Strategy-Level vs. Portfolio-Level Drawdowns: When Diversification Works

A single strategy can experience a 25% drawdown while your portfolio experiences only 12%. The difference is diversification—other strategies or positions that are uncorrelated or negatively correlated with the drawdown. A strategy-level drawdown is the loss from one strategy, viewed in isolation. A portfolio-level drawdown is the loss across all capital, accounting for how different pieces move together. Understanding the distinction guides capital allocation and helps you prepare for the drawdown you will actually experience (portfolio) rather than the theoretical worst case (single strategy).

Lede

Strategy-level and portfolio-level drawdowns are distinct metrics, often wildly different in magnitude. A trend-following strategy might experience a 30% drawdown (historical norm) while a diversified portfolio of trend-following plus mean-reversion plus covered calls experiences only an 18% portfolio drawdown, because the strategies move differently in different market conditions. Strategy drawdown comparison is crucial for multi-strategy traders; it determines how much capital to allocate to each strategy and whether adding a new strategy improves or worsens overall portfolio risk. A portfolio with multiple uncorrelated strategies typically experiences smaller drawdowns, shorter recovery times, and smoother returns than a portfolio concentrated in a single strategy.

Quick definition: Strategy-level drawdown is the peak-to-trough decline of an individual strategy in isolation; portfolio-level drawdown is the peak-to-trough decline of the entire portfolio, accounting for diversification across multiple strategies or assets.

Key Takeaways

• A portfolio-level drawdown is typically 30–60% smaller than the worst individual strategy drawdown if strategies are uncorrelated (e.g., trend-following and mean-reversion)
• Correlation during drawdowns is the critical variable; strategies that are uncorrelated in normal times often become correlated during market shocks, reducing diversification benefit
• A portfolio with one 25% strategy and one 15% strategy may experience a portfolio drawdown of only 18% if capital is allocated 50/50 and strategies are uncorrelated
• Concentration in a single strategy or asset class eliminates diversification; portfolio drawdowns become identical to strategy drawdowns
• The portfolio-level drawdown is the risk you actually live with; strategy-level drawdowns are important for understanding individual strategy risk but not for planning your experience

The Mathematics of Diversification: How Two Strategies Beat One

Suppose you split $100,000 equally between two strategies: Strategy A (trend-following, 25% historical max drawdown) and Strategy B (mean-reversion, 20% historical max drawdown).

Scenario 1: Strategies are perfectly correlated (move together) Both strategies drawdown simultaneously at their worst. The portfolio experiences a 22.5% drawdown (simple average). Diversification provides no benefit.

Scenario 2: Strategies are uncorrelated (move independently) Strategy A experiences a 25% drawdown while Strategy B experiences 0% drawdown (or gains). The portfolio drawdown is:

• Strategy A loses: $50,000 × 25% = $12,500
• Strategy B gains or flats: $50,000 × 0% = $0
• Portfolio loss: $12,500 / $100,000 = 12.5% drawdown

The portfolio drawdown is half the worst-case strategy drawdown.

Scenario 3: Strategies are negatively correlated (move oppositely) When Strategy A is in a 25% drawdown, Strategy B gains. If Strategy B gains 10% while A loses 25%:

• Strategy A loses: $50,000 × 25% = $12,500
• Strategy B gains: $50,000 × 10% = $5,000
• Portfolio loss: $12,500 - $5,000 = $7,500
• Portfolio drawdown: $7,500 / $100,000 = 7.5%

The portfolio drawdown is one-third the worst-case strategy drawdown. The counter-movement of Strategy B provides a hedge.

This is the mathematical case for diversification. The key variable is correlation—how closely the strategies move together.

Understanding Correlation: The Hidden Risk in Diversification

Correlation is the degree to which two strategies move together. Correlation ranges from -1 (perfect negative correlation; they move in exact opposite directions) to +1 (perfect positive correlation; they move identically).

Zero correlation: Strategies move independently. When one is up, the other is down, flat, or up differently. Diversification is maximized.

Positive correlation: Strategies move somewhat together. When one is down, the other is likely down too, but not as much. Diversification benefit is reduced.

High positive correlation: Strategies move almost together. Diversification benefit is minimal.

The problem is that correlation is not constant. In normal markets, a trend-following strategy and a mean-reversion strategy might be uncorrelated (0.1 correlation). But during market shocks (COVID crash, 2022 rate hikes), most strategies drawdown together. Correlation during crisis rises to 0.6–0.8. This is called "correlation break-down" and it is the reason diversification fails when you need it most.

