Premier Anomaly: A Proven Framework for Beating SPY Year After Year
Beating SPY was supposed to be impossible. We did it anyway.
In the summer of 1831, a volcanic eruption released sulfur gas into the atmosphere, cooling Northern Hemisphere temperatures by 1°C, leading to famine and crop failures in Europe. The composer Felix Mendelssohn, while traveling through the Alps in August, remarked, 'It is as cold as in winter; there is already deep snow on the nearest hills' (Wizevich, 2025).
The stock market, like the weather, operates within probabilities shaped by trends, and shifting variables. While outcomes are unpredictable, probabilities frame what is likely versus rare.
Based on probabilities, the Q72 Stratum Index utilizes a modular architecture and business cycle indicators to tactically allocate assets across sectors—while staying prepared for the unexpected, like snow in summer. Using rules-based algorithms, it harnesses the momentum effect, a well-documented phenomenon suggesting past returns can indicate future performance. This principle is supported by centuries of evidence across asset classes (Geczy & Samonov, 2012; Hurst, Ooi, & Pedersen, 2012; Antonacci, 2014).
Since 2013, the model, powered by Stratum Index signals, has demonstrated how $10K could grow to over $2.1 million, delivering a 21,041% return (61% annualized, Sharpe 2.72, beta 0.64) while outpacing SPY with lower volatility and smaller drawdowns.
These results reflect modeled trades based on Stratum Index-driven allocations; actual returns may vary depending on execution, fees, and capital.
Q72 Stratum Index Returns (Oct 25, 2013 – Dec 27, 2024)
A Modular Approach to the Momentum Strategy Design
The Q72 Stratum Index is a systematic, rules-based approach that trades weekly. Using a modular framework where multiple algorithms independently select ETFs, it reduces overfitting, hindsight bias, and trading costs. This results in a diversified portfolio with lower risk and superior risk-adjusted returns. Over 11 years, the Q72 Stratum Index would have returned 210.41x the initial value, compared to just 4.17x for a simple buy-and-hold SPY investor.
SPY vs Stratum Index — Log Growth (Jun 20, 2013 – Dec 27, 2024)
Algorithmic Trading Challenges
Financial markets, shaped by human behavior, are prone to recurring follies. For centuries, crashes have stemmed from speculative bubbles, fueled by overconfidence and flawed assumptions.
A cautionary parallel is the South Sea Bubble of 1720, where rampant speculation led to a catastrophic market collapse. Sir Isaac Newton—brilliant mathematician and inventor of derivative calculus—fell prey to the frenzy, losing £20,000 after reinvesting at the bubble’s peak. Adjusted for inflation, his loss would amount to roughly £40 million today.
His reflection, "I can calculate the motions of the heavenly bodies, but not the madness of men," underscores the enduring challenge of predicting market behavior. Just as Newton misjudged the power of speculation, modern trading—no matter how sophisticated—often fails to anticipate herd mentality, abrupt shifts, and the deep irrationality that still governs financial markets.
"I can calculate the motions of the heavenly bodies, but not the madness of men.” - Sir Isaac Newton
Data Mining Challenges
Efforts to time financial markets and identify price trends have long divided practitioners and academics, with the debate hinging on whether patterns in past data can truly predict future outcomes. Before the 2000s, most academics dismissed market timing as futile, exemplified by Malkiel's (1995) critique of technical analysis, which argued that such methods added little value beyond randomness.
However, Brock et al. (1992) shifted perspectives by applying 26 technical trading rules to the Dow Jones Industrial Average (DJIA) from 1897 to 1986. Their robust methodology accounted for data-snooping bias, used a long historical dataset, and demonstrated potential support for technical strategies. Yet, follow-up studies like Fang, Jacobsen, and Qin (2013) found poor out-of-sample performance for these same rules, raising questions about their reliability.
The challenges of data mining came into sharper focus with Bajgrowicz and Scaillet’s (2012) monumental effort to evaluate 7,846 trading rules on the DJIA from 1897 to 2011. Using the false discovery rate (FDR) to correct for data-snooping bias, they showed that investors could not have identified the best-performing rules in advance. Introducing transaction costs eliminated any remaining profitability.
