What Is Correlation?

What Is Correlation? Definition and Key Terms

Correlation is a statistical measure that describes the degree and direction of a relationship between two variables. In trading and investing contexts, correlation typically refers to how the price movements of one asset relate to those of another. A correlation coefficient ranges from –1 to +1:

• +1 (Perfect Positive Correlation): When one asset’s price rises or falls, the other asset always moves in the same direction by a proportional amount.
• –1 (Perfect Negative Correlation): When one asset’s price moves in one direction, the other moves in the exact opposite direction by a proportional amount.
• 0 (Zero Correlation): There is no consistent linear relationship between the price movements of the two assets.

Key Correlation Measures

Pearson’s r (Pearson Correlation Coefficient):

1. Definition: Measures the strength and direction of a linear relationship between two continuous variables.
2. Range: –1 to +1. Values near +1 or –1 indicate a strong linear relationship, while values near 0 suggest a weak or no linear relationship.
3. Interpretation:
   • 0.7 ≤ r ≤ 1.0: Strong positive correlation
   • 0.3 ≤ r < 0.7: Moderate positive correlation
   • –0.3 < r < 0.3: Weak or negligible correlation
   • –0.7 < r ≤ –0.3: Moderate negative correlation
   • –1.0 ≤ r ≤ –0.7: Strong negative correlation

Spearman’s ρ (Spearman Rank Correlation):

1. Definition: A non-parametric measure that assesses how well the relationship between two variables can be described by a monotonic function. Instead of raw values, it uses the ranks of data points.
2. Range: –1 to +1, with similar interpretation to Pearson’s r but more robust to outliers and non-linear relationships.

Other Less Common Measures (for Reference):

1. Kendall’s τ: Also a rank-based measure that evaluates the ordinal association between two variables.
2. Point-Biserial Correlation: Used when one variable is dichotomous (e.g., up/down classification) and the other is continuous.

Glossary of Key Terms

• Positive Correlation: Two variables move in the same direction—e.g., if Asset A’s return increases, Asset B’s return tends to increase.
• Negative Correlation: Two variables move in opposite directions—e.g., if Asset A’s return increases, Asset B’s return tends to decrease.
• Zero (or Near-Zero) Correlation: No discernible linear relationship; price movements appear independent.
• Spurious Correlation: A coincidental relationship where two variables appear correlated but are actually linked by a third, unseen factor or purely by chance.
• Rolling Correlation: A method of calculating correlation over a moving time window (e.g., 30-day rolling Pearson r) to observe changes over time.
• Correlation Matrix: A table showing pairwise correlation coefficients among a group of assets, often used to gauge overall portfolio diversification.

Why Correlation Matters to Traders

Understanding correlation helps traders and portfolio managers to:

• Diversify Portfolios: By combining assets with low or negative correlation, overall portfolio risk can be reduced without sacrificing expected returns.
• Hedge Positions: Traders can offset risk—for example, if two currency pairs move negatively correlated, a long position in one may hedge a short position in the other.
• Identify Market Regimes: Correlations between asset classes often spike during market stress (e.g., equities and credit markets during the 2008 crisis), signalling heightened systemic risk.

Limitations and Common Pitfalls

While correlation is a powerful tool for understanding relationships between assets, there are several important caveats to bear in mind:

