Why Correlations Rise During Market Crashes—and Diversification Can Fail

Diversification is supposed to reduce risk.

But during severe market selloffs, something uncomfortable can happen:

assets that normally move differently can suddenly start falling together.

This is known as correlation convergence.

It helps explain why a portfolio that looks diversified in normal markets can experience much larger losses during a crisis.

Educational research only. This article is not investment advice.

What Is Correlation?

Correlation measures how two assets tend to move relative to each other.

It ranges from -1 to +1.

+1 correlation
The assets tend to move in the same direction.

0 correlation
Their movements have little consistent relationship.

-1 correlation
They tend to move in opposite directions.

Diversification works best when a portfolio contains assets that do not all respond to the same forces in the same way.

Why Correlations Can Rise During Crashes

During normal conditions, investors distinguish between:

  • sectors;
  • countries;
  • industries;
  • growth prospects;
  • individual companies.

During a crisis, one factor can suddenly dominate everything:

the need to reduce risk.

Investors may sell many assets at once.

That can cause previously different investments to start moving together.

Forced Selling Makes It Worse

Leverage can amplify this effect.

If prices fall sharply, leveraged investors may face margin calls.

They may then sell whatever assets they can—not necessarily the assets they want to sell.

The chain can become:

Prices fall

Losses increase

Margin calls appear

Investors sell more assets

Correlations rise

This can spread stress across markets.

Liquidity Can Disappear

Diversification also depends on markets functioning normally.

During severe stress, buyers can become harder to find.

Investors seeking cash may sell their most liquid holdings first.

As selling spreads, different assets can decline together.

This means the problem is not always that the investments suddenly became economically identical.

Sometimes they are simply reacting to the same liquidity shock.

Why Diversification Can Disappoint

Imagine a portfolio containing:

  • technology stocks;
  • banks;
  • small caps;
  • international equities;
  • crypto.

On paper, that may look diversified.

But all of those positions still carry significant exposure to risk appetite.

During a major market shock, they may all fall simultaneously.

The portfolio contained many assets.

But it may not have contained many different sources of risk.

That distinction matters.

Asset Diversification vs Risk Diversification

Owning more investments does not automatically create better diversification.

For example:

10 technology stocks
may still represent one broad technology risk.

Stocks across several countries
may still share global equity risk.

Stocks plus crypto
may still be heavily exposed to liquidity and investor risk appetite.

True diversification asks:

What makes these assets behave differently?

Not simply:

How many assets do I own?

Why Historical Correlation Can Mislead

Correlation is not fixed.

Two assets may show low correlation during normal periods but become strongly correlated during stress.

That means historical averages can hide important behavior.

Suppose two assets usually have low correlation.

During a crisis, both suddenly fall 20%.

Their long-run correlation may still look reasonable.

But the period when diversification mattered most was exactly when the relationship changed.

This is why risk analysis should consider stress conditions, not only averages.

How Risk Simulation Helps

TradingSimuLab’s Risk Simulation looks beyond one expected outcome.

Important outputs include:

VaR

Where does severe downside begin?

CVaR

How large are losses once the portfolio moves beyond that severe-loss threshold?

Max Drawdown

How far could the investment fall from peak to trough?

Probability of Gain

How often do simulated paths finish above their starting level?

Terminal Price Range

How wide is the range of possible ending outcomes?

These measures help reveal something that simple diversification claims can miss:

the path matters.

A portfolio may have an attractive average outcome while still carrying significant downside during stressful scenarios.

Does Diversification Stop Working?

Not necessarily.

Diversification can still reduce risk.

But its protection can become weaker precisely when markets are under the most pressure.

The solution is not to abandon diversification.

It is to understand that:

correlations change,

liquidity changes,

and:

risk regimes change.

Different assets should therefore be evaluated by how they behave under stress—not only during calm markets.

A Simple Diversification Checklist

Ask:

Do my assets depend on the same economic conditions?

Would they all suffer if interest rates rise?

Would they all fall if liquidity disappears?

How did their relationships change during previous selloffs?

What happens to portfolio drawdown if correlations increase?

