146 | The False Promise of Drawdown Rules: New Evidence and a Better Framework November 2025 greater losses than a single, managed drawdown would have produced. This counterintuitive result warrants systematic investigation. This article examines the conditions under which drawdown rules work as intended versus when they fail, providing a framework for more effective implementation. We approach this analysis with a focus on practical applications, avoiding unnecessarily complex econometric models in favor of clear, actionable insights. Our goal is to develop guidelines that portfolio managers can readily incorporate into existing risk management systems, with principles that can be understood and applied across different market environments and investment strategies. LITERATURE REVIEW The management of drawdowns represents a complex challenge in investment management that has attracted attention from both academics and practitioners. This literature spans theoretical models, empirical studies, practical implementations, and behavioral insights, collectively providing a nuanced picture of drawdown management’s evolution. The theoretical underpinnings of drawdown management derive from optimal stopping problems in finance. Grossman and Zhou (1993) developed a foundational framework for addressing when investors should optimally exit positions based on cumulative losses. Although their work did not specifically target drawdown management, it established mathematical principles that would later be applied to this domain. Building on this foundation, Carr, Zhang, and Hadjiliadis (2011) expanded the theory to explicitly incorporate drawdown constraints, demonstrating that these constraints effectively function as path-dependent options within a portfolio context. The practitioner community has generally embraced structured approaches to drawdown management. Ang (2014) highlights the value of disciplined risk management frameworks that incorporate drawdown constraints, particularly for institutional portfolios with specific liability structures or regulatory requirements. These approaches gained increased adoption following the 2008 financial crisis, as investors sought more systematic methods to control downside risk. When properly implemented, these rules can provide important guardrails for portfolios during periods of market stress. Despite their intuitive appeal, several researchers have identified limitations in mechanical drawdown rules. Chekhlov, Uryasev, and Zabarankin (2005) presented evidence that rigid implementation can be counterproductive, especially for strategies that require sustained exposure to risk premiums to generate returns. More recently, Kaminski and Lo (2014) examined when stop-loss rules (conceptually similar to drawdown thresholds) effectively protect capital versus when they trigger unnecessary exits. Their research indicates these rules’ efficacy depends critically on market regime, return autocorrelation, and implementation parameters—highlighting the need for context-aware application rather than universal rules. The behavioral finance literature offers compelling insights into why drawdown rules remain prevalent despite mixed empirical evidence. Kahneman (2011) and Thaler (2015) suggest that investors’ preference for explicit risk management rules stems from loss aversion and the psychological need for agency during uncertainty. This desire for control can lead to intervention even when patience might yield better outcomes—a cognitive bias that influences both individual and institutional decision making. This literature collectively suggests that although drawdown management serves an important function in portfolio risk control, its implementation requires nuance November 2025 The Journal of Portfolio Management | 147 beyond simple thresholds. Our work builds on these insights by empirically evaluating when traditional drawdown rules succeed versus when they fail, and by developing a framework that maintains their intuitive appeal while addressing their documented limitations. CONDITIONAL PROBABILITY AND THE NEED FOR A COHERENT METRIC Understanding drawdowns requires examining the conditional probabilities that underlie risk management decisions. When practitioners implement drawdown rules, they are implicitly making assumptions about these probabilities—often without recognizing the biases these assumptions introduce. Drawdowns represent the decline from previous peak portfolio value and possess unique properties that distinguish them from other risk measures. They are inherently path-dependent, meaning two portfolios with identical current values may have entirely different drawdowns based on their history. They persist until a new maximum is reached, creating memory in the measurement. And they capture both magnitude and duration aspects of market declines, making them intuitively appealing for risk management. The fundamental question driving drawdown rules is straightforward but profound: Given a