The empirical evidence across all fourteen tables converges on three findings that directly address the shortcomings identified in the prior literature. The first concerns survivorship bias. Table 1 demonstrates that restricting the sample to institutions that remained in the S&P 500 throughout the full period consistently inflates measured performance across all strategies, but disproportionately so for systemic-risk-aware strategies. The CoSR portfolio with the more demanding crisis threshold shows an uplift of more than 51% in final wealth when moving from the bias-corrected baseline to the survivor-only dataset. This magnitude is large enough to alter the qualitative verdict on strategy superiority. The implication is that studies reporting strong outperformance for systemic-risk-based portfolios without controlling for sample selection should be interpreted with caution: a meaningful portion of the documented excess returns may reflect the exclusion of distressed or delisted firms that bore the highest systemic risk and the lowest subsequent returns. Taken together, the evidence suggests that systemic risk matters for portfolio allocation, but only when treated carefully. The key is not simply to “avoid systemic risk” in the aggregate. Rather, investors should distinguish between the destabilizing contagion component and the compensated common-exposure component. Once survivorship bias is controlled and systemic risk is decomposed, the empirical picture becomes much clearer: investors are rewarded for minimizing contagion, while moderate exposure to common factors may be beneficial in tranquil states. The second finding concerns the value of decomposition. The evidence makes clear that contagion and common exposure do not affect portfolio performance in the same direction, and that treating systemic risk as a single scalar can obscure offsetting forces. The minimum-contagion portfolio, constructed using the Diebold-Yilmaz connectedness index, generates the highest full-sample terminal wealth, the highest Sharpe and Sortino ratios, and the best crisis-period downside protection among all strategies tested. This performance gap is not merely economic: Tables 3 and 4 show that it is statistically significant under both HAC-adjusted parametric tests and moving-block bootstrap inference. The certainty-equivalent return analysis in Table 5 further confirms that this performance translates into welfare gains for investors across a wide range of risk-aversion coefficients, with the minimum-contagion strategy dominating for all moderate risk preferences. By contrast, the common-exposure and idiosyncratic-exposure signals point in the opposite direction: investors benefit from holding assets with high common exposure and low idiosyncratic risk, not from avoiding them. The minimum-idiosyncratic portfolio is the strongest performer in Table 9, and the maximum-common portfolio ranks second. These opposing findings across components confirm that aggregate systemic risk measures can yield misleading results by netting out effects that individually point in different directions. Conclusion This study examines the portfolio implications of systemic risk for large U.S. financial institutions from 2006 to 2025, focusing on three key limitations in prior research: survivorship bias in sample construction, the aggregation of distinct systemic risk components into a single index, and the lack of formal statistical testing to substantiate claims of strategy superiority. By addressing these issues concurrently, the analysis demonstrates that survivorship-corrected, decomposition-based portfolio strategies yield results that are both economically significant and statistically robust. Notably, these findings diverge substantially from those produced by aggregate, survivor-only analyses. The first finding demonstrates that survivorship bias is a quantitatively significant concern in systemic-risk-based portfolio evaluation. When strategies are applied only to institutions that remained continuously in the S&P 500, measured performance is substantially higher than when the same strategies are evaluated using a point-in-time baseline that includes subsequently delisted firms. This distortion is most pronounced for systemic-risk-aware strategies, especially those with stringent crisis-trigger thresholds, where final wealth is overstated by more than 50%. This outcome aligns with the observation that firms exiting the index typically exhibit both poor subsequent returns and elevated systemic risk exposure, so their exclusion artificially inflates the returns of portfolios favoring safer institutions. The implication for the literature is clear: the apparent outperformance of systemic-risk strategies in survivor-only samples cannot be accepted without survivorship-corrected replication. The second finding indicates that contagion is the component of systemic risk most strongly penalized by investors. Among all strategies tested on the survivorship-corrected baseline, the minimum-contagion portfolio achieves the highest terminal wealth, the highest Sharpe and Sortino ratios, and the strongest downside protection during stress episodes. Its compound annual growth rate (CAGR) of 20.26% substantially exceeds those of the equal-weight (9.28%), global minimum variance (GMV) (7.93%), and maximum-contagion (6.19%) portfolios. These differences are supported by HAC-adjusted mean-difference tests and moving-block bootstrap inference, which confirm that the minimum-contagion strategy significantly outperforms GMV, Sharpe, and equal-weight benchmarks in both mean return and Sharpe ratio. Welfare analysis further demonstrates that investors with moderate risk aversion would prefer the minimum-contagion portfolio on a certainty-equivalent basis. The underlying economic mechanism is clear: contagion amplifies shocks through interconnected balance sheets, undermining diversification when it is most needed. Portfolios that systematically avoid the most contagion-exposed institutions are thus rewarded with higher risk-adjusted returns. The third finding reveals that common exposure and idiosyncratic exposure influence portfolio construction in opposite directions. Investors tend to reward common exposure: the minimum-idiosyncratic portfolio achieves the highest full-sample CAGR among