Those questions reveal more than simply counting how many investments are in a portfolio.

Final Takeaway

Diversification works because different assets behave differently.

But during market crashes, fear, leverage and liquidity pressure can push correlations higher.

The risk chain is:

Market shock → forced selling → liquidity pressure → rising correlations → weaker diversification

That does not make diversification useless.

It means diversification should be tested under stress, not judged only during normal markets.

The better question is not:

“How many assets do I own?”

It is:

“How differently will those assets behave when markets are under pressure?”

For more market risk tools, Monte Carlo simulations and downside analysis, sign up to TradingSimuLab and explore the platform.

Continue exploring TradingSimuLab.

  • Terminal Price Range Explained: How to Read Simulation Outcome Bands

    A terminal price range shows where simulated price paths finish at the end of a selected time horizon. Instead of giving one price forecast, it presents a range of possible outcomes. That matters because one Expected Price can look more precise than the underlying simulation really is. The terminal range helps answer: How wide is…

  • Tail Risk, VaR and CVaR Explained Inside Risk Simulation

    Tail risk is the risk of unusually severe losses in the adverse end of an investment-return distribution. Inside TradingSimuLab’s Risk Simulation, two metrics help describe that downside: VaR estimates where severe modeled downside begins. CVaR estimates how severe losses become, on average, once outcomes move beyond that VaR threshold. The distinction matters because an investment…

  • Slope Health and Distance Health Explained in Trend Detector

    TradingSimuLab’s Slope Health and Distance Health turn raw trend structure into easier-to-read labels. They answer two different questions: Slope Health: Is the underlying trend base rising, falling, flat, or becoming unusually steep? Distance Health: Is price sitting at a reasonable distance from that trend base, or has it become stretched? Together, they help users distinguish…

  • Risk Simulation Explained: VaR, CVaR, Drawdown and MonteCarlo Paths

    TradingSimuLab’s Risk Simulation uses Monte Carlo paths to examine possible future outcomes and, especially, the downside hidden behind an attractive expected return. The most useful risk metrics answer different questions: VaR: Where does severe modeled downside begin? CVaR: How bad are losses deeper in that adverse tail? Maximum Drawdown: How difficult can the path become…

  • Risk Simulation Workflow: Combine Risk, Trend, Persistence and Timing

    A strong trend is not automatically a good risk setup. TradingSimuLab’s Risk Simulation workflow combines direction, durability, timing and downside analysis so one attractive signal does not become the entire research conclusion. The practical sequence is: Trend Detector → Trend Persistence → Timing Model → Risk Simulation This answers four different questions: Is the trend…

  • Risk Simulation Explained: How to Read Monte Carlo Paths,VaR, CVaR and Drawdown Risk

    TradingSimuLab’s Risk Simulation is the downside-path layer of the five-model framework. It uses simulated future price paths to help answer: Is the potential reward attractive enough relative to the modeled downside? Instead of focusing only on upside, Risk Simulation examines: The goal is not to predict one exact future price. It is to understand how…

  • Reversal Warning and Extension Watch: How to Read Trend Maturity Without Overreacting

    A Reversal Warning and Extension Watch are caution layers inside TradingSimuLab’s Trend Persistence model. They help answer two related questions: Reversal Warning: Is the trend showing possible signs of cooling or losing durability? Extension Watch: Has the move become mature or stretched enough to deserve closer attention? Neither means the trend must reverse. A strong…

  • Range and Chop Risk Explained: When Timing Conditions AreNoisy

    Range and Chop Risk describes market conditions where price action is sideways, repetitive, or too noisy to produce a clean directional timing signal. Inside TradingSimuLab’s Timing Model, it acts as the noise layer. A high Range/Chop Risk reading does not mean a large move cannot happen. It means: the immediate market structure is less clean,…

  • Probability of Gain Explained: How to Read Simulation Win-Rate Context

    Probability of Gain measures the percentage of simulated paths that finish above their starting value. If 570 out of 1,000 simulated paths end higher than where they began, the simulation would show a Probability of Gain of approximately: 57% That makes the metric easy to understand—but also easy to misuse. A 57% Probability of Gain…