1. Correlation Does Not Imply Causation
   • Just because two assets move together does not mean one causes the other to move. For example, during a commodity boom, both oil and certain equities might rise, but this could be driven by broader economic factors rather than oil price changes directly causing equity gains.
   • Spurious Correlations: Occasionally, two seemingly unrelated variables exhibit high correlation purely by coincidence or because of a hidden third factor. For instance, ice cream sales and stock market performance might both rise in summer, but that does not mean buying ice cream improves portfolio returns.
2. Correlations Change Over Time (Regime Shifts)
   • Markets evolve, and relationships that held during one period may break down in another. For example, equities and government bonds traditionally exhibit low or negative correlation in normal conditions but can move in tandem during extreme market stress.
   • Rolling Correlation Variability: A static, historical correlation (e.g., based on the past year) may not capture sudden shifts. Traders often use rolling windows—say, 30- or 90-day Pearson r—to monitor how correlations drift, ensuring they adjust positions if asset relationships decouple.
3. Sample Size and Data Window Selection
   • A small sample size (for example, daily returns over just one month) can produce unstable correlation estimates. Short windows often exaggerate noise and may give misleading signals.
   • Conversely, using an excessively long window (several years) can mask recent changes. It is important to strike a balance: choose a window that matches your holding period or trading horizon.
   • Look-Back Bias: Be cautious not to optimise based on past data that may no longer be representative of future behaviour.
4. Assumption of Linearity
   • Pearson’s r measures linear relationships. If two assets exhibit a non-linear relationship—say, one moves with the square or logarithm of the other—Pearson’s r may understate the true dependence.
   • In such cases, rank-based measures like Spearman’s ρ or more advanced techniques (e.g., copulas) may capture non-linear dependencies more accurately.
5. Over-Reliance on Correlation Matrices
   • A correlation matrix provides pairwise relationships but does not account for multivariate interactions among three or more assets. In complex portfolios, factor models (e.g., principal component analysis) can reveal deeper insights into how groups of assets move together.
   • Diversification Fallacy: Even if two assets show low correlation historically, an extreme market event could synchronise them. For instance, during the 2008 Financial Crisis, many supposedly uncorrelated assets surged to high positive correlation, reducing diversification benefits.

If you’d like to learn how to monitor evolving correlations in real time, try our demo account and experiment with rolling-correlation indicators on the AvaTrade MT4/MT5 platforms.

Practical Applications in Portfolio Construction

Correlation analysis is a cornerstone of effective portfolio construction. By understanding how different assets move in relation to one another, traders and investors can make more informed decisions regarding diversification, hedging, and risk management.

Below are key ways to apply correlation in constructing and maintaining a well-balanced portfolio.

Using a Correlation Matrix for Diversification

A correlation matrix displays pairwise correlation coefficients among a set of assets (for example, equities, bonds, gold and foreign-exchange pairs).

By examining this matrix, you can identify which assets are highly correlated (near +1), uncorrelated (around 0), or negatively correlated (near –1).

Example Correlation Matrix (hypothetical 6-month rolling Pearson r):

In this illustration, EUR/USD and USD/JPY exhibit a strong negative correlation (–0.78).

If a trader is long EUR/USD and concerned about a yen-driven appreciation in USD/JPY, they might use USD/JPY as a partial hedge.

Conversely, equities and bonds show a slight negative correlation (–0.23), which could help smooth overall portfolio volatility.

For a deeper dive into portfolio diversification theory, see our Diversification Strategies guide.

Balancing Risk Through Asset Selection

After constructing a correlation matrix, you can identify asset combinations that reduce overall portfolio volatility without substantially diluting expected returns.

For instance, if gold exhibits near-zero correlation with equities, allocating a portion of capital to gold can act as a buffer during equity drawdowns.

Sector and Geographical Diversification: Correlation analysis isn’t limited to asset classes; you can apply it to sectors (e.g., technology versus utilities stocks) or regions (e.g., emerging-market equities versus developed-market bonds).

By choosing sectors or regions with low correlations, you mitigate concentration risk.

Scenario: Balancing a Sample Portfolio

Imagine you have a portfolio with 60 per cent equities, 30 per cent bonds and 10 per cent gold.

Over the past quarter, the correlation matrix shows that equities and bonds have moved increasingly in tandem (average Pearson r of +0.5).

To reduce potential drawdown risk, you might:

1. Trim Equity Exposure: Lower equity allocation to 50 per cent.
2. Increase Gold Allocation: Shift to 20 per cent gold, since gold’s correlation with equities remains near 0.15.
3. Introduce a Low-Correlated Forex Pair: Add a small position in USD/CHF, which historically has shown a negative correlation with equities in risk-off scenarios.

By making these adjustments, the portfolio’s overall volatility can decrease. Periodically re-running the correlation matrix (for example, monthly) ensures you capture shifting relationships as market regimes evolve.

Real-World Case Studies and Historical Scenarios

Examining past market events and asset relationships can illustrate how correlation behaves during different conditions.