current drawdown of magnitude x%, what is the probability of experiencing a larger drawdown in the future? This conditional probability question lies at the heart of all drawdown-based decision frameworks. The origin of most practitioners’ intuition about drawdowns deserves closer examination. Many investment professionals develop their risk management instincts through formative experiences with US large-cap equities, particularly during significant market corrections. When managing portfolios tracking indexes, such as the S&P 500, drawdown rules often appear to work as intended—creating a powerful impression that generalizes to other contexts. For drawdown rules to be rational, the conditional probability of experiencing a drawdown of y% must increase more rapidly than one would expect under normal distribution assumptions as the initial drawdown x% grows larger. Despite claims of sophisticated risk models, most investment frameworks—whether using Sharpe ratios, Sortino ratios, volatility measures, or correlation matrixes—implicitly rely on normality assumptions. For all practical purposes, any time we use standard deviation for risk assessment, we are effectively assuming some form of normal or lognormal distribution. These assumptions often fail to capture how drawdowns actually behave, particularly during market stress periods—precisely when accurate risk assessment becomes most critical. This generalization proves problematic when applied across different asset classes, market regimes, and investment strategies. The conditional probability structure that works for large-cap US equities may not translate to other assets or portfolio structures, creating systematic biases in risk management. These theoretical insights translate directly to practical challenges for portfolio managers: § Risk assessment: Conventional drawdown metrics often underestimate risk during severe market stress periods and overestimate risk during minor corrections. § Position sizing: Traditional drawdown-based rules can lead to counterintuitive outcomes, increasing exposure when the probability of further losses is highest. 148 | The False Promise of Drawdown Rules: New Evidence and a Better Framework November 2025 § Performance evaluation: Conventional measures can rank catastrophic performance ahead of moderate underperformance, distorting risk assessment precisely when accuracy matters most. To illustrate these failures, our analysis of the S&P 500 ETF (SPY) during the COVID crash reveals dramatic inconsistencies in traditional metrics. For the 22-trading-day period from February 21, 2020–March 23, 2020, SPY produced an annualized return of −99% with a maximum drawdown of −33%. The traditional drawdown-adjusted performance (DAP) ratio—calculated as annualized return divided by maximum drawdown—yielded −2.96, ranking this catastrophic period at merely the 25th percentile of risk. Even more troubling, the period from February 11, 2020–March 12, 2020, with slightly better performance (annualized return of −96.7%, maximum drawdown of −26.7%) received a worse DAP score of −3.63, ranking it in the 22nd percentile. Similar inconsistencies appear in other common risk metrics. For the period from February 24, 2020–March 24, 2020, with a −97% annualized return and −32.7% maximum drawdown, Sharpe ratio readings fell only in the bottom 5% of their historical distribution, whereas Sortino ratios reached only the bottom 3%—nowhere near reflecting the true severity of the fastest market drawdown in modern history. This measurement inconsistency stems from a fundamental flaw: When returns are negative, traditional metrics produce counterintuitive results in which deeper drawdowns paradoxically improve the ratio. To address this issue while preserving the intuitive appeal of return-todrawdown measures, we propose the coherent drawdown-adjusted performance (CDAP) metric: r ×|d|−sgn(r ) (1) where r represents return, |d| is the absolute value of maximum drawdown, and sgn refers to the sign function sgn(r), which equals +1 when r > 0, and −1 when r < 0. This formulation ensures that when returns are positive, CDAP divides by drawdown (higher drawdown = lower CDAP), and when returns are negative, CDAP multiplies by drawdown (higher drawdown = lower CDAP). The result maintains logical consistency across all market conditions. The effectiveness of this approach becomes evident when applied to the COVID crash periods. For all the periods previously mentioned, CDAP correctly ranked them in the bottom 0.2% of all observations—precisely where risk metrics should place such extraordinary market stress. This consistent behavior extends across the entire set of trading periods from February 11, 2020–March 12, 2020, through March 6, 2020–April 6, 2020, with CDAP-values consistently in the bottom 0.4% of all observations, accurately reflecting the extreme risk environment. Our empirical investigation demonstrates that although the drawdown rule intuition may occasionally work in specific contexts (particularly with S&P 500 investments during certain periods), it fails to generalize reliably across different market environments, asset classes, and investment strategies. In the next section, we visualize this phenomenon through probability ratio analyses across different assets, as shown in Exhibits 1 and 2. These visualizations reveal how the S&P 500’s conditional probability structure—which appears to support drawdown rules—differs markedly from that of other major exchange-traded funds (ETFs) such as EWJ (Japan) and EWU (United Kingdom), helping to explain why intuition derived from US equity markets often leads to suboptimal decisions when applied elsewhere. 