Similarly, Fang, Qin, and Jacobsen (2014) tested 93 market indicators on S&P 500 data and found that none outperformed a simple buy-and-hold strategy after accounting for costs. These findings underscore the inherent challenges of data mining, where the search for meaningful patterns often leads to overfitting or chasing noise disguised as signal. Such outcomes reinforce the value of simple, robust strategies like absolute momentum, as highlighted by Antonacci (2014, pp. 131–132), which prioritize enduring principles over fleeting statistical curiosities.
Momentum: The Premier Anomaly
Unlike a coin toss, stock movements reflect patterns shaped by investor behavior and fundamentals. The S&P 500 selects the top 500 U.S. companies by market cap, favoring firms with sustained growth. Recurring deposits from retirement accounts into these stocks further amplify returns, creating reliable yet semi-predictable growth trends. These contributions are intentional, not random, driving capital into these companies.
This highlights the market's semi-random nature, where factors like earnings, industry performance, and geopolitical events create impacts that are semi-predictable yet inconsistent. News—a subjective force—might drive sharp movement in a stock one week and have no effect the next.
Amid this, the momentum effect stands out as one of the most robust and widely studied phenomena in financial markets, with evidence spanning various asset classes and historical periods. Research, including the foundational work of Jegadeesh and Titman (1993, 2001), has validated the predictive power of momentum, both in relative strength—where an asset's performance relative to its peers predicts its future performance—and in absolute momentum, where an asset's own past returns forecast its future performance.
Absolute momentum, which measures an asset's excess return over a specific look-back period, has shown strong applicability across asset classes such as commodities, equity indices, bond markets, and currency pairs, even back to the early 1800s (Geczy and Samonov, 2012; Hurst, Ooi, and Pedersen, 2012; Antonacci, 2014).
Profiting from the ‘Madness of Men’
Momentum investing thrives on the "madness of men," despite ongoing debates about its underlying causes. Behavioral explanations, such as anchoring, herding, and the disposition effect, suggest that cognitive biases fuel momentum profits. These biases lead investors to underreact to new information, overreact during trends, and mishandle winning and losing investments. Additionally, risk-sharing mechanisms and structural market dynamics, as highlighted by Brown and Jennings (1989) and Zhou and Zhu (2014), offer further insights into the persistence of momentum, blending human psychology with market mechanics (Antonacci, 2014, pp. 153–154).
Modular Script Architecture
The Q72 Stratum Index consists of multiple algorithms that measure relative and absolute momentum across varying timeframes. Script 1 might allocate 10% to an equities ETF, while Script 2 might allocate 10% to a defensive ETF, and so on. The final strategy combines the outputs of all the scripts, resulting in a balanced and diversified portfolio.
Example Allocation for the Q72 Stratum Index
Methodology
Each script analyzes historical performance across a given time series and a selection of ETFs. Script 1 may not have the same universe of ETFs to choose from as Script 2. This approach captures short-term shifts, medium-term trends, and long-term market cycles. For instance, in the chart below showing yearly returns for three sectors, a given script might allocate to Treasury Bonds if they outperform SPY and QQQ on a year-over-year basis.
Tactical Portfolio Composition
By combining outputs from multiple rules-based algorithms, the Q72 Stratum Index allocates a percentage of total assets to tactical positions. For example, one algorithm might select Gold, another Utilities, and another the S&P 500—each behaving differently for distinct reasons, ensuring a balanced and adaptive approach.
The Business Cycle
A key contribution to the modular script architecture is State Street's business cycle framework, which analyzes decades of market sector relationships across economic cycles. For example, during recoveries, sectors like Consumer Discretionary often outperform the broader market. These historical relationships serve as crucial data points that inform the Q72 Stratum Index’s strategic allocations.
State Street's Business Cycle framework, based on the Conference Board's Leading Economic Index (LEI) introduced in the 1960s, categorizes economic activity into four phases: Recession, Recovery, Expansion, and Slowdown. By integrating Kenneth French’s data, it analyzes sector performance across seven recessions, 12 expansions, and 11 slowdowns using metrics like average returns and hit rates (periods of outperformance).
Recession: The LEI troughs, signaling contraction. Consumer Staples, Utilities, and Health Care thrive with 86%-100% hit rates.
Recovery: The LEI rebounds, signaling growth. Consumer Discretionary and Real Estate lead with 57%-86% hit rates.
Expansion: The LEI rises above trends, driven by demand and profits. Technology and Financials excel with 82%-91% hit rates.