Below are two illustrative case studies—one covering a broad market stress scenario and another focusing on currency pairs—that demonstrate the practical implications of correlation dynamics.

Case Study: The 2008 Financial Crisis

During the 2008 Financial Crisis, correlations among a wide range of asset classes converged toward +1, fundamentally altering the benefits of diversification. Key observations include:

• Equities and Credit Instruments:
  • Prior to 2007, major equity indices (for instance, the S&P 500) and investment-grade credit (e.g., corporate bond indices) often exhibited moderately low correlations (Pearson r roughly between 0.2 and 0.4).
  • As Lehman Brothers collapsed in September 2008, fears of systemic failure drove a flight to safety. Both equities and credit spreads widened dramatically, causing their returns to move almost in lockstep—correlation coefficients spiked above +0.8 for several months.
  • Implication: Portfolios that held both equities and credit, believing they were uncorrelated, saw simultaneous drawdowns, emphasising that “safe” diversification can erode under extreme stress.
  • Data Source: For a detailed analysis of cross-asset correlations during the crisis, see the Bank for International Settlements’ Working Paper No. 284 (“Co‐movement in commodity and equity markets,” BIS, 2008).
• Equities and Commodities (Oil and Gold):
  • Historically, oil and equities sometimes exhibited a low-to-moderate positive correlation (r between 0.3 and 0.6).
  • During late 2008, as global demand collapsed, oil prices plunged. Equities similarly fell, but by different magnitudes. The correlation between crude oil futures and equity indices briefly reached near +0.7.
  • Gold—commonly perceived as a safe haven—initially rose in mid-2008, showing a negative correlation (around –0.4) with equities. However, as liquidity concerns intensified, gold was sold alongside other assets, driving its correlation with equities toward +0.5 by November 2008.
  • Implication: Even traditionally negatively correlated “safe havens” can become positively correlated when forced liquidations and margin calls prevail.
  • Reference: See the International Monetary Fund’s Global Financial Stability Report (IMF, October 2008) for empirical data on asset co-movements during stress.
• Lessons Learned:
  1. Diversification Gaps: During market panics, many asset classes converge, reducing diversification effectiveness.
  2. Dynamic Risk Management: Constantly monitor rolling correlations rather than relying on historical averages.
  3. Stress Testing: Incorporate worst-case scenarios into portfolio stress tests—simulate correlation spikes to +1 to see potential maximum drawdowns.

Case Study: Currency Pair Relationships (EUR/USD vs USD/CHF)

Certain major currency pairs have historically exhibited strong negative correlation, largely due to fundamental linkages between European and Swiss monetary policy, as well as safe-haven flows. Key points:

• Long-Term Negative Correlation:
  • Over the period 2000–2019, daily returns of EUR/USD and USD/CHF often produced Pearson r values around –0.85 to –0.95. When the euro strengthened against the dollar, the Swiss franc tended to weaken by a similar magnitude, and vice versa.
  • Fundamental Drivers: Switzerland’s economy and the Swiss National Bank (SNB) traditionally maintained lower interest rates than the Eurozone, leading to capital flows that produced this mirror-like relationship.
• Use in Hedging and Trading Strategies:
  • Hedging Example: A trader holding a long position in EUR/USD wishing to hedge euro exposure could short USD/CHF. Because of the strong negative correlation, profit on one position generally offset losses on the other when exchange rates shifted.
  • Correlation Breakdown Episodes:
    • During the 2015 “Swiss Franc Shock,” when the SNB abruptly abandoned its EUR/CHF floor, the CHF strengthened sharply against the EUR and the USD. In the ensuing turbulence, the negative correlation between EUR/USD and USD/CHF temporarily weakened to around –0.4 as sudden liquidity constraints and capital flows distorted typical patterns.
    • Implication: Even well-established negative correlations can break down during policy shocks or extreme volatility. Traders should set alert thresholds (for example, if Pearson r moves above –0.6) to reassess hedges.
  • Data Source and Further Reading: For an in-depth historical correlation analysis, consult the BIS’s “Exchange Rate Co‐movements and Coordination” study (BIS Working Paper No. 123, 2015).
  • Trader Insight: During calmer market conditions, maintaining a simple EUR/USD/CHF hedge could reduce position risk by up to 60 per cent compared with holding a single currency pair. However, during the SNB shock, leveraged positions in both pairs led to substantial margin calls for many trading accounts.

