Quality Tilt Strategy

The thesis

Some businesses earn far more on the capital they employ than others, and keep doing so. A quality tilt simply prefers them: rank companies on profitability, require a sound balance sheet and real cash generation, and avoid the ones priced as if nothing could go wrong.

This is not a secret method. It is the part of long-term fundamental investing that can be written down and ranked. What the screen cannot supply is the judgment about whether the profitability is durable — so treat its output as a short list worth researching, not a verdict.

Academic basis

Novy-Marx (2013) showed that gross profits-to-assets has roughly the same power as book-to-market in predicting the cross-section of stock returns, and that profitable firms earn higher returns despite trading at higher valuations.

Asness, Frazzini, and Pedersen (2019) defined quality more broadly — profitable, growing, and safe — and found that a Quality-Minus-Junk factor earned significant risk-adjusted returns in the US and across 24 countries, tending to do well when markets fall.

Frazzini, Kabiller, and Pedersen (2018) connect this to the best-known record in investing. Berkshire Hathaway’s Sharpe ratio of 0.79 from 1976 to 2017 is largely explained by a tilt toward cheap, safe, high-quality stocks combined with leverage of roughly 1.7-to-1 financed cheaply by insurance float. Once those exposures are accounted for, the remaining alpha is statistically insignificant.

Greenblatt (2005) popularised pairing return on capital with earnings yield as the “magic formula”. An equal-weight quality + value rank in that spirit was this strategy’s first design. We changed it after testing it — see the next section.

What our own backtest found

We rebuilt the screen at every quarter end from March 2012 to October 2026 using only what was knowable on each date: S&P 500 membership as of that day, SEC XBRL fundamentals filtered by filing date, and market capitalisation from the share price actually quoted that day. Portfolios are equal-weighted, rebalanced quarterly, with 10 bps charged on every unit traded. The universe is every non-financial index member that passes the same safety gates.

The strategy did not beat its universe, and it trailed the S&P 500 by 1.4 percentage points a year. Its drawdown was somewhat shallower than the universe’s. None of the differences in the table is statistically significant: the largest, for quality with no value floor, has a t-statistic of about 1.0 over 14.5 years.

The pattern across rows is consistent, though: the more that value constrained the selection, the worse the result. That is why value was demoted from half of the rank to a floor. We did not remove the floor to adopt the best-looking row — choosing a rule after seeing which one won is how backtests mislead. The floor was fixed at 25% before this test was run.

Limits of the test: companies later acquired or delisted mostly could not be included (71% of index members could be evaluated in 2012, rising to 99% by 2026); the ratios are rebuilt from raw SEC filings and differ slightly from the data-vendor ratios the live scanner reads; the live universe is a curated list of about 100 names, not the whole index; and 2012–2026 was an unusually strong period for large growth companies. Past results, including these, do not predict future returns.

How Alpha Suite implements it

• Non-financial universe — ~100 liquid US large and mid caps across technology, healthcare, staples, discretionary, industrials, energy, materials, and communications. Financials, utilities, and REITs are excluded: their leverage is structural, so enterprise-value and operating-return ratios do not rank on the same scale.
• Quality rank — the average cross-sectional rank of return on assets, return on equity, and operating margin, as proxies for return on capital. ROE is capped at 60% because buybacks shrink book equity and inflate the ratio without the business improving. At least two of the three must be reported.
• Value floor — EBITDA / enterprise value, averaged with free-cash-flow yield when it is reported. Value does not drive the ranking; it only removes the most expensive quarter of the universe.
• Safety gates — positive trailing EPS, positive free cash flow, and net debt / EBITDA of 3.5x or less.
• Selection — top quintile of the quality rank, minus any name in the most expensive quarter on value. Cheapness alone never qualifies a name.
• Entry checks — a name trading more than 10% below its 200-day average loses 6 points; RSI above 75 loses 5.
• Regime-aware scoring — +5 in DEFENSIVE and +3 in CAUTIOUS regimes. Like every directional long, new signals are suppressed when the macro regime blocks new longs.
• Six-month horizon (180 days) — the stop is anchored to each name’s own volatility over the horizon (bounded 8%–18%) with the take-profit at 1.5x the stop. These are risk-management levels, not return forecasts, and they were not part of the backtest above.

When it fires

The quality tilt is an always-on screen, not an event trigger. Fundamentals update quarterly, so the list changes slowly: expect around ten names, with a few rotating as earnings are reported and prices move.

Because the list is slow-moving and carries no demonstrated edge on its own, it is most useful as context for other signals: a high-quality name that also shows insider buying or a new buyback authorization has an event-driven reason to act, and the quality rank tells you what kind of business it is.