the common and idiosyncratic exposure strategies at 16.45%, while the maximum-common portfolio ranks second at 11.95%, both exceeding the equal-weight benchmark. However, this reward is contingent on market conditions. During crisis periods, maximum-common and minimum-idiosyncratic strategies experience the largest losses, with drawdowns and volatility surpassing those of contagion-based or conservative strategies. In contrast, during calm periods, the minimum-idiosyncratic strategy achieves a CAGR of 37.20% and the maximum-common strategy follows at 30.05%, both significantly above benchmarks. This regime dependence underscores the inadequacy of aggregate systemic-risk measures for portfolio decision-making, as the common and contagion components of aggregate risk move in opposite directions across market states, and averaging them obscures critical information for portfolio selection. Collectively, these findings refine the prescriptive guidance for practitioners and researchers. Systemic-risk-aware portfolio management should not aim to indiscriminately reduce aggregate systemic risk exposure. Instead, investors should determine which channel of systemic risk is contributing most to portfolio vulnerability at any given time. In stable market conditions, holding assets with substantial common factor exposure and low idiosyncratic risk is advantageous. During periods of emerging systemic stress, minimizing contagion exposure becomes paramount. A portfolio framework that differentiates between these channels, monitors their relative contributions over time, and adjusts asset weights accordingly is more likely to achieve both strong long-term growth and effective downside protection than one that treats systemic risk as a single, aggregate metric. This study contributes to the literature in several ways. For systemic risk research, it provides the first portfolio-choice analysis that simultaneously controls for survivorship bias and disaggregates systemic risk into contagion, common, and idiosyncratic components, using a sample that spans major stress episodes of the past two decades. For the portfolio optimization literature, it demonstrates that survivorship-corrected, decomposition-based strategies can yield statistically significant improvements over traditional benchmarks, addressing a gap highlighted in recent evidence reviews. More broadly, the study underscores the importance of rigorous sample construction and formal inference as essential complements to graphical and economic analysis when evaluating new investment strategies. Several limitations warrant consideration. The sample is limited to large U.S. financial institutions within the S&P 500, so the findings may not generalize to smaller firms, non-financial sectors, or equity markets in other countries with different systemic risk dynamics and institutional structures. The results are also sensitive to the specific proxies selected for contagion and connectedness, as well as to the procedures used for identifying turmoil windows and setting rebalancing rules. Alternative threshold choices could affect the reported magnitudes, although the directional conclusions remain robust across various checks. Future research could extend this framework by exploring alternative decomposition methodologies, testing the approach in international and multi-asset contexts, incorporating real-time or high-frequency data to refine regime identification, and investigating whether dynamic portfolio rules that adapt to changing systemic conditions can further enhance performance relative to the static strategies analyzed here. In sum, the analysis in this paper demonstrates that the portfolio relevance of systemic risk is substantive, but its value becomes apparent only after correcting for survivorship bias and distinguishing among its underlying components. A portfolio strategy that minimizes contagion provides the strongest combination of long-term returns, statistical support, and investor welfare, while common exposure is primarily rewarded during stable periods. These findings help bridge the gap between systemic risk literature and portfolio theory by illustrating that investors are concerned not only with the level of systemic risk, but also with the specific mechanisms through which that risk enters the financial system.

The empirical findings from all fourteen tables converge to establish three core results that directly address the methodological limitations identified in existing literature. The first pertains to survivorship bias. Table 1 reveals that constraining the sample to institutions maintaining continuous S&P 500 membership throughout the entire observation period systematically overstates performance metrics across all strategies, with particularly pronounced effects for systemic-risk-aware approaches. The CoSR portfolio employing the more stringent crisis threshold exhibits a final wealth increase exceeding 51% when transitioning from the bias-adjusted baseline to the survivor-restricted dataset. This distortion proves substantial enough to fundamentally alter conclusions regarding strategy effectiveness. These results indicate that research documenting superior performance for systemic-risk-based portfolios without addressing sample selection effects should be evaluated cautiously, as a considerable portion of reported excess returns may stem from excluding financially distressed or delisted firms that exhibited the greatest systemic risk exposure and poorest subsequent performance. Collectively, the evidence establishes that systemic risk represents a meaningful consideration for portfolio construction, provided it receives appropriate treatment. The critical insight involves moving beyond simplistic aggregate systemic risk avoidance toward distinguishing between the destabilizing contagion element and the compensated common-exposure element. When survivorship bias receives proper control and systemic risk undergoes decomposition, the empirical landscape becomes substantially clearer: investors receive compensation for contagion minimization, while moderate common factor exposure may prove advantageous during periods of market stability. The second result emphasizes the importance of decomposition analysis. The evidence demonstrates conclusively that contagion and common