Explore how these historical examples might apply to your own trading by opening a demo account and back-testing past correlation regimes.

Step-by-Step “How To” Guides: Computing and Interpreting Correlations

Understanding how to calculate and interpret correlation coefficients is essential for applying correlation analysis to your trading or investment decisions.

In this section, we outline two practical workflows—one using a spreadsheet (for example, Excel) and another using a charting platform (for example, MT4/MT5)—to help you retrieve historical data, compute rolling correlations, and draw meaningful conclusions.

Retrieving Historical Price Data

Choose Your Data Source:

• Download historical price data from a reputable provider (for example, Investing.com, Yahoo Finance or your broker’s data download tool).
• Select the date range that aligns with your analysis horizon (for example, past 1 year for a 30-day rolling correlation, or past 5 years for a long-term view).

Format Your Data:

• Ensure both datasets share the same frequency (for example, daily closing prices).
• Align dates by creating a master column of dates (for example, in column A) in Excel and using VLOOKUP (or INDEX/MATCH) to pull each symbol’s closing price into separate columns (for example, columns B and C).
• Remove any rows with missing values (for example, public holidays) so that you have a clean, continuous series for both symbols.

Computing a Simple Correlation Coefficient in Excel

1. Select Your Data Range:
   • Suppose column B contains “Asset A” daily returns (for example, EUR/USD) and column C contains “Asset B” daily returns (for example, USD/CHF).
   • Ensure cells B2:B and C2:C contain your aligned price returns (calculate returns as (today’s close – yesterday’s close) / yesterday’s close, or use Log Returns = LN(today’s close/yesterday’s close)).
2. Use the CORREL Function:
   • In a blank cell, type =CORREL(B2:B[n], C2:C[n]).
   • Press Enter to obtain the Pearson correlation coefficient for the entire sample window.
   • Interpretation: A result of –0.85 suggests a strong negative linear relationship; +0.30 suggests a weak positive relationship.
3. Tip on Return Calculation (Optional but Recommended):
   • Instead of raw prices, calculate daily percentage returns or log returns in separate columns. Use these columns as the input for the CORREL function to get a more accurate measure of co-movement.

Computing Rolling Correlations in Excel

1. Decide on a Rolling Window:
   • Common choices include 30-day, 60-day or 90-day windows, depending on your trading horizon. A 30-day window is often used for short-term traders; a 90-day window suits medium-term analysis.
2. Set Up the Rolling Calculation:
   • In cell D31 (assuming row 2 is your first return observation), enter: =CORREL(B2:B31, C2:C31)
   • This computes the 30-day correlation for rows 2 through 31.
   • Copy the formula down column D, so that D32 computes =CORREL(B3:B32, C3:C32), and so on, until the end of your dataset.
3. Plot the Rolling Correlation (Optional):
   • Highlight column D (the rolling correlation values) and insert a line chart to visualise how the correlation changes over time.
   • Interpretation: Watch for periods when the line approaches +1 (strong positive) or –1 (strong negative); sudden swings often signal regime shifts.
4. Interpret the Results:
   • A consistently negative rolling correlation (for instance, –0.8 or below) suggests reliable hedging potential.
   • Periods when the rolling correlation moves toward zero or changes sign indicate that historic relationships may be breaking down—consider adjusting your hedges or diversification accordingly.

Computing Correlations on MT4/MT5

Below is a step-by-step guide for both platforms, covering how to retrieve price data, apply built-in or custom correlation indicators, export data for spreadsheet analysis, and set alerts.

