The Conservative Formula in US equities

What we test

Pim van Vliet and David Blitz propose selecting low-volatility stocks and favouring those with stronger momentum and higher net shareholder payouts. We ask what a US adaptation with explicit rules produces and how it compares with holding SPY.

How we build the portfolio

We start with Portfolio123’s historical Russell 1000 proxy. After requiring sufficient history, we select the 500 companies with the lowest volatility over 36 monthly returns. Within that group, we combine the 12–1 momentum ranking — excluding the most recent month — with net payout yield (NPY): dividends and a net buyback proxy based on shares outstanding.

The portfolio holds 100 stocks, each weighted at 1% at quarterly rebalancing. It starts with USD 1 million and enters at the first close after the cut-off. Dividends are included once through total return; modelled IBKR/ARCA commissions are charged on net position changes, without artificially selling and repurchasing the entire portfolio.

Full-period results

Between 4/1/2002 and 9/25/2026, the adaptation delivers a slightly higher annualised return than SPY, with lower volatility and a shallower observed maximum drawdown. The table shows the scenario after modelled commissions, not returns net of all costs.

The period changes the picture

The accumulated advantage over SPY comes mainly from 2002–2016. In 2017–2021, the gross curves record a CAGR of 11.85% for HIDRA versus 18.20% for SPY, with a deeper observed drawdown. These are descriptive windows, not out-of-sample validation.

The original study’s 15.1% compound return covers 1929–2016 and is not directly comparable with our period. Between the closes on 5/31/2002 and 11/29/2024, Robeco’s series records a gross CAGR of 11.09% and HIDRA 10.57%, using the same monthly annualisation basis. When the comparison ends in 2016, HIDRA is slightly ahead. Aligning dates reduces the apparent gap but does not identify the cause of the remaining difference between implementations. Table 3 in the paper.

Study notes

This is an adaptation, not an exact replication of the original CRSP universe. Historical fundamental-data availability has not been certified issuer by issuer. Four cut-offs retain 499 valid NPY observations, and five corporate-event assumptions remain open with measured sensitivities. The weekly curve may omit drawdowns between dates. Spread, slippage, market impact and taxes are not included.

Conclusion

The adaptation achieves a slightly higher annualised return than SPY and a shallower observed maximum drawdown over the full period. However, that advantage is concentrated in the first segment and does not hold across all windows: the conservative filter does not guarantee lower risk or better results in every crisis.

Comparing the same metric over the same dates substantially reduces the gap with Robeco’s series. The study provides a more consistent reference for assessing the strategy, but it does not confirm an exact replication or a persistent advantage. The results should be read as historical research, not as a real investment track record.