I ran extensive backtests to find out whether adding a ‘take-profit’ sell rule—selling any stock that rises above X%—changes outcomes. The answer may surprise readers.

Subscribe now

When One Stock Takes Over

Big unrealized winners are great—until portfolio allocation becomes lopsided. When a single stock compounds from a 2% position to 15% of a portfolio, the dynamics change. Monthly returns become increasingly driven by that one name. The psychological pressure intensifies—every pullback feels magnified, and the temptation to “protect gains” grows.

The baseline Alpha Engineer model (with no take-profit rule) showed exactly this pattern. As of December 31, 2025, the top holding—Andean Precious Metals (APM)—represented 15.7% of the portfolio after an exceptional run. This concentration isn't inherently bad for returns, but it does create challenges for real-world investors managing their own portfolios. The concentration would be even more pronounced in a typical 20-stock portfolio. At some point, an investor may want to limit the impact of a single stock—but does doing so hurt returns? Let's find out.

The Tests

I tested six different take-profit thresholds against the baseline model: selling any stock once it rises 50%, 100%, 200%, 300%, 400%, or 500% from purchase. All other rules remained identical—the same universe with its quality filters, the same ranking methodology, and the same rank-based sell rule (exit when a stock drops below the top 10%).

The backtest period spans January 2010 through December 2025—a 16-year window that includes multiple market cycles, the COVID crash, and the 2022 bear market. All backtest reports are available for paid subscribers in the Backtest Reports archive.

The Results

What the Data Tells Us

The most striking finding is how little the take-profit threshold matters for long-term returns. Every variant—from the aggressive 50% cap to the lenient 500% cap to the rank-qualified version—delivered a CAGR between 38.70% and 41.11%. The differences from baseline range from -1.50% to +0.91%, well within the noise expected from backtesting.

Risk metrics tell a similar story. Sharpe ratios range from 1.60 to 1.75, Sortino ratios from 2.50 to 2.62, and maximum drawdowns cluster tightly around -37%. None of these variations would change an investment thesis.

The one metric that does move meaningfully is turnover. The baseline model has 174.57% annual turnover. Adding a 500% take-profit rule barely changes this (175.00%), but a 100% threshold pushes it to 202.57%, and a 50% threshold drives it to 240.46%. More aggressive profit-taking means more trading, more transaction costs, and more opportunities for slippage in illiquid microcaps.

What If Sold Stocks Get Repurchased?

One inefficiency, especially with an aggressive take-profit rule: a sold stock might still rank at the very top (rank 98 or 99) and get repurchased a week or two later when another position is sold. This creates unnecessary round-trip transactions.

To address this, I tested a refined variant of the 100% sell rule: sell at 100% gain only if the stock has also dropped below rank 98. This ensures the stock has genuinely weakened in the ranking before triggering a sale.

The result? A 40.35% CAGR—virtually identical to the baseline (40.20%) and the simple 100% rule (41.11%). Turnover dropped slightly to 198.78% (vs. 202.57% for the simple 100% rule), but the performance difference remains well within noise. Even optimizing the take-profit logic to avoid unnecessary trades doesn’t create a meaningful edge.

The Concentration Benefit

Where take-profit rules do make a difference is in managing position sizing. Look at the top holdings across variants as of December 31, 2025:

• Baseline (no cap): Top position at 15.7%

• 500% cap: Top position at 5.1%

• 100% cap: Top position at 3.3%

A 300-500% take-profit rule dramatically reduces single-name concentration without meaningfully impacting returns or risk-adjusted performance. For investors who find it psychologically difficult to watch a single position dominate their portfolio, this represents a legitimate implementation consideration. For the unusual big winners (significantly above 500%), it may make sense to have a sell threshold anyway—at some point, 20-30% of a 20-stock portfolio ends up in a single name. If history is any guide, these are once-in-a-decade gains, but I hope to face this issue more often in the future!

Conclusion

After analyzing 16 years of backtested data across seven different configurations, my conclusion is straightforward:

Take-profit rules neither help nor hurt returns in a statistically meaningful way.

