The Best Edge We Found, and How It Died
Key takeaways
- Gross profitability was the most credible factor we tested: IC +0.0218, t=1.69, positive in every regime.
- Against the cap-weighted index the edge was only ~+0.6%/yr — and survivorship bias (~+3%/yr) likely exceeded it.
- Smarter construction (a risk-aware optimizer) lifted the information ratio but not the underlying strength.
- The honest read: even the best factor we found was marginal and largely already harvested by the index.
- What survived every test was a risk control — volatility-targeting — not a return forecast.
If any single factor was going to convince us, it was gross profitability. It was the most consistent edge in the entire program: a positive information coefficient in every market regime, a t-statistic of 1.69, grounded in theory as the Fama-French profitability factor, and — unlike option-skew — it did not collapse when we scaled to the full universe. So we put it through the tests that kill most strategies, and watched the most credible edge in the study turn marginal.
How strong was it, really?
Strong relative to everything else, which is a low bar. Against an equal-weight basket a profitability tilt added about +2.7%/yr. But against the cap-weighted index a client can actually buy, that shrank to roughly +0.6%/yr — because 2016–2024 cap-weighting had already concentrated in high-profitability mega-cap technology, harvesting the premium for free. And the survivorship bias in the universe was worth about +3%/yr on its own, plausibly larger than the entire real edge. The most credible factor in the study was a fraction of a percent against the honest benchmark.
Can better construction rescue it?
Partly, and instructively. Building the same signal with a risk-aware optimizer — characteristic-portfolio weighting with a turnover penalty and a properly estimated covariance — lifted the dollar- and sector-neutral long/short Sharpe from roughly zero (naive quintile sorting) to about 0.28. A real +0.25 of information ratio with no new signal. But it tops out around +0.5%/yr in absolute terms. Construction recovers information that naive sorting throws away; it does not manufacture strength from a weak signal. The ceiling is the signal, and the signal was marginal.
| Gross profitability | Result |
|---|---|
| t-statistic, full universe | +1.69 (best in study) |
| Edge vs equal-weight basket | +2.7%/yr |
| Edge vs cap-weighted index | ~+0.6%/yr |
| Survivorship bias in universe | ~+3%/yr (likely larger than the edge) |
| Naive quintile → optimized L/S Sharpe | ~0 → ~0.28 |
What survived instead?
Not a return forecast — a risk control. Volatility-targeting was the one result that did not weaken under honest testing; re-run fully out-of-sample on the most recent era, it strengthened rather than decayed. It works for a structural reason that has nothing to do with predicting returns: volatility clusters and is forecastable, while returns are not. That asymmetry is the foundation of how we build strategies — manage the risk you can predict, and stop betting on the return you cannot.
The thesis, earned the hard way
Manage volatility, don’t forecast returns. We did not start there. A research program that killed its own best factor — and showed the survivor was a risk control, not a return signal — is what led us there.
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.
The discipline behind our strategies
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.
See VOLT™ on a real position.
The tax-smart option overlay behind this paper, for concentrated stock.