Retrieving Historical Data via the History Centre

1. Open the History Centre:
   • In MT4, click Tools → History Centre (or press F2).
   • In MT5, click View → Symbols, then select your symbol and choose Download. Alternatively, click Tools → History Centre.
2. Select the Instrument and Timeframe:
   • In the History Centre window, choose the desired symbol (for example, EURUSD) and timeframe (e.g., D1 for daily bars, H1 for hourly bars).
   • Click Download (MT4) or Download/Update (MT5) to fetch missing data from the broker’s server.
3. Export Data to CSV (If Required):
   • Highlight the relevant timeframe (e.g., EURUSD D1) and click Export (MT4) or Export to CSV (MT5).
   • Save the file to a local folder. Repeat this process for your second symbol (for example, USDCHF).
4. Format Exported Data:
   • The CSV will contain columns like Date, Time, Open, High, Low, Close, Volume. To compute correlations in a spreadsheet, import both CSV files into Excel (or Google Sheets).
   • Align dates in a master-date column, calculate daily returns (e.g., (Close_t – Close_t–1) / Close_t–1), and proceed with Excel’s =CORREL function. as described in our Excel Correlation Guide.

Using the Built-In Correlation Coefficient Indicator

Both MT4 and MT5 offer a built-in “Correlation Coefficient” indicator that calculates rolling Pearson r between two instruments:

1. Add Both Instruments to Your Workspace:
   • Open two chart windows (for example, EURUSD in one and USDCHF in the other). In MT5, you can also tile charts vertically by clicking Window → Tile Vertically.
2. Attach the Correlation Indicator:
   • In MT4:
     1. Open the Navigator panel (Ctrl+N) and expand Indicators.
     2. Scroll down to Correlation Coefficient (under “Built-in” or “Custom” if not present by default).
     3. Drag Correlation Coefficient onto your EURUSD chart.
   • In MT5:
     1. Open Navigator (Ctrl+N), expand Indicators, and locate Correlation Coefficient.
     2. Double-click to attach to your EURUSD chart.
3. Configure Indicator Parameters:
   • Symbol to Compare: Enter the second instrument’s ticker (for example, “USDCHF”).
   • Period/Length: Define the look-back window (for instance, 30 for a 30-day rolling correlation on daily bars, or 100 for a 100-period rolling correlation on H1 bars).
   • Applied Price: Choose which price to use (Close is standard, but you can also use Typical Price or Median Price).
   • Shift (Optional): Leave at 0 unless you wish to compare with shifted data.
   • Click OK to apply. The indicator will appear in a sub-window below your main chart, plotting values between –1 and +1.
4. Interpreting the Correlation Plot:
   • Values near +1 indicate a strong positive linear relationship; values near –1 indicate a strong negative relationship.
   • Hover your cursor along the correlation line to see exact correlation values on specific dates.
   • Trading Insight: If the line moves toward zero from historically extreme levels (for example, your usual –0.8 to –0.6 threshold), evaluate whether hedges remain effective.

Applying Custom or Multi-Pair Correlation Indicators

If you wish to compare more than two instruments or display a correlation matrix heat-map:

1. Install a Custom Correlation Matrix Indicator:
   • Search online (for example, the MQL5 Market or MQL4 Code Base) for “Correlation Matrix MT4” or “Correlation Heat Map MT5.”
   • Download the .ex4 (for MT4) or .ex5 (for MT5) indicator file.
   • Copy the downloaded file into your MQL4/Indicators or MQL5/Indicators directory (for example, C:\Program Files\MetaTrader 4\MQL4\Indicators).
2. Compile and Attach the Indicator:
   • Restart your MT4/MT5 terminal.
   • Open Navigator → Indicators → Custom, and locate the new correlation matrix indicator.
   • Drag it onto any chart; a settings window will prompt you to select multiple symbols (up to 10 pairs or instruments) and the data period (e.g., M15, H1, D1).
   • Configure colour thresholds (for example, green for correlations > +0.7, yellow for between +0.3 and –0.3, red for < –0.7) or accept defaults, then click OK.
3. Understanding the Heat-Map Display:
   • The indicator will display a grid showing correlation coefficients among all selected symbols. Each cell’s colour intensity signals the strength and direction of the correlation.
   • Use this at a glance to identify diversification opportunities—green cells (strong positive correlation) warn that assets move similarly, while red cells (strong negative correlation) highlight potential hedges.