The choice comes down to implementation preferences:

• Investors comfortable with occasional concentrated positions who want to minimize turnover may prefer no take-profit rule

• Those who prefer more even position sizing and can tolerate slightly higher turnover may find a 300-500% threshold offers a reasonable balance

• Thresholds below 200% meaningfully increase trading activity without improving returns

For The Alpha Engineer model portfolio, I’ve chosen not to implement a fixed take-profit rule. However, some investors may prefer to cap winners at 300% or 400% for behavioral reasons. The data suggests this is a perfectly reasonable approach that doesn’t cost returns over time.

Subscribe now

The key insight: systematic investing isn’t about optimizing every parameter to the third decimal place. It’s about finding robust approaches that work across market conditions and sticking with them. Whether multibaggers are sold at 100%, 300%, or allowed to run further, the system keeps generating new candidates to replace them.

The Alpha Engineer --- Investing with a quantitative edge

Disclaimer: The Alpha Engineer shares insights from sources I believe are reliable, but I can’t guarantee their accuracy—data’s only as good as its inputs! This content (whether on Substack, via email newsletters, X, or elsewhere) is for informational and educational purposes only—it’s not personalized investment advice. I’m not a registered investment advisor, just an engineer crunching numbers for alpha. My opinions are my own and may shift without notice. Investing involves risks, including the chance of losing money. Past performance, whether from back-testing or historical data, does not guarantee future results—outcomes can vary. So, please consult your financial advisor to see if any strategy fits your situation. Full disclosure: I may own positions in the securities I mention, as I actively manage my own portfolio based on these strategies.

Welcome to The Alpha Engineer's comprehensive backtest reports archive. This collection represents the quantitative foundation of my systematic investment approach, documenting the rigorous testing behind the strategies I discuss. Each report provides transparent, data-driven evidence of how these methodologies have performed historically across various market conditions.

This archive includes both detailed PDF reports, summary spreadsheets and a link to the research article. Arranged chronologically with the most recent reports at the top, this archive serves as both a reference library and a demonstration of my commitment to evidence-based investing.

About Portfolio123

All backtest research in this archive was compiled using Portfolio123, a powerful quantitative research and backtesting platform I use for strategy development and testing. Portfolio123 provides access to comprehensive financial data spanning +20 years, allowing for robust historical analysis of investment strategies across different market cycles.

Try Portfolio123 Yourself

Want to run your own quant research? Get a 35-day trial of Portfolio123 through my affiliate link and test your investment ideas with professional-grade backtesting tools.

Get Your 35-Day Portfolio123 Trial

Backtest Reports Collection

Read more

Why I Ran These Tests

I wanted to stress-test the Alpha Engineer stock-picking model across a variety of market ecosystems—from an "all regions" universe of nearly 3,800 names down to niche 10-stock portfolios in the UK and Canada.

The goal was twofold:

1. Robustness: Does the factor blend—dominated by small market cap, value factors, fundamental and price momentum factors, and quality screens—still work when we confine it to specific geographies?

2. Practicality: How much does portfolio breadth (or the lack thereof) alter returns, risk and trading costs once we bake in a healthy 1% per-trade friction?

I back-tested six universes:

Sell and buy rules are the same as The Alpha Engineer Model Portfolio. A position is trimmed when it drops below the 90th percentile of the global ranking. With the freed-up funds, the top ‘liquid’ stock (>$50K avg daily liquidity and >$0.5 price) that is not yet in the portfolio is purchased. Portfolio size is limited to 20 stocks to reduce turnover for the smaller universes. For the two smallest universes (UK only and Canada only), I had to limit the portfolio to 10 stocks to manage churn.

The Numbers

1 | Annualised Trailing Returns (net %)

2 | Risk Profile & Turnover

3 | U.S. "Outlier Effect"

I removed two +1,000% moon-shots (KOSS & ZDGE) from the baseline U.S. run to avoid overstating typical performance.

The outliers add roughly 6-8 percentage points of CAGR but double volatility, knocking the Sharpe lower even as the Sortino ticks higher (bigger upside moves improve the upside-downside asymmetry).

