A 26% Bug: When ROE Is Nonsense

Return-on-equity is a textbook quality metric, so when an ROE-based signal showed a positive information coefficient of +0.013 we nearly added it to the keep pile. Then we looked at what was actually driving the number, and found a bug wearing the costume of a factor.

What broke the metric?

ROE is earnings divided by book equity. The problem is the denominator. After large buybacks, write-downs, or years of accumulated losses, a company’s book equity can go negative — and roughly 26% of the S&P 500 had negative book equity at some point in our sample. When equity is negative, the ratio becomes nonsense: a profitable company can show a large negative ROE, and a money-loser can show a positive one. A handful of these outliers, sorted into the wrong bucket, were quietly producing the entire “signal.”

What did the clean metrics say?

Once we looked past ROE, the quality story changed. Return-on-assets — which uses total assets, always positive, as the denominator — had an IC of −0.006: no edge. Generic book-value/price “value” was about zero. The one quality metric that held up was gross profitability: IC +0.0218, t=1.69, positive in every regime, and grounded in theory as the Fama-French profitability factor. The real edge was specific and modest; the impressive-looking ROE result was an accounting artifact.

The lesson generalizes well beyond ROE. A plausible information coefficient is not evidence of an edge; it is a prompt to ask what is producing it. Before trusting any factor we now stress-test it for data artifacts — negative denominators, outliers, unit errors, stale or restated fields. In our experience, most “surprising” edges are bugs first and signals never.

About this series: every figure comes from a leak-free research harness on US equities — point-in-time index membership, fundamentals keyed to filing date, expanding-window walk-forward, and transaction costs charged. Statistics are gross and in-sample unless noted, and describe published anomalies, not a Yayati product. Standing caveats: roughly a third of true historical index members are unpriced by the naive data source (survivorship); a 2 bps cost assumption is optimistic; fundamentals are post-2009 XBRL.

This article is for educational and informational purposes only and is not investment, tax, or legal advice. It describes findings from an internal research program about publicly documented market anomalies and research methodology; it is not a description of any Yayati product or its results. Research statistics are gross, in-sample illustrations subject to survivorship, data-coverage, transaction-cost, and modeling limitations described in the text, and do not represent actual trading or any client account. Past performance and backtested results are not indicative of future results. Yayati Asset Management is a Registered Investment Adviser. © Yayati Asset Management. VOLT™ and PLASMA™ are trademarks of Yayati.