150 | The False Promise of Drawdown Rules: New Evidence and a Better Framework November 2025 EMPIRICAL EVIDENCE FROM LONG PORTFOLIOS To empirically test our theoretical understanding of drawdown behavior, we analyze how drawdown thresholds perform across various equity markets. We begin with the S&P 500, which serves as the foundation for many practitioners’ risk management intuition, before examining whether these patterns generalize to international markets. Exhibit 1 presents a revealing analysis of the SPDR S&P 500 ETF Trust (SPY). This chart shows the ratio between actual drawdown probabilities from historical data versus theoretical probabilities assuming a normal distribution with identical mean and standard deviation. Each colored line represents a different starting drawdown condition: The pink line shows probabilities given a current drawdown of 0%, blue represents a current drawdown of 20%, and so on. The patterns in this exhibit are striking. For smaller eventual drawdowns (left side of chart), actual historical frequencies are lower than normal distribution would predict—suggesting that modest drawdowns are less common than theory would indicate. As eventual drawdown size increases (moving right on the chart), however, actual frequencies become multiples of what normal distribution theory predicts. For an eventual drawdown of approximately 45%, the true probability is up to seven times higher than standard models would expect, regardless of the starting drawdown condition. This S&P 500 analysis seemingly validates the intuition behind drawdown rules: If large drawdowns occur substantially more frequently than normal distribution theory predicts, implementing cutoffs might protect against tail risks that standard models underestimate. This pattern, however—and this represents a critical finding—does not generalize reliably to other markets. Exhibit 2 presents the same probability ratio analysis for iShares MSCI Japan ETF (EWJ) and iShares MSCI United Kingdom ETF (EWU). Although the Japan ETF shows some pattern similarity to SPY, the probability ratios for the UK ETF demonstrate no clear pattern. Our comprehensive analysis (Supplemental Exhibits S1–S3) in the online supplement examines what happens as you vary the drawdown cutoff threshold for various ETFs, with rows colored green where drawdown cutoffs improved performance. As documented there, for neither EWJ nor EWU does a drawdown cutoff yield better results, despite EWJ showing probability patterns somewhat similar to SPY. The scarcity of green-highlighted results in our comprehensive analysis is striking—drawdown rules rarely outperform simple buy-and-hold strategies across the diverse set of ETFs we examined. Exhibit 3 further illustrates this finding by plotting the mean CDAP difference between drawdown-rule strategies and buy-and-hold approaches across various threshold levels. This visualization demonstrates that drawdown cutoffs rarely improve performance, and when they do, the magnitude of improvement is typically modest. These empirical results reveal an important disconnect: Even when conditional probability patterns seem to support drawdown rules (as with SPY and somewhat with EWJ), the practical implementation of such rules frequently fails to deliver improved risk-adjusted performance. This suggests that additional factors beyond conditional probability structures—such as market regime, recovery patterns, and reentry timing—significantly impact the effectiveness of drawdown-based strategies. For portfolio managers, this evidence carries important implications: Risk management approaches developed through experience with US large-cap equities may not translate effectively to other markets, even when those markets display somewhat similar probability structures. The data suggests that practitioners should approach drawdown thresholds with caution, recognizing their limited effectiveness across diverse market environments. November 2025 The Journal of Portfolio Management | 151 EXHIBIT 3 Mean CDAP Difference vs. Buy and Hold for Various Drawdown Cutoffs: Long Only Mean Difference in CDAP for Drawdown Cutoff Thresholds Ranging from 5% to 35% versus B&H 0.05 0.00 Mean CDAP Difference –0.05 –0.10 EWA,Mar96:May25 EWC,Mar96:May25 EWD,Mar96:May25 EWG,Mar96:May25 EWH,Mar96:May25 