A trader might build a portfolio expecting correlation of 0.2 during normal times, yielding portfolio drawdown of 15% with 50/50 allocation to two 25% max-drawdown strategies. But during a market shock, correlation rises to 0.7, and the portfolio drawdown reaches 22%—much higher than expected.

Historical backtests typically show more diversification benefit than live trading because backtests miss regime shifts and correlation breakdown.

Correlation During Calm Markets vs. Drawdowns

The correlation of two strategies is not constant over time. It varies dramatically.

Calm market periods (2013–2018, 2019, 2021–2022 mid-period): Trend-following and mean-reversion strategies are often uncorrelated. One thrives in trends; the other thrives in ranges. Correlation is 0.1–0.3. Diversification is high; portfolio drawdown is well-controlled.

Shock periods (2020 COVID, 2022 rate shock, 2023 banking crisis): Most strategies drawdown simultaneously. The source of loss (systemic risk: liquidity drying up, volatility spiking, leverage forced liquidations) affects all strategies. Trend-following and mean-reversion both lose. Correlation rises to 0.6–0.8. Diversification fails; portfolio drawdown approaches the worst-strategy drawdown.

Crisis-driven regime shifts: A persistent regime shift (not a short shock) can degrade strategy performance across the board. Example: the 2022 rate-hike regime shifted correlations and asset class trends, making trend-following simultaneously less profitable and mean-reversion simultaneously less profitable. Portfolio correlation across strategies rose and stayed high.

The critical insight: The diversification benefit you see in calm backtests is not the diversification benefit you will experience in a real drawdown. Real drawdowns are driven by shocks, and shocks cause correlation breakdown. This is why diversification is less powerful in practice than in theory.

Single-Strategy vs. Multi-Strategy Portfolios: Real Examples

Single-Strategy Portfolio: The Concentrated Case A trader runs only a trend-following strategy on S&P 500 futures with $100,000 capital. Historical maximum drawdown: 28%. In a drawdown, the trader experiences the full 28% loss ($28,000). Recovery takes 12–16 weeks. Psychological toll is severe because all capital is at risk simultaneously.

Two-Strategy Portfolio: Partial Diversification The same trader splits capital: $50,000 in trend-following (S&P), $50,000 in mean-reversion (Treasury yield curve). Backtests show:

• Trend-following max drawdown: 28%
• Mean-reversion max drawdown: 18%
• Expected portfolio drawdown: 16% (assuming 0.2 correlation in calm markets)

In reality, during a 2022-like shock:

• Correlation rises to 0.6
• Actual portfolio drawdown: 21% (higher than expected)
• Loss: $21,000

The trader expected $16,000 loss but experienced $21,000 because correlation rose during the shock.

Three-Strategy Portfolio: More Diversification The same trader with $100,000 allocates:

• $40,000 trend-following (S&P, 28% max drawdown)
• $40,000 mean-reversion (Treasuries, 18% max drawdown)
• $20,000 covered calls (equities, 12% max drawdown)

Calm market correlation across all three strategies: 0.25 average. Expected portfolio drawdown: 14%.

Shock period correlation: 0.65 average. Actual portfolio drawdown during shock: 19%.

Even with the correlation breakdown, the three-strategy portfolio experiences less drawdown than the single-strategy case (19% vs. 28%) or the two-strategy case (19% vs. 21%).

Adding the third strategy provides a hedge because covered calls typically rise when equities fall (implied volatility increases), creating some negative correlation during shocks.

Capital Allocation Across Strategies

How much capital should you allocate to each strategy? The answer depends on correlation and individual strategy drawdown.

Equal-weight allocation (50/50, 33/33/33): Simple, no optimization needed. Works well if strategies have similar historical drawdown and similar expected correlation. Easy to manage. Downside: may not minimize portfolio drawdown.

Risk-parity allocation (allocate capital inversely to historical volatility): If Strategy A has 25% drawdown and Strategy B has 15% drawdown, allocate 37.5% to A and 62.5% to B. This weights strategies so they contribute equally to overall portfolio risk. More complex but theoretically optimizes drawdown reduction.

Correlation-aware allocation (optimize using expected correlation): If you estimate correlation at 0.3, optimize capital allocation to minimize expected portfolio drawdown. Requires forecasting correlation, which is unreliable. Backtested allocations often fail in live trading because correlation was misestimated.

Hierarchical allocation (add strategies based on marginal benefit): Start with one strategy. Add a second strategy only if it reduces portfolio drawdown below the first strategy's drawdown. Add a third only if it further reduces portfolio drawdown. This is pragmatic and data-driven.

