace price-driven growth indices with a rule that selects stocks based on real business growth (R&D, sales, gross profit) and weights by the dollar magnitude of that growth

• In 1969-2024 backtests, the composite “Fundamental Growth 1000” beat market cap-weighted growth by 1.7-1.8% per year with significant alpha

How is “fundamental growth” different from common “growth stock” definitions

Most growth style indexes call a stock “growth” if it scores high on a small set of ratios and forecasts, then they weight by market cap. The ratios often include a version of P/E and price momentum. (See S&P index methodology, for example)

The paper defines growth by realized business expansion (multi-year R&D, sales, and gross profit) and then weights by the dollar size of that growth. It keeps price and analyst forecasts out of both steps. That is a material departure from the big providers.

What the authors did

Step 1: they defined growth from business results, not price

The authors pulled a five-year history of each stock’s R&D, sales, and gross profit, and computed two things:

1. a rate score that measured how fast those items grew on a per-share basis, scaled by the firm’s size using sales;

2. a magnitude score that measured how many actual dollars those items increased over the same window.

(If a company’s sales declined, that part was set to 0 so shrinking sales did not help the score.)

Stocks qualified for the portfolio based on the rate score and then were sized by the magnitude score.

Quick example

If two companies both grew quickly but one added $50M of gross profit while the other added $5B, both passed the rate screen, but the $5B improver received a larger weight because its growth was larger in dollars. That is the core idea: select on speed, weight on muscle.

Step 2: they turned the scores into portfolios

They built the composite “FG 1000” by taking the 1,000 U.S. stocks with the highest combined rate score and then weighting those names by the combined magnitude score. The portfolio was rebalanced each March. 

For benchmarks, they used a market cap-weighted top-1000 and a simulated market cap-weighted growth index. They also tested tighter U.S. versions (FG 500, FG 250, FG 100) and ran the same rules in the UK, Europe ex-UK, and Japan.

The results

US broad portfolio

The FG 1000 composite beat market cap-weighted growth-style indices by 1.7-1.8% per year with CAPM alpha of 1.6-1.7% (t-stat ~4). The four-factor alpha remained significant with low factor loadings.

• The four-factor alpha means that they ran a Carhart four-factor regression, which explains a portfolio’s returns using four common factors: market, size (SMB), value (HML), and momentum. The regression’s intercept is the “four-factor alpha.” If that intercept is statistically different from zero, the portfolio earned return that those factors do not explain.

• The “factor loadings” are the betas on each factor in that regression. Low loadings mean the portfolio had only small exposures to market, size, value, and momentum. So the result was not just a hidden tilt to, say, small or value. The exposures were small, and the alpha was still there.

Driver mix: excess return comes mainly from stronger EPS growth, not revaluation or dividends. For example, FG portfolios’ log excess returns show most of the spread from EPS growth; valuation change is small to negative.

Single-measure variants: 

• R&D Growth CAPM alpha ~1.9% (t~2.1)

• Gross Profit Growth alpha ~1.4% (t~3.0)

• Sales Growth alpha ~1.5% (t~2.6)

International results

• UK: FG delivers 7.9% vs 6.8% for both cap-weighted market and cap-weighted growth.

• Europe ex-UK: FG adds ~2.0% excess.

• Japan: FG outperforms by >3% per year, with CAPM alpha t-stat ~4.1; more concentrated FG portfolios reach 4-5% annual excess.

What it fixes vs traditional growth indices

Price is a noisy proxy for growth. Traditional style indices still lean on valuation multiples and short-horizon forecasts, which can label slow-growing expensive names as “growth.” FG ignores price at both selection and weighting, so it does not pay up for glamour without fundamentals. Sector concentration is also lower and more stable than in cap-weighted growth. (Research Affiliates)

Why it matters & how to use it

For allocators: If you want “growth,” target real business growth. Ask managers how they define growth and how much price sneaks in. Favor strategies that select on multi-year R&D, sales, and gross profit trends, and weight by fundamental magnitude rather than market cap. Expect alpha to be larger when stress is high.

For stock pickers and quants: uild a composite from R&D, sales, and gross profit. Use per-share rates for selection and dollar magnitudes for weights. Normalize and diversify across signals. Keep price out of both steps.

Disclaimer

This publication is for informational and educational purposes only. It is not investment, legal, tax, or accounting advice, and it is not an offer to buy or sell any security. Investing involves risk, including loss of principal. Past performance does not guarantee future results. Data and opinions are based on sources believed to be reliable, but accuracy and completeness are not guaranteed. You are responsible for your own investment decisions. If you need advice for your situation, consult a qualified professional.