Want the complete backtest reports? Detailed reports with yearly performance breakdowns, benchmark comparisons, current stock holdings, sector allocations, and comprehensive trade statistics are available exclusively for paid subscribers.

Subscribe now

A Quick Primer on the Risk Metrics

For those new to quantitative investing, here's what these risk measures mean for your portfolio:

• Standard Deviation (Std Dev) — The total volatility of monthly returns. Higher numbers mean a bumpier ride, whether up or down.

• Downside Deviation — Volatility of negative monthly returns only; punishes painful moves rather than joyful ones. This focuses exclusively on the "bad" volatility.

• Sharpe Ratio — Annualised excess return divided by Std Dev; treats upside and downside volatility symmetrically. Generally, anything above 1.0 is considered good, above 1.5 is excellent.

• Sortino Ratio — Annualised excess return divided by Downside Deviation; rewards upside volatility and therefore often runs higher than the Sharpe when returns are positively skewed. This is even more telling than Sharpe since it only penalizes downside risk.

• Maximum Drawdown (Max DD) — Peak-to-trough drop during the test period; a gut-check for risk tolerance. For reference: The Alpha Engineer model portfolio (50 stocks) had a max drawdown of -36% during the same period, while the S&P 500 experienced -34%.

• Turnover — Estimated percentage of the portfolio traded per year; multiplied by the 1% per-trade cost gives the rough fee drag baked into these results. Higher turnover means more trading costs.

What the Data Tells Me

1. Breadth Matters

The larger the universe, the more freedom the model has to cherry-pick top-ranked names. The All-regions test leads almost every column—and boasts the standout Sharpe (1.62) and Sortino (2.22)—suggesting that factor robustness thrives on variety and that its risk-adjusted return profile is superior.

This makes intuitive sense: when we can fish in a larger pond, we have more opportunities to catch the biggest fish. My model can select the very best stocks from the global ranking regardless of which country they come from.

2. Narrow Pools — Two Different Stories

• Canada: With only 354 names and a 10-stock limit, turnover explodes to ~259% p.a. and risk-adjusted returns suffer. The smaller universe means fewer high-quality options, forcing more frequent trading as rankings shift.

• United Kingdom: Capping the book at 10 names does cut turnover to 146% p.a. but returns trail broader universes because the very best global-ranked stocks are simply not available within the UK market. However, the UK portfolio still delivers respectable performance with lower churn.

3. Understanding the U.S. Market Dynamics

The dynamic U.S. micro- and small-caps market generally screens pricier on value factors than Europe and exhibits less momentum persistence (more efficient market) and therefore stocks slip below the 90th-percentile sell bar sooner, driving a 262% annual turnover.

The trade-off: access to occasional mega-winners (if you are lucky). This explains the dramatic difference between the baseline and outlier-inclusive numbers. These "lottery tickets" can turbocharge returns but at the cost of higher volatility.

It's worth noting that U.S. returns have also benefited from the strong bull markets over the past 15 years that have lifted all boats, helping deliver great returns even when value and momentum factors appear less attractive compared to other markets. This broader market strength can compensate for the higher churn and potentially lower factor efficiency in the U.S. universe.

Recent market conditions have particularly challenged U.S. stocks, with a soft market for microcaps since the 2021 Covid boost and recently new tariff policies impacting performance of small company stocks considerably. This explains the stark -11.11% one-year and -14.80% YTD figures, compared to Europe's impressive 45.59% and 32.10% for the same periods.

4. Risk-Return Sweet Spots

All country baskets clear a Sortino of 1 with a big margin even after a generous 2% round-trip cost assumption. Europe excels over the 10-year window (with a 38.17% CAGR), but the all-regions portfolio is the clear all-rounder on both absolute and risk-adjusted metrics.

The data aligns with portfolio theory: while stocks across different geographical markets maintain some correlation, this correlation is typically lower than among stocks within the same region. This reduced correlation between regional markets improves the overall risk-reward profile. Being able to select from a multi-regional pool as regional factors shift gives the all-regions approach its edge.