EWI,Mar96:May25 EWJ,Mar96:May25 EWK,Mar96:May25 EWL,Mar96:May25 EWM,Mar96:May25 EWN,Mar96:May25 EWO,Mar96:May25 EWP,Mar96:May25 EWQ,Mar96:May25 EWS,Mar96:May25 EWT,Jun00:May25 EWU,Mar96:May25 EWW,Mar96:May25 EWX,May08:May25 EWY,May00:May25 EWZ,Jul00:May25 IWM,May00:May25 QQQ,Mar99:May25 SPY,Feb93:May25 TLT,Jul02:May25 XBI,Feb06:May25 XLB,Dec98:May25 XLE,Dec98:May25 XLF,Dec98:May25 XLI,Dec98:May25 XLK,Dec98:May25 XLP,Dec98:May25 XLU,Dec98:May25 XLV,Dec98:May25 XLY,Dec98:May25 XME,Jun06:May25 XRT,Jun06:May25 EMPIRICAL EVIDENCE FROM LONG–SHORT PORTFOLIOS Having examined traditional long-only equity portfolios, we now investigate whether drawdown rules offer advantages for long–short strategies. This analysis is particularly valuable since long–short portfolios often require more active risk management and are commonly used by sophisticated investors. For this investigation, we construct various long–short pairs using ETFs, analyzing their performance both over their entire history and during specific periods when particular investment theses would have been most relevant. This approach allows us to separate general statistical patterns from context-specific performance. Importantly, we construct these pairs with complete hindsight—not to create investment strategies, but to provide the most favorable possible conditions for drawdown rules to demonstrate value. Our initial analysis examines these long–short pairs across their entire history. As documented in the online supplement (Supplemental Exhibit S2), drawdown cutoffs appear to improve performance in most cases. Exhibit 4 visualizes this finding, showing the mean CDAP difference between drawdown-rule implementation and simple buy-and-hold approaches. This initially suggests that drawdown cutoffs might be beneficial for long–short portfolios. 152 | The False Promise of Drawdown Rules: New Evidence and a Better Framework November 2025 EXHIBIT 4 Mean CDAP Difference vs. Buy and Hold for Various Drawdown Cutoffs: Long-Short Mean Difference in CDAP for Drawdown Cutoff Thresholds Ranging from 5% to 35% versus B&H 0.10 Mean CDAP Difference 0.05 0.00 –0.05 XLK-XLU,Dec98:May25 XLE-XLK,Dec98:May25 XLV-XLF,Dec98:May25 QQQ-XLB,Mar99:May25 XME-XRT,Jun06:May25 XLY-XLP,Dec98:May25 EWH-EWZ,Jul00:May25 XLF-XLU,Dec98:May25 EWJ-EWY,May00:May25 XBI-XLV,Feb06:May25 EWU-EWP,Mar96:May25 XLE-XLU,Dec98:May25 QQQ-IWM,May00:May25 Further investigation reveals, however, an important contextual factor: As shown in Exhibit 5, most of these long–short combinations generate minimal returns when evaluated across their entire history. This makes intuitive sense—long–short pairs are typically implemented only when a specific investment thesis is active, not as permanent allocations. This observation raises a critical question: Does the apparent benefit of drawdown rules for these portfolios stem merely from limiting losses in generally unprofitable combinations? To answer this, we analyze the same pairs during specific time periods when their underlying investment theses would have been most compelling, as detailed in Exhibit 6. During these thesis-specific periods (shown in Exhibit 7), the benefits of drawdown rules largely disappear. The online supplement (Supplemental Exhibit S3) documents this in detail, showing a significant reduction in instances in which drawdown cutoffs improve performance. Even more revealing is the inconsistency of any apparent benefits. Examining standard 5% and 10% drawdown thresholds, only four pairs (XME-XRT, EWH-EWZ, XBIXLV, and EWU-EWP) show improvement. This improvement demonstrates, however, remarkable instability across different threshold levels. For XBI-XLV, the benefit disappears beyond a 5% threshold. For EWU-EWP, it vanishes above 10%. For XME-XRT, improvements cease beyond 15%. Even the strongest performer, EWH-EWZ, loses its advantage beyond a 30% threshold. Some pairs, such as XLK-XLU, show improvement November 2025 The Journal of Portfolio Management | 153 EXHIBIT 5 Annualized Return for Various Long–Short Strategies Profitability of Various Long vs. Short ETF Pairs 0.02 0.00 Annualized Return –0.02 –0.04 –0.06 –0.08 –0.10 XLK-XLU,Dec98:May25 XLE-XLK,Dec98:May25 XLV-XLF,Dec98:May25 QQQ-XLB,Mar99:May25 XME-XRT,Jun06:May25 XLY-XLP,Dec98:May25 EWH-EWZ,Jul00:May25 XLF-XLU,Dec98:May25 EWJ-EWY,May00:May25 XBI-XLV,Feb06:May25 EWU-EWP,Mar96:May25 XLE-XLU,Dec98:May25 QQQ-IWM,May00:May25 EXHIBIT 6 Selected ETF Long–Short Pairs with Justification for the Trade Long ETF Short ETF Start Date End Date Market Context XLK XLU Jan. 2019 Dec. 2021 Tech outperformance during COVID XLE XLK Jan. 2022 Dec. 2022 Energy surge during inflation XLV XLF Jan. 2008 Dec. 2008 Healthcare stability versus financial crisis QQQ XLB Jan. 2017 Dec. 2019 Tech boom versus materials underperformance XME XRT Jan. 2021 Dec. 2021 Metals surge versus retail weakness XLY XLP Mar. 2020 Dec. 2021 Discretionary versus staples during COVID recovery EWH EWZ Jan. 2021 Dec. 2022 Hong Kong versus Brazil during China reopening