exposure exert divergent effects on portfolio performance, and that treating systemic risk as a unified scalar measure can mask counteracting influences. The minimum-contagion portfolio, developed using the Diebold-Yilmaz connectedness framework, achieves the highest terminal wealth across the full sample period, the superior Sharpe and Sortino ratios, and the most effective crisis-period downside protection among all examined strategies. This performance advantage extends beyond economic significance: Tables 3 and 4 demonstrate statistical significance under both HAC-corrected parametric testing and moving-block bootstrap procedures. The certainty-equivalent return examination in Table 5 validates that this performance translates into measurable welfare improvements for investors spanning diverse risk-aversion parameters, with the minimum-contagion strategy proving optimal for all moderate risk preferences. Conversely, the common-exposure and idiosyncratic-exposure indicators suggest opposite conclusions: investors derive benefits from maintaining assets characterized by elevated common exposure and reduced idiosyncratic risk rather than avoiding such positions. The minimum-idiosyncratic portfolio demonstrates the strongest performance in Table 9, while the maximum-common portfolio achieves second-place ranking. These contrasting results across components validate that aggregate systemic risk measures can produce misleading conclusions by offsetting effects that individually indicate different optimal directions. Conclusion This investigation analyzes the portfolio implications of systemic risk for major U.S. financial institutions spanning 2006 to 2025, addressing three fundamental limitations in existing research: survivorship bias in sample construction, the consolidation of distinct systemic risk components into unified indices, and the absence of rigorous statistical testing to validate claims of strategy superiority. Through concurrent treatment of these issues, the analysis establishes that survivorship-adjusted, decomposition-based portfolio strategies produce results that achieve both economic meaningfulness and statistical reliability. These findings represent substantial departures from those generated through aggregate, survivor-restricted analyses. The initial finding establishes survivorship bias as a quantitatively meaningful concern in systemic-risk-based portfolio assessment. When strategies apply exclusively to institutions maintaining continuous S&P 500 inclusion, measured performance substantially exceeds results obtained when identical strategies undergo evaluation using point-in-time baselines incorporating subsequently delisted firms. This distortion reaches maximum intensity for systemic-risk-aware strategies, particularly those employing stringent crisis-trigger criteria, where final wealth experiences overstatement exceeding 50%. This pattern aligns with the recognition that index-exiting firms typically demonstrate both poor subsequent returns and heightened systemic risk exposure, rendering their exclusion artificially beneficial for portfolios favoring safer institutions. The literature implications prove unambiguous: apparent outperformance of systemic-risk strategies in survivor-restricted samples requires survivorship-corrected validation. The second finding indicates that contagion represents the systemic risk component receiving the strongest investor penalty. Among all strategies examined using the survivorship-corrected baseline, the minimum-contagion portfolio attains the highest terminal wealth, superior Sharpe and Sortino ratios, and the most robust downside protection during stress periods. Its compound annual growth rate of 20.26% substantially surpasses those achieved by equal-weight (9.28%), global minimum variance (7.93%), and maximum-contagion (6.19%) portfolios. These differentials receive support through HAC-adjusted mean-difference testing and moving-block bootstrap procedures, confirming that the minimum-contagion strategy significantly exceeds GMV, Sharpe, and equal-weight benchmarks in both mean return and Sharpe ratio metrics. Welfare analysis additionally demonstrates that investors exhibiting moderate risk aversion would favor the minimum-contagion portfolio on a certainty-equivalent basis. The underlying economic rationale proves transparent: contagion amplifies shocks through interconnected balance sheet structures, compromising diversification precisely when protection becomes most critical. Portfolios systematically avoiding the most contagion-exposed institutions consequently receive compensation through enhanced risk-adjusted returns. The third finding demonstrates that common exposure and idiosyncratic exposure affect portfolio construction through opposing mechanisms. Investors typically reward common exposure: the minimum-idiosyncratic portfolio attains the highest full-sample compound annual growth rate among common and idiosyncratic exposure strategies at 16.45%, while the maximum-common portfolio achieves second position at 11.95%, both exceeding the equal-weight benchmark. However, this compensation depends on prevailing market conditions. During crisis episodes, maximum-common and minimum-idiosyncratic strategies suffer the largest losses, with drawdowns and volatility exceeding those of contagion-based or conservative approaches. Conversely, during tranquil periods, the minimum-idiosyncratic strategy achieves a compound annual growth rate of 37.20% while the maximum-common strategy follows at 30.05%, both significantly surpassing benchmarks. This regime sensitivity highlights the inadequacy of aggregate systemic-risk measures for portfolio decision-making, as the common and contagion components of aggregate risk exhibit opposing behaviors across market states, with their averaging obscuring essential portfolio selection information. These findings collectively refine prescriptive guidance for practitioners and researchers. Systemic-risk-aware portfolio management should avoid indiscriminate aggregate systemic risk reduction. Instead, investors should identify which systemic risk channel contributes most significantly to portfolio vulnerability at specific times. During stable market conditions, maintaining assets with substantial common factor exposure and minimal idios