Observations from the Back-test

1. Breadth and opportunity set. The back-test data suggests that broader opportunity sets produced better risk-adjusted results; the all-regions portfolio achieved the highest Sharpe (1.62) and Sortino (2.22) ratios. Both Europe-only and US-only strategies also delivered strong historical performance. As for any investing style in stocks, a time horizon of 5 years or longer appears necessary as the US results demonstrate.

2. Transaction cost impact. The back-test assumed 1% per-side transaction costs. With the high turnover observed (especially in the US portfolio at 262% annually), transaction costs significantly impact net returns. For example, reducing costs from 1% to 0.5% per trade in a portfolio with 262% turnover would theoretically add approximately 2.6% to annual returns in the historical simulation.

3. Sell rule considerations. For single-region portfolios in the back-test—particularly the U.S. portfolio—modifying the sell threshold from the 90th to the 85th percentile will reduce turnover. My previous study on parameter optimization explores this relationship in greater detail.

4. Drawdown characteristics. The back-test showed that all portfolio configurations experienced significant drawdowns ranging from 40-55% peak-to-trough, generally in line with their respective local market indices. The model portfolio (50 stocks) had a max drawdown of -36% during the same period, while the S&P 500 experienced -34%.

5. Portfolio size in smaller markets. The back-test results for smaller regions indicate that using fewer stocks (10 instead of 20) can potentially improve returns and risk-adjusted performance metrics, although results varied significantly between regions. The UK's 10-stock model demonstrated better turnover characteristics than Canada's.

Conclusion

This back-testing study examined the Alpha Engineer model's historical performance across different geographical markets during the test period from January 2010 through May 2025. The data shows that the all-regions approach produced the strongest overall results in terms of both absolute and risk-adjusted returns.

The test results illustrate notable differences in performance between regions, with significant divergence particularly evident in recent periods. Europe's considerable outperformance over the U.S. in the most recent year (+45.59% vs -11.11%) demonstrates how regional economic conditions and policies can influence returns in different market environments.

While the model generated positive long-term results across all tested markets during the study period, the variations in performance metrics, turnover, and risk characteristics between regions highlight the potential impact of geographical diversification in a systematic investment approach.

The historical data presented in this back-test should be viewed in context, with the understanding that past performance patterns may not continue in future market conditions.

Learn more about my investing framework:

• Read about my overall portfolio strategy.

• See the full details of my systematic method for stock picking.

The Alpha Engineer — Investing with a quantitative edge

Disclaimer: The Alpha Engineer shares insights from sources I believe are reliable, but I can't guarantee their accuracy—data's only as good as its inputs! This content (whether on Substack, via email newsletters, X, or elsewhere) is for informational and educational purposes only—it's not personalized investment advice. I'm not a registered investment advisor, just an engineer crunching numbers for alpha. My opinions are my own and may shift without notice. Investing involves risks, including the chance of losing money. Past performance, whether from back-testing or historical data, does not guarantee future results—outcomes can vary. So, please consult your financial advisor to see if any strategy fits your situation. Full disclosure: I may own positions in the securities I mention, as I actively manage my own portfolio based on these strategies.

The standard methodology described in My Investment Method works well as a baseline approach for most situations. However, individual investors have different constraints—portfolio size, time availability, transaction costs, and risk preferences. This analysis allows you to explore potential adjustments based on your personal situation.

It's important to note that this research presents historical backtested results rather than individual investment advice. The data shows how different parameter configurations performed in the past, but each investor should make their own decisions about implementation based on their unique circumstances.

The Reference Point

To establish a consistent baseline for this sensitivity analysis, I've used a reference model that's nearly identical to the standard Alpha Engineer model portfolio settings, with one modification: the removal of the 60-day minimum holding period. This allows stocks to be sold after just one week if they fall below the ranking threshold.

This modification creates minimal performance difference in backtesting compared to the standard model—primarily affecting turnover rather than returns. All percentage changes discussed below are measured relative to this reference model for analytical consistency.