Most retail traders use equal-weight or hierarchical allocation because they are simple and do not require correlation forecasting.

The Correlation Breakdown Risk: What Kills Diversification

Diversification breaks down during the periods you need it most. This is the diversification paradox.

Reasons for correlation breakdown:

Systemic risk events: A banking crisis, credit freeze, or liquidity crisis affects all strategies simultaneously. Everything drops at once.

Margin pressure: If one strategy experiences a drawdown, your broker may increase margin requirements. You are forced to liquidate the other strategies (or reduce positions) to meet margin calls. Forced simultaneous reduction makes drawdowns worse.

Forced rebalancing: If one strategy plummets, you may be forced to rebalance (to maintain target allocation) by selling other strategies (to raise capital for the falling strategy). This forces correlation that would not otherwise exist.

Common external factor: If all strategies trade the same underlying asset (all trade SPY, for example), they all suffer during SPY drawdowns. Apparent diversification is illusory.

Leverage and leverage withdrawal: If strategies use leverage, a drawdown in one strategy may trigger leverage reduction globally (risk-off regime). All strategies reduce positions, creating correlation.

The solution is to stress-test correlation. Backtest your portfolio through periods of known correlation breakdown (2008 financial crisis, 2020 COVID crash, 2022 rate shock). What was the actual portfolio drawdown? Did correlation rise? Did your diversification benefit persist? If not, reconsider the capital allocation.

Strategy Drawdown vs. Account Drawdown: The Role of Capital Allocation

A strategy's drawdown and your account's drawdown are different if you do not allocate 100% of capital to that strategy.

If you allocate 40% of capital to a strategy with 25% maximum drawdown, the strategy's loss is 25%, but the account's loss from that strategy is 10% (25% × 40%). If other capital is in cash or other strategies, the account drawdown is less than the strategy drawdown.

Example:

• $100,000 account
• $40,000 in trend-following strategy (25% max drawdown)
• $30,000 in mean-reversion strategy (20% max drawdown)
• $30,000 in cash/bonds (0% drawdown)

If both strategies hit their max drawdown simultaneously:

• Trend-following loss: $40,000 × 25% = $10,000
• Mean-reversion loss: $30,000 × 20% = $6,000
• Total loss: $16,000
• Account drawdown: 16%

Despite strategy drawdowns of 25% and 20%, the account drawdown is only 16%, because capital is distributed.

This is why portfolio construction matters. Even with a single strategy, allocating only 60–70% to the strategy and keeping 30–40% in cash or bonds reduces portfolio drawdown below the strategy's drawdown.

Real-World Comparison: Two Traders' Drawdown Experiences

Trader A (Single Strategy) Runs a trend-following strategy exclusively. $100,000 capital, 100% deployed. Historical maximum drawdown: 26%.

In Q2 2024, strategy experiences a 24% drawdown. Trader experiences:

• Account drawdown: 24%
• Dollar loss: $24,000
• Recovery time: 14 weeks
• Psychological toll: Severe (all capital at risk)

Trader B (Diversified Portfolio) Runs three strategies with uncorrelated expected performance. $100,000 capital:

• $35,000 in trend-following (26% max historical)
• $35,000 in mean-reversion (18% max historical)
• $30,000 in covered calls (12% max historical)

In Q2 2024, correlation during the shock is 0.6 (higher than calm-market average of 0.25):

• Trend-following experiences: 24% drawdown
• Mean-reversion experiences: 16% drawdown
• Covered calls experience: 10% drawdown
• Portfolio result:
  • Loss from trend: $35,000 × 24% = $8,400
  • Loss from mean-reversion: $35,000 × 16% = $5,600
  • Loss from covered calls: $30,000 × 10% = $3,000
  • Total loss: $17,000
  • Portfolio drawdown: 17%

Trader B experiences:

• Account drawdown: 17%
• Dollar loss: $17,000 (vs. Trader A's $24,000)
• Recovery time: 10 weeks (vs. Trader A's 14 weeks)
• Psychological toll: Moderate (diversification reduced impact)

The same market shock affected both traders' underlying strategies similarly, but Trader B's diversification reduced the portfolio impact by 7 percentage points and 4 weeks of recovery time.

When Diversification Fails: Market Regimes Where All Strategies Lose

There are market conditions where diversification fails entirely because all strategies lose simultaneously.

Flash crash (March 2020): Most trading strategies lost on the same day due to forced liquidations and liquidity drying up. A portfolio of trend, mean-reversion, and options strategies all lost 5–15% in a single day. Correlation was nearly 1.0.