XLF XLU Jan. 2016 Dec. 2018 Financials versus utilities during rate hikes EWJ EWY Jan. 2018 Dec. 2019 Japan versus Korea during trade tensions XBI XLV Jan. 2019 Dec. 2020 Biotech versus healthcare during vaccine race EWU EWP Jul. 2016 Dec. 2017 United Kingdom versus Spain during Brexit aftermath XLE XLU Jan. 2021 Dec. 2022 Energy versus utilities during inflation QQQ IWM Jan. 2018 Dec. 2019 Large tech versus small caps during tech dominance at isolated thresholds (15%) but not at others, suggesting statistical noise rather than robust improvement. Exhibit 8 visualizes this instability, showing that across most threshold levels and pairs, drawdown rules either provide no significant benefit or actively reduce performance. In the few cases in which benefits appear, they tend to be modest and inconsistent across threshold levels. EXHIBIT 8 EXHIBIT 7 Mean CDAP Difference Annualized Return –0.4 –0.3 –0.2 –0.1 0.0 0.0 0.2 0.4 0.6 0.8 1.0 XLK-XLU,Jan19:Dec21 XLK-XLU,Jan19:Dec21 XLE-XLK,Jan22:Dec22 XLE-XLK,Jan22:Dec22 XLV-XLF,Jan08:Dec08 XLV-XLF,Jan08:Dec08 QQQ-XLB,Jan17:Dec19 QQQ-XLB,Jan17:Dec19 XME-XRT,Jan21:Dec21 XME-XRT,Jan21:Dec21 XLY-XLP,Mar20:Dec21 Mean Difference in CDAP for Drawdown Cutoff Thresholds XLY-XLP,Mar20:Dec21 EWH-EWZ,Jan21:Dec22 EWH-EWZ,Jan21:Dec22 XLF-XLU,Jan16:Dec18 XLF-XLU,Jan16:Dec18 154 | The False Promise of Drawdown Rules: New Evidence and a Better Framework Annualized Return for Various Long–Short Strategies for Specific Periods EWJ-EWY,Jan18:Dec19 EWJ-EWY,Jan18:Dec19 XBI-XLV,Jan19:Dec20 XBI-XLV,Jan19:Dec20 Profitability of Various Long vs. Short ETF Pairs EWU-EWP,Jul16:Dec17 EWU-EWP,Jul16:Dec17 Ranging from 5% to 35% versus B&H XLE-XLU,Jan21:Dec22 XLE-XLU,Jan21:Dec22 QQQ-IWM,Jan18:Dec19 QQQ-IWM,Jan18:Dec19 Mean CDAP Difference vs. Buy and Hold for Various Long–Short Strategies for Specific Periods November 2025 November 2025 The Journal of Portfolio Management | 155 These findings substantially challenge the case for drawdown rules in long– short investing. Even in these highly favorable conditions—constructed with perfect hindsight during periods when specific investment theses should work—drawdown cutoffs fail to provide consistent improvement. The inconsistency of results across threshold levels further suggests that any observed benefits likely reflect statistical noise rather than fundamental risk management principles. This evidence reinforces our earlier findings from long–only portfolios: Despite their intuitive appeal, simple drawdown thresholds rarely provide reliable risk-adjusted performance benefits. Even for sophisticated long–short portfolios during their most favorable implementation periods, drawdown rules generally fail to deliver meaningful improvement. A FRAMEWORK FOR EFFECTIVE DRAWDOWN MANAGEMENT Our empirical findings demonstrate that traditional drawdown cutoff rules frequently fail to deliver their intended benefits. This raises an important question for practitioners: What approaches might work better? This section develops a constructive framework for more effective drawdown management. Conceptual Limitations of Simple Rules The most obvious alternative to binary cutoffs—simply scaling positions as drawdowns increase—unfortunately suffers from the same fundamental limitation. Any continuous scaling rule can be mathematically decomposed into a combination of discrete thresholds. Since these threshold-based approaches consistently underperform, their weighted combinations are similarly unlikely to provide systematic benefits. A more effective approach recognizes that pre-trade risk management offers greater benefits than reactive drawdown rules. Specifically, limiting position sizes, managing overall portfolio leverage, and establishing appropriate time horizons before initiating positions avoid the performance costs associated with mid-trade interventions. These ex ante controls allow positions to express their intended risk– return characteristics without imposing artificial constraints that can disrupt their performance lifecycle. Elements of an Enhanced Framework When post-initiation risk management is necessary, several enhancements can substantially improve upon simple drawdown thresholds: 1. Multivariate response functions: Rather than binary responses to single metrics, effective risk management incorporates multiple risk indicators. For equity positions, these might include yield curve dynamics, credit spread movements, and volatility regime assessments alongside drawdown metrics. This approach allows for graduated responses proportional to genuine risk conditions. 2. Market context integration: The interpretation of any drawdown should vary with market conditions. A 10% drawdown during periods of market stress (elevated Chicago Board Options Exchange (CBOE) Volatility Index (VIX), widening credit spreads) carries different implications than the same drawdown during normal conditions. During stress periods, drawdowns typically show greater persistence and cross-asset correlation, whereas normal periods more frequently exhibit mean reversion. 