Slowdown: The LEI peaks as growth slows. Health Care and Consumer Staples perform defensively with 73% hit rates.
By combining LEI cycles with sector data, State Street offers insights for constructing portfolios tailored to economic conditions. (Bartolini et al., 2019)
Methodology
By examining key sectors and their relative performance, we can make high probability educated guesses about the economy's current phase.
For instance, if the Utilities sector outperforms Technology, it may indicate an approaching recession. However, such indicators are not definitive; they could signal de-leveraging or a shift in investor preference from high-growth sectors like the Nasdaq to more stable options like Utilities.
Selection Criteria
Selection is based on both 12-month rolling correlations and intentional diversification—some scripts deliberately exclude assets included in others. This approach reduces redundant exposure, enhances diversification, and maintains balanced risk without overfitting. Roughly 75% of the portfolio follows a 65/30/5 allocation across a primary asset, secondary asset, and 5% allocated to cash. Asset order adjusts with market dynamics—for example, QQQ and TLT might be held at 65/30, but in a high-volatility environment, that could shift to TLT and GLD.
Correlation is not Causation
Correlation reflects how assets tend to move in relation to one another, but it doesn’t guarantee consistent behavior. Combining assets with varying correlations builds portfolio resilience—at the cost of total expected returns, which is the core premise of hedging. This positions the Q72 Stratum Index to perform across a range of market environments.
A correlation matrix measures how assets move in relation to each other, with values ranging from -1 to 1:
1.00 = Perfect positive correlation (move together)
0.00 = No correlation (independent)
-1.00 = Perfect negative correlation (move in opposite directions)
Market Regimes
The 2020 COVID Crash
Timeframe: February 21, 2020, to March 27, 2020
Q72 Stratum Index Performance During COVID-19 Market Turmoil:
Phase 1 – Initial Stock Crash (Feb 21–Mar 6, 2020): The Q72 Stratum Index declined 2.01%, demonstrating resilience during the early equity selloff.
Phase 2 – Treasury Bond Crisis (Mar 9–Mar 18, 2020): After gaining 12.6% from early bond positioning, the strategy fell 14.7% during the sudden Treasury market dislocation, reflecting systemic stress across asset classes.
Benchmark Comparisons:
SPY (S&P 500 ETF): Declined 31.67% during the broader market downturn.
SPXL (3x Leveraged S&P 500 ETF): Plummeted 74.67%, illustrating the amplified risk of leveraged exposure.
Performance Comparison During the 2020 COVID Crash
Jan–Feb 2018 'Volmageddon'
Timeframe: January 26, 2018, to February 16, 2018
The Q72 Stratum Index showcased its adaptive approach by using tactical allocation and dynamic sector adjustments to navigate flash crashes and minimize volatility. However, like many momentum strategies, it is not immune to whipsaw effects—sharp market reversals that can temporarily reduce its effectiveness and highlight a key weakness in such approaches.
Comparative Performance:
Q72 Stratum Index: Declined 15.87%
SPXL (3x Leveraged S&P 500 ETF): Dropped 20.56%
SPY (S&P 500 ETF): Declined 6.86%
Performance of the Q72 Stratum Index: 2017-2018
Starting January 26, 2018, without prior warning, and just before the significant VIX spike, the Q72 Stratum Index was strategically allocated across a mix of offensive and defensive sectors, including 14.4% in Cash, 14.3% in Bonds and 3.7% Gold.
Allocation of the Q72 Stratum Index Jan 26 2018
By the week of February 23, 2018, allocations had shifted to 16.2% shorts, 8.8% Cash, and 25% Gold, highlighting the algorithm’s tactical hedging capabilities. Notably, bonds underperformed during ‘Volmageddon,’ and the algorithm had already reallocated assets accordingly. Activating under high-probability conditions. Had the market continued to decline, the algorithm would have progressively adjusted further to safeguard the portfolio.
Allocation of the Q72 Stratum Index Feb 23 2018
10-Year Monte Carlo Simulations
The Monte Carlo analysis highlights the Q72 Stratum Index’s adaptability and consistent performance across diverse market conditions, instilling confidence in its long-term application. In the original dataset, the strategy successfully navigated a wide spectrum of market phenomena, including record inflation, flash crashes, volatility spikes, and recessionary periods.