The 2022 regime shift: Transition from low-inflation, low-rate environment to high-inflation, rapidly rising-rate environment. Trend-following on equities and bonds both lost. Mean-reversion on both equities and bonds struggled. A diversified portfolio still lost because the diversifying factors (different strategies) faced the same headwind (regime change).

The 2008 financial crisis: Credit markets seized; leverage was withdrawn globally. All leveraged strategies lost simultaneously. Diversification across strategies provided minimal help because the shock affected the entire financial system.

These are the drawdowns diversification cannot prevent. These are tail-risk events where all correlations rise to 1.0 and the portfolio drawdown approaches the worst-case strategy drawdown (or worse, if leverage is involved).

The defense against these rare events is position sizing so small that even a worst-case portfolio drawdown is survivable.

FAQ

How do I calculate the expected portfolio drawdown from individual strategy drawdowns?

You need three inputs: (1) historical maximum drawdown for each strategy, (2) expected correlation between strategies, (3) capital allocation. The formula is approximate:

Portfolio max DD ≈ sqrt(sum of (allocation% × strategy DD)^2 + 2 × allocation_A × allocation_B × correlation × DD_A × DD_B)

In practice, this is complex. Use historical simulation instead: run a backtest of the combined portfolio and observe the actual maximum drawdown across the period.

How do I measure correlation between my strategies?

Calculate returns (daily, weekly, or monthly) for each strategy separately. Then calculate the correlation coefficient between the return series. Use a rolling correlation window (last 1–2 years) to see if correlation changes over time. Be aware that backtested correlation is often lower than real correlation because backtests miss regime shifts.

Should I use correlation or volatility to decide capital allocation across strategies?

Both matter. Volatility determines the dollar amount risked per unit of capital. Correlation determines whether drawdowns overlap. A low-volatility, high-correlation strategy is less risky but provides poor diversification. A high-volatility, low-correlation strategy is risky per trade but provides strong diversification. For most traders, prioritize correlation: allocate more to uncorrelated strategies, less to correlated strategies.

What is the minimum correlation needed for diversification to work?

Correlation below 0.5 provides noticeable diversification benefit. Correlation below 0.2 provides strong benefit. Above 0.7, diversification benefit is minimal. However, remember that calm-market correlation is not shock-market correlation. A strategy pair with 0.1 correlation in normal markets may have 0.6–0.7 correlation during shocks, reducing the real-world benefit.

Is a portfolio of 10 strategies better than a portfolio of 3 strategies?

Not necessarily. After 3–4 uncorrelated strategies, adding more strategies provides diminishing benefit. Additional strategies add complexity and potential correlation that would not otherwise exist (you run out of uncorrelated markets to trade). For most traders, 2–4 well-chosen strategies provide the optimal risk-adjusted returns.

What is the worst portfolio drawdown I should prepare for?

Prepare for a portfolio drawdown equal to the worst individual strategy drawdown (assume correlation goes to 1.0 during shock). If your strategies have historical max drawdowns of 28%, 20%, and 15%, mentally prepare for a 28% portfolio drawdown. The diversification benefit is a bonus; the worst case is no benefit. This prevents shock when correlation breaks down.

Related Concepts

• Reading Underwater Equity Curves
• Maintaining Discipline Through a Drawdown
• Setting Drawdown Circuit Breakers
• What Ruin Means
• Defining Investment Risk

Summary

Strategy-level drawdowns (the loss experienced by an individual strategy in isolation) are typically larger than portfolio-level drawdowns (the loss experienced across all strategies and capital). Diversification reduces portfolio drawdown if strategies are uncorrelated; a portfolio of strategies with 25% and 20% individual max drawdowns can experience a portfolio drawdown of only 15% if capital is equally allocated and correlation is low. However, correlation is not constant; it rises dramatically during market shocks (0.2 in calm markets, 0.6–0.8 during crises), reducing diversification benefit when you need it most. Capital allocation matters: allocating 40% to a strategy with 25% drawdown limits that strategy's impact to 10% of portfolio. The portfolio-level drawdown is the risk you actually experience; this is the drawdown to prepare for psychologically, to size positions around, and to plan circuit breakers against. Diversification is powerful but not perfect; it reduces drawdown by 30–50% in normal times and 10–20% in crisis periods. The goal is not to eliminate drawdown but to reduce it to levels that maintain discipline and capital preservation.

Next

→ The Worst Historical Drawdowns in Markets