26 years against the index: what eight systematic strategies reveal

All eight strategies outperformed the index in average annual return, by 3 to 16 percentage points per year. Six of the eight also outperformed it on the Sharpe ratio.

EMERALD has the best risk/return profile. Its return is almost on par with the most aggressive strategy, POWER, while its maximum drawdown is half that of the index. EMERALD's Calmar ratio is 0.91, compared with 0.17 for the S&P 500. POWER delivers the highest cumulative return, but at the cost of drawdowns on the scale of the index. STABILITY lives up to its name: with volatility on par with the S&P 500, it delivers twice the return with shallower declines.

The table and charts also include the blended strategy - a portfolio of eight equally sized subportfolios, each corresponding to one of the strategies. The subportfolio allocations are reset to equal weights once a year. We explain its purpose below.

A single-strategy portfolio holds about 10 stocks, and a significant share of its risk comes from individual companies. The natural next step is to combine strategies, and the blended strategy is the simplest way to do this. The strategy portfolios overlap in part, leaving 30-⁠40 stocks in the combined portfolio. The stock weights are unequal: a stock held by several strategies receives a correspondingly larger weight.

The blended strategy delivered an annual return of 18.7%, with volatility of 17.9%, a maximum drawdown of -39.0% and a Sharpe ratio of 0.95. This is lower than for the best individual strategies, but there is no need to pick the winner in advance, and the outcome depends much less on luck in selection. The blended strategy outperformed the index in 22 of 26 calendar years and underperformed it in 2016, 2018, 2022 and 2024.

Equal allocations are the simplest way to blend strategies. Combinations that account for correlations between strategies and their behavior across different market regimes produce more interesting results. This is a topic in its own right, which we will discuss separately.

For an experienced investor, average returns are less important than what happens during difficult periods. Defensive characteristics vary across the range, and this is deliberate. POWER, EMERALD and AVICORE are the growth engines. POWER pays for the highest return with drawdowns on a par with the index: in 2008, it lost almost as much as the market. EMERALD is the exception in this group: with returns close to POWER's, it also proved the most resilient in crises. STABILITY and DIVIDEND serve as anchors, as does gold outside the strategy range: its correlation with the S&P 500 is virtually zero, and it rose in 2000-⁠2003 and 2007-⁠2009.

Not every crisis unfolds in the same way. In 2022, most strategies lost at least as much as the index: the sell-⁠off in growth companies also affected our portfolios.

Splitting the period into two subperiods shows that the result does not depend on a single favorable stretch. In 2000-⁠2012, the S&P 500 returned 2.1% annually and the blended strategy 14.3%. In 2013-⁠2025, the index returned 14.6% and the blended strategy 23.4%. The excess return persisted both during the "lost decade" for stocks and in a strong bull market.

DIVIDEND is the exception: over the last ten years, its return fell to 3.9% annually. This is consistent with the broader underperformance of the dividend factor in an era dominated by growth companies and calls for separate analysis.

The strategies have been traded in live accounts at Interactive Brokers since October 22, 2024. Account values are recorded on the 22nd of each month. The launch took place in three stages: POWER, INDIGO, YELLOW and GREEN have been traded since October 2024, STABILITY, EMERALD and DIVIDEND since October 2025, and AVICORE since April 2026.

All results are net of brokerage commissions and withholding taxes on dividends. Capital gains tax is not included.

We will analyze only strategies that have been traded for more than a year. Over a shorter period, there is too little data and too much noise: a few months of results say more about market randomness than about a strategy's quality. That is why only the four strategies from the first launch phase are shown below. The others will be included in the analysis once each has accumulated a year of live trading.

Over 23 months, these four strategies returned between 63% and 82%, or 29-⁠37% annually. The S&P 500 TR rose by 35.9% over the same period, or 17.4% annually.

We recognize that the data is still limited: with annual volatility of 20-⁠25%, the range of outcomes is very wide even over two years, and it is too early to draw conclusions about excess returns - whether favorable or unfavorable. The track record is updated monthly.

All results for 2000-⁠2025 were obtained using historical data. We know that a backtest typically overestimates future returns, and we have eliminated the common sources of overstatement. The backtest uses historical data, including companies that were later delisted: historical index constituents for strategies based on the S&P 500, and the full company database for DIVIDEND. Delistings and spin-⁠offs are correctly accounted for. Each rebalancing uses only company financial statements already published by that date. Rebalancing once a year means low turnover, so trading costs have little effect on the result. We also tested other annual rebalancing dates: the results vary only slightly, meaning they do not depend on a fortunate choice of month.

Some limitations cannot be eliminated. Eight strategies were tested, so the best performer cannot be assessed in isolation: adjusting for multiple testing (for example, using the deflated Sharpe ratio) reduces the statistical significance of the leading strategies. In this respect, the blended strategy's result is more reliable than that of any individual strategy. A portfolio of 10 stocks is concentrated, and a significant share of its risk is company-specific. Finally, 26 years is a long period, but still only a single historical path.

That is why we launched the strategies with real capital (see the "Live Trading" section above). The track record is still too short to draw statistical conclusions, but this is precisely the test that a backtest cannot replace.