156 | The False Promise of Drawdown Rules: New Evidence and a Better Framework November 2025 EXHIBIT 9 3. Cross-asset confirmation: Isolated drawdowns in Cross-Asset Confirmation Matrix single positions often represent different risks than simultaneous drawdowns across multiple Scenario Primary Asset CDAP Related Assets CDAP assets or strategies. The former may indicate Isolated Risk <5th percentile >10th percentile idiosyncratic factors that could represent potenConfirmed Risk <5th percentile <10th percentile tial opportunities, whereas the latter more likely Systemic Risk <5th percentile <5th percentile signals systematic risk requiring reduction. The confirmation logic is summarized in Exhibit 9. CDAP provides a foundation for implementing these enhancements in a coherent framework that correctly ranks risk across both positive- and negative-return environments. CDAP-Based Risk Classification System Based on our empirical analysis using 22-day measurement periods (approximately one month in calendar time), we propose the following percentile-based classification system: § Critical risk: CDAP below 1st percentile § High risk: CDAP between 1st and 5th percentile § Elevated risk: CDAP between 5th and 25th percentile § Normal risk: CDAP between 25th and 50th percentile § Low risk: CDAP above 50th percentile The value of this classification system becomes particularly evident during market crises, when traditional metrics systematically misclassify risk. Empirical Validation During Market Crises 2008 financial crisis. During the 2008 financial crisis, CDAP correctly identified extreme risk conditions that traditional metrics failed to capture: § For investment-grade bonds (LQD), the 22-day period starting September 26, 2008: annualized return of −94.60%, with maximum drawdown of −30.53% § Traditional DAP: −3.10 (misleadingly ranked at only the 15.8th percentile) § CDAP: −0.29 (correctly ranked in bottom 0.7% of observations) COVID-19 market crash. The COVID-19 crisis provides even more dramatic evidence of CDAP’s superior risk identification: § For investment-grade bonds (LQD), the 22-day period starting February 21, 2020: CDAP of −0.17 (bottom 0.09%) § For high-yield bonds (HYG), the 22-day period starting on the same date: CDAP of −0.21 (bottom 0.2%) § Traditional DAP ranked these periods at only the 25.8% and 21.3% percentiles, respectively § CDAP began showing bottom 1% readings for 22-day periods beginning as early as February 18, 2020, providing early warning signals before the worst declines These empirical results demonstrate that CDAP correctly ranks extreme risk events at the appropriate percentiles, whereas traditional metrics systematically understate their severity. November 2025 The Journal of Portfolio Management | 157 PRACTICAL IMPLEMENTATION OF CDAP Implementing CDAP within existing risk management systems requires structured protocols that can be readily integrated into current frameworks. We propose the following implementation guidelines: Monitoring Protocols and Triggers § Daily monitoring: all positions § Enhanced monitoring (twice daily): when CDAP falls below 10th percentile § Risk committee notification: when CDAP falls below 5th percentile § Emergency risk assessment: when CDAP falls below 1st percentile Deterioration Assessment The rate of change in CDAP provides crucial additional information: § Calculate five-day rate of change in CDAP percentile § Rapid deterioration (> 15 percentile points in five days): accelerate response protocol by one level § Example: A position with CDAP in the 12th percentile would normally receive standard monitoring, but if it has declined from the 30th percentile within five days, it should trigger enhanced monitoring despite not yet crossing the 10th percentile threshold Cross-asset Confirmation Framework Risk classification should incorporate information from related assets. Response Framework Response protocols should be graduated based on risk severity and confirmation: § Critical risk + Confirmation: consider immediate 25%–50% position reduction § Critical risk without confirmation: enhanced daily monitoring, stress testing, 10%–25% reduction § High risk + Deteriorating trend: position-specific assessment, potential 10%–15% reduction § High risk + Stable/improving trend: enhanced monitoring without immediate action Time Horizon Adjustments Risk thresholds should be calibrated to investment time horizons: § One-to-three-year horizon: Increase percentile thresholds by five percentage points. § > 3 year horizon: Increase percentile thresholds by 10 percentage points. § Example: For a five-year strategic allocation, critical risk would be defined as CDAP below the 11th percentile rather than the 1st percentile. 