The short version
From 2012 to 2026, none of the 6 screens beat the S&P 500, which compounded at 14.3% a year. The fairer test is against the whole pool of stocks they were picking from, which returned 12.3% a year: 1 of 6 screens beat that. We are publishing the results either way, because a screen you cannot test honestly is not worth trusting.
Growth of $10,000, equal-weighted, rebalanced quarterly. Hover or tap the chart for values.
Results by screen
| Screen | Annual return | vs S&P 500 | Worst drawdown | Avg. stocks held |
|---|---|---|---|---|
| Magic Formula | 12.4% | -1.9 pts | -52% | 20.0 |
| Dividend Quality | 11.2% | -3.1 pts | -53% | 20.0 |
| Quality Compounder | 10.7% | -3.6 pts | -43% | 20.0 |
| Momentum / Breakout | 9.4% | -4.8 pts | -47% | 20.0 |
| Graham Defensive | 8.7% | -5.6 pts | -47% | 5.2 |
| Deep Value / Net-Net | 2.1% | -12.2 pts | -64% | 0.2 |
| S&P 500 (SPY, with dividends) | 14.3% | -34% | ||
| Equal-weight universe | 12.3% | -2.0 pts | -41% |
2012-04-02 to 2026-10-01, quarterly, top 20 equal-weighted, no costs. Screens that find few large companies (Net-Net, Graham) hold cash in empty quarters. Hypothetical results.
How we tested it
- Point in time. On the first trading day of each quarter we ran the screen using only SEC filings that had already been filed, and prices through the day before. Nothing from the future leaks in.
- The top 20, equal weight. We bought the 20 best-ranked names (the same ranking the screener uses), held them for the quarter with dividends reinvested, then did it again.
- Size filter. Companies under $300M in market cap are excluded, because tiny, thinly traded stocks are hard to buy at these prices. We have no trading-volume data, so market cap is our stand-in for tradability.
- Two benchmarks. The S&P 500 with dividends (via SPY), and an equal-weighted basket of every eligible stock we screened. The second one answers a fairer question: did the rules beat simply owning the whole pool?
- No trading costs or taxes. Real-world results would be lower, and more so for screens that turn their holdings over quickly, like Momentum.
What this does and doesn't prove
- Hypothetical results; not investment advice. Past performance does not predict future results.
- Survivorship bias: the universe is today's S&P 500 and Russell 2000 constituents. Companies that were delisted or went bankrupt are not included, which flatters cheap-stock strategies.
- Fundamentals use only SEC filings filed before each rebalance (no look-ahead). Buys happen at the close of the first trading day of the quarter; no transaction costs, taxes or slippage.
- Insider Conviction is not included: SEC Form 4 history in the dataset begins in 2025.
- Magic Formula is the screener's approximation. No volume data, so a minimum market cap is the liquidity proxy.
- Universe: S&P 500 + Russell 2000 (current constituents).
One more thing worth saying plainly: a screen is a filter, not a forecast. Cheap, profitable, safe-looking companies can stay cheap for years, and a 20-stock portfolio will have long stretches of underperformance even when the idea is sound. For more on that, see Is value investing dead? and Value traps.
Run the screens yourself
Every match in the screener shows the arithmetic that put it there. Free to use.
Open the screenerEducational content, not investment advice. Past performance — of people, papers, or backtests — doesn't guarantee future results. Do your own homework before buying anything.