The table below shows the historical performance metrics of the three benchmark models used in this analysis:

A Note on Interpretation

This analysis explores the historical impact of different parameters based entirely on backtested data. It's crucial to understand that backtested results are not predictive of future performance. Markets evolve constantly, and strategies that performed well in the past may not continue to do so.

Due to outliers in the historical data, differences of 2-3% in performance metrics should not be considered statistically significant. The standard deviation of approximately 20-25% across most variants reflects the model's tendency to capture multibaggers—a characteristic of the methodology. The downside deviation is much closer to the S&P 500 benchmark (see spreadsheet).

This research is presented to help understand potential trade-offs when customizing parameters, not to suggest optimal settings for future implementation.

1. The Effect of Portfolio Size

The backtested data shows how varying the number of stocks in a portfolio influenced historical performance metrics. Portfolios with 15-25 stocks achieved similar CAGR (Compound Annual Growth Rate) to the reference model with 50 positions.

The historical backtested results across different portfolio sizes were:

Several patterns emerged from the historical data:

• CAGR remained relatively strong across most portfolio sizes (with 20-stock portfolios showing the highest returns in the test period)

• Smaller portfolios (5-10 stocks) exhibited substantially higher volatility (29.53% standard deviation for 5 stocks vs. 21.34% for 50 stocks)

• Risk-adjusted returns (Sortino ratio) generally improved with increased diversification

• Portfolios with 15-25 stocks showed only moderately different risk metrics than the 50-stock reference model

The data suggests that historically, concentrating the strategy into 15-25 stocks maintained comparable returns while requiring fewer positions to monitor and manage.

Read more

The Non-energy Materials sector has followed a volatile path since 2022, initially seeing signs of life in 2023 as growth resumed and inflation cooled (while remaining moderately elevated), before facing headwinds in 2024. Since 2025 materials showed resilience in a market dominated by the word ‘tariff’. Now, with Trump's extraordinary tariffs causing market turbulence and raising stagflation concerns, the sector sits at a critical juncture.

The current market environment presents a complex picture for Materials stocks. The proposed tariffs could potentially trigger a stagflationary environment with mixed impacts across the sector. However, unless we experience an outright deflationary environment with minimal government intervention (an unlikely scenario), inflationary pressures will likely persist as governments rely on fiscal spending to support their economies, return to modest growth and reduce the impact of deglobalization.

A weakening USD could also reignite emerging markets and potentially spark a commodity boom. When we examine the data objectively, it's conceivable that large segments of the Materials sector could outperform over the next decade, especially compared to the post-financial crisis period where the sector largely underperformed financials and technology.

Quantitative Assessment

I conducted a 10-stock portfolio backtest of The Alpha Engineer model focusing exclusively on this sector. The methodology followed a straightforward, rules-based approach that can be replicated live using the weekly ranking:

• Filter to include only liquid stocks from the non-energy materials sector

• Consider only stocks with a ranking of 95 or higher for purchase

• Sell positions when rankings dropped below 90 (or when stocks disappeared from the ranking entirely)

The portfolio maintained generally 10 stocks except for brief periods. The results demonstrate clear outperformance compared to both the broader market and sector-specific ETFs:

• 23% CAGR for the Materials-only portfolio

• Compared to XLB (Materials Select Sector SPDR Fund) and S&P 500, both significantly lower: 5% and 12% respectively

• However, still below the 36% CAGR achieved by the full Model Portfolio

While it is not the best example due to past macro headwinds, there is a key insight here: sector-specific portfolios, while potentially strong performers, still benefit from the diversification provided by the complete model. This is because a diversified portfolio typically buys stocks with a higher average ranking than a portfolio limited to a specific sector.

Recent Performance Shift

Since 2023, we've observed a change in performance trajectory for the top material stocks, delivering 36% CAGR (including the market crash of April 3th and 4th). The YTD results are particularly strong. However, if growth suddenly stalls due to tariffs, we could experience downside pressure similar to 2018 or 2022. The near-term outlook remains uncertain.

Read more