158 | The False Promise of Drawdown Rules: New Evidence and a Better Framework November 2025 THE CDAP DECISION TREE To provide practitioners with an intuitive implementation framework, in Exhibit 10, we propose a decision tree approach that systematically evaluates drawdown risk while incorporating market context and cross-asset confirmation. Implementation Example: The COVID-19 Market Crash To demonstrate the practical application of this decision tree, we examine how it would have performed during the COVID-19 market crash of February 2020–March 2020, a period that traditional metrics systematically misclassified. (See Exhibit 9 for the confirmation matrix that the decision tree implements.) February 24, 2020 (early crash phase). § CDAP data: SPY CDAP falls to 6th percentile rank § Decision tree application: • CDAP in bottom 5%–20% range: YES • CDAP deteriorating over past five days: YES (dropped from 35th to 6th percentile) § Result: YELLOW ZONE (moderate risk with negative momentum) EXHIBIT 10 CDAP Risk Assessment Decision Tree November 2025 The Journal of Portfolio Management | 159 March 9, 2020 (acceleration phase). § CDAP data: SPY CDAP falls to 2nd percentile rank § Confirmation: International equity markets (EFA, EEM) also showing CDAP below fifth percentile § Volatility: VIX at 54, representing the 92nd percentile of its three-year range § Decision tree application: • CDAP in bottom 5%: YES • Confirmation in related assets: YES § Result: RED ZONE (high probability of continued deterioration) March 16, 2020 (severe crisis phase). § CDAP data: SPY CDAP falls below 0.5th percentile § Confirmation: Credit markets (HYG, LQD) also showing extreme stress with CDAP below 1st percentile § Volatility: VIX reaches 82, highest level since 2008 § Decision tree application: • CDAP in bottom 5%: YES • Confirmation in related assets: YES § Result: RED ZONE (high probability of continued deterioration) During this same period, traditional metrics were sending dangerously contradictory signals: § Traditional DAP ranked the most severe crisis periods as only moderately concerning (25th percentile). § Sharpe ratio showed improvement during periods of worsening crisis. § These misleading signals would have prevented proper risk escalation during a critical market dislocation. Using the CDAP decision tree would have triggered appropriate risk management responses approximately 1–2 weeks earlier than traditional metrics, providing crucial time for risk mitigation through emergency risk committee meetings, comprehensive stress testing, liquidity evaluation, and potential defensive positioning. Dashboard Integration For practical implementation, we recommend risk dashboards display: § Current CDAP-value § Historical percentile ranking § Five-day change in CDAP percentile § Traditional metrics (DAP, Sharpe, Sortino) for comparison § Cross-asset confirmation status IMPLICATIONS FOR PRACTITIONERS These findings have several important implications for portfolio management practice. First, risk management systems should move beyond simple rules to incorporate market context and cross-asset signals. The CDAP framework provides a structured 160 | The False Promise of Drawdown Rules: New Evidence and a Better Framework November 2025 methodology for this transition that can be implemented within existing risk assessment infrastructure. Second, performance evaluation should incorporate more sophisticated metrics than traditional drawdown-adjusted ratios. CDAP offers more consistent risk-adjusted performance measurement across varying market conditions, particularly during stress periods when accurate assessment is most critical. Third, client communication needs to evolve. Although the intuition behind drawdown rules appears straightforward, particularly given common experiences with US large-cap equities, more sophisticated risk management approaches require more nuanced explanations. Developing this communication framework presents challenges but offers substantial benefits compared to maintaining suboptimal approaches simply because they are easier to explain. By implementing these enhancements, practitioners can develop more robust risk management frameworks that maintain the intuitive appeal of drawdown-based approaches while avoiding their documented limitations. A NOTE ON METHODOLOGY Academic papers traditionally include extensive statistical testing, robustness checks, and formal analysis of limitations. We have deliberately chosen a different path. This choice reflects our focus on practical significance: The central insight of this article—that simple drawdown rules often lead to worse outcomes—doesn’t require complex statistical validation to be meaningful for practitioners. Indeed, if the effect required sophisticated statistical tests to become apparent, it would offer questionable practical value. This approach connects to a broader consideration in investment management: Statistical significance does not imply economic significance, and vice versa. Many established practices in portfolio management persist not because they have been rigorously proven optimal, but because they align with intuition or become institutional norms. The