10-Year Monte Carlo Simulation: Metrics Grouped by Percentiles of Total Returns
Monte Carlo Simulations: Analyzing 10-Year Performance
10,000 Monte Carlo simulations covering a 10-year period.
Results were grouped into deciles, with each decile representing the average performance of 1,000 simulations.
Key Insights:
High-Performing Deciles: The strategy exhibited the potential for exceptional returns in the top-performing deciles.
Low-Performing Deciles: Even in the lowest-performing deciles, the strategy demonstrated resilience, underscoring its ability to manage risk across varying market conditions.
Appendix
Antonacci, Gary (2014). Dual Momentum Investing: An Innovative Strategy for Higher Returns with Lower Risk. McGraw-Hill Education, Kindle Edition (pp. 131–132).
Antonacci, Gary (2014). Dual Momentum Investing: An Innovative Strategy for Higher Returns with Lower Risk. McGraw-Hill Education, Kindle Edition (pp. 153–154).
Bajgrowicz, Pierre, and Olivier Scaillet (2012). Technical Trading Revisited: False Discoveries, Persistence Tests, and Transaction Costs. Journal of Financial Economics, 106(3), 473–491.
Bartolini, M., Dong, A., & SPDR Americas Research. (2019). Sector Business Cycle analysis (By US Department of Commerce & Bloomberg Finance L.P.). https://www.ssga.com/library-content/products/fund-docs/etfs/us/insights-investment-ideas/sector-business-cycle-analysis.pdf
Brock, William, Josef Lakonishok, and Blake LeBaron (1992). Simple Technical Trading Rules and the Stochastic Properties of Stock Returns. Journal of Finance, 47(5), 1731–1764.
Brown, David P., and Robert H. Jennings (1989). On Technical Analysis. Review of Financial Studies, 2(4), 527–551.
Fang, Jiali, Ben Jacobsen, and Yafeng Qin (2013). Predictability of the Simple Technical Trading Rules: An Out-of-Sample Test. Review of Financial Economics, 23(1), 30–45.
Fang, Jiali, Yafeng Qin, and Ben Jacobsen (2014). Technical Market Indicators: An Overview. Working paper.
Geczy, Christopher, and Mikhail Samonov (2012). 212 Years of Price Momentum (The World’s Longest Backtest 1801–2012). Working paper.
Isaac Newton pay order (1719). Isaac Newton pay order. The New York Public Library. Retrieved from https://www.nypl.org/events/exhibitions/galleries/empire-imagination/item/11057
Jegadeesh, Narasimhan, and Sheridan Titman (1993). Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency. Journal of Finance, 48(1), 65–91.
Jegadeesh, Narasimhan, and Sheridan Titman (2001). Profitability of Momentum Strategies: An Evolution of Alternative Explanations. Journal of Finance, 56(2), 699–720.
Malkiel, Burton G. (1995). Returns from Investing in Equity Mutual Funds. Journal of Finance, 50(2), 549–572.
The South Sea Bubble of 1720 (2023). The South Sea Bubble of 1720. Historic UK. Retrieved from https://www.historic-uk.com/HistoryUK/HistoryofEngland/South-Sea-Bubble/
Wizevich, E. (2025). Scientists find the mysterious source of the massive 1831 volcanic eruption that cooled Earth and made the sun appear blue. Smithsonian Magazine. Retrieved from https://www.smithsonianmag.com/smart-news/scientists-find-the-mysterious-source-of-the-massive-1831-volcanic-eruption-that-cooled-earth-and-made-the-sun-appear-blue-180985784/
Zhou, Guofu, and Yingzi Zhu (2014). A Theory of Technical Trading Using Moving Averages. Working paper.
Disclaimer: The content in this post is for informational purposes only and does not constitute financial or investment advice.
Dear Quanta 72, thanks for your superb work. Every time we buy an ETF, there are bid-ask spreads, bank costs, etc that add up a lot. In your Stratum backtest, I assume a minimum cost of -0,40% for each rotation (-0.20% sell ETF, and another -0,20% buy ETF), otherwise Stratum is not real money but, only paper money. That can add to -7% performance per year, depending on the rotation/costs, which in your model seems to be achieved with huge rotation. Could you please explain this? Thisis what I need before applying. There are models out there that with huge rotation and real life commissions do not work at all.