drawdown rule exemplifies this pattern—it appears intuitively sound, especially given experiences with US large-cap equities, yet fails to generalize effectively across other markets and time periods. Some readers might consider traditional academic concerns such as survivorship bias or sample selection. These factors, however, have limited relevance to our analysis precisely because we explicitly use retrospective selection—constructing portfolios with complete historical knowledge—to demonstrate that even under these favorable conditions, simple drawdown rules still fail to deliver consistent value. If they perform poorly even with the benefit of historical knowledge, they are unlikely to perform better in forward-looking applications. Similarly, considerations about rebalancing frequency or transaction costs, although important in many contexts, remain secondary to our core argument. The fundamental limitation of drawdown rules isn’t their implementation cost or timing sensitivity—it’s their conceptual framework. They attempt to impose a simple, universal solution on a complex, context-dependent problem. This doesn’t suggest that drawdown management is impossible or unnecessary. Rather, it indicates that effective drawdown management requires more sophisticated approaches that incorporate market context, cross-asset relationships, and dynamic adjustment. Our goal has been to demonstrate, using clear and direct evidence, why practitioners should approach simple drawdown rules with appropriate skepticism, and why more nuanced frameworks such as CDAP offer superior alternatives. November 2025 The Journal of Portfolio Management | 161 CONCLUSION Effective risk management requires nuance that extends beyond simple mechanical rules. Our empirical evidence demonstrates that traditional drawdown cutoff approaches frequently fail to achieve their intended purpose of limiting losses. In practice, such rules often truncate potential recovery periods while allowing multiple smaller drawdowns to accumulate into larger cumulative losses than a single managed drawdown might have produced. The CDAP framework introduced in this article offers a more robust alternative by addressing the fundamental inconsistencies in traditional metrics. By properly accounting for drawdown severity across both positive and negative return environments, CDAP enables more accurate risk assessment—particularly during market crises when traditional metrics systematically misclassify risk. Implementing effective drawdown management requires recognizing when risk reduction is truly warranted versus when apparent drawdowns represent potential opportunities. This distinction demands contextual awareness, cross-asset confirmation, and dynamic adjustment rather than static thresholds. The decision tree approach presented provides a structured methodology for navigating these complex decisions while maintaining implementation feasibility. Portfolio managers and risk officers face increasing pressure to justify their decision frameworks to clients and investment committees. Moving beyond simple rules to more sophisticated approaches requires both analytical rigor and communication skill. The potential benefits, however—more accurate risk assessment, improved performance evaluation, and enhanced decision making during market stress—offer compelling justification for this evolution. The art of risk management ultimately lies not in avoiding difficult decisions through mechanical rules, but in developing frameworks that enhance judgment with systematic analysis. By integrating CDAP into existing risk management systems, practitioners can better distinguish between transitory drawdowns and genuine risk events, leading to more resilient portfolios and more consistent risk-adjusted performance. REFERENCES Ang, A. 2014. Asset Management: A Systematic Approach to Factor Investing. New York: Oxford University Press. Carr, P., J. Zhang, and O. Hadjiliadis. 2011. “Maximum Drawdown Insurance.” International Journal of Theoretical and Applied Finance 14 (8): 1195–1230. Chekhlov, A., S. Uryasev, and M. Zabarankin. 2005. “Drawdown Measure in Portfolio Optimization.” International Journal of Theoretical and Applied Finance 8 (1): 13–58. Grossman, S. J., and Z. Zhou. 1993. “Optimal Investment Strategies for Controlling Drawdowns.” Mathematical Finance 3 (3): 241–276. Kahneman, D. 2011. Thinking, Fast and Slow, 1st ed. New York: Farrar, Straus, and Giroux. Kaminski, K. M., and A. W. Lo. 2014. “When Do Stop-Loss Rules Stop Losses?” The Journal of Financial Markets 18: 234–254. Thaler, R. H. 2015. Misbehaving: The Making of Behavioral Economics, 1st ed. New York: W. W. Norton & Company. Disclaimer The material presented is for informational purposes only. The views expressed in this material are the views of the author and are subject to change based on market and other conditions and factors; moreover, they do not necessarily represent the official views of VS Asset Management and its affiliates.