Exporting Correlation Data and Alerts

Exporting Correlation Values

1. Enable Indicator Data Logging:
   • Some built-in and custom indicators allow you to log the correlation values to the Experts or Journal tabs.
   • In the indicator’s properties window, look for an option such as “Write to File” or “Enable Logging,” and specify a file name (e.g., EURUSD_USDCHF_corr.csv).
   • If no logging option exists, you can manually copy values:
     1. Right-click the sub-window where the correlation line is plotted.
     2. Choose Properties → Common, and check “Allow DLL imports” (if required by the custom indicator).
     3. After enabling logging, correlation values for each bar should be saved to the specified file in your MQL4/Files or MQL5/Files folder.
2. Open the Saved CSV File:
   • Via File → Open Data Folder in MT4/MT5, navigate to MQL4/Files (or MQL5/Files), and locate your CSV.
   • Import this into Excel for further analysis—plot trends, calculate summary statistics, or overlay with other time series.

Setting Alerts on Correlation Thresholds

1. Right-click on the Correlation Indicator:
   • In the correlation sub-window (either built-in or custom), right-click the indicator line and select Trading → Alert (MT5) or Alert on Indicator (MT4).
   • If this option isn’t immediately visible, ensure the indicator code includes Alert() or AlertCondition() functions—otherwise, you may need to modify the indicator’s source or use a custom alert script.
2. Configure the Alert Condition:
   • In the Alert tab, set:

• • • Condition: For example, “Correlation Coefficient (EURUSD vs USDCHF) > –0.6” or “Correlation Coefficient < +0.8.”
    • Value: Enter the numerical threshold (e.g., –0.6).
    • Source: Ensure the indicator’s name appears.
    • Action: Choose Pop-up, Sound, Email, or Notification (for MT5 mobile app).

1. • Click OK to confirm. When the correlation line crosses the specified level, MT4/MT5 will trigger the chosen alert.
2. Verifying the Alert:
   • Open the Terminal window (Ctrl+T) and select the Alerts tab. Confirm that your new alert appears, with status set to Enabled.
   • You can test the alert by temporarily lowering the threshold to a level the current correlation already meets. If it fires, reset it to your desired threshold.

Interpreting Correlation Outputs on MT4/MT5

• High Positive Correlation (Near +1): Assets are moving almost identically. In such a case, adding both to a portfolio offers little diversification benefit.
• High Negative Correlation (Near –1): Assets act as natural hedges. For instance, if the EURUSD vs USDCHF correlation is –0.85, profits on one position may offset losses on the other in risk-off scenarios.
• Correlation Drift Toward Zero: When a historically strong positive or negative relationship weakens (for example, moving from –0.85 to –0.4), consider rebalancing hedges or diversification strategies.
• Volatility Clusters: During periods of heightened volatility (for example, around major news releases or central bank decisions), correlations may spike either positive or negative. Always interpret correlation values in the context of broader market conditions.

Summary, FAQs, and Next Steps

Key Takeaways

• Correlation Is Not Causation: Correlation measures co-movement but does not prove that one asset’s price change causes another. Always watch for spurious relationships or third-factor influences.
• Relationships Evolve Over Time: Historical correlations can shift during regime changes. A diversified portfolio today may become highly correlated under market stress, so use rolling windows and monitor regularly (Section 5).
• Diversification and Hedging Benefits: A correlation matrix helps identify low- or negatively correlated assets. Combining such assets can reduce overall portfolio volatility. Conversely, strongly negative pairs (for example, EUR/USD vs USD/CHF) can serve as hedges when markets turn.
• Choosing the Right Look-Back Period: Short windows (e.g., 30-day rolling) capture recent dynamics but can be noisy. Longer windows smooth out noise but may lag behind changing conditions. Select a window aligned with your trading horizon.
• Platform Implementation Matters: Whether using MT4/MT5 or another platform, ensure you understand indicator settings (symbol selection, look-back length, applied price) and set alert thresholds to catch correlation breakdowns .
• Always Contextualise Correlation: Correlation values can spike during events such as central-bank announcements, geopolitical shocks, or liquidity crunches (Section 4). Interpret numbers alongside volatility measures and market news.

Next Steps

To get started, open a free demo account with AvaTrade and apply the Correlation Coefficient indicator on MT4 or MT5.

Try different look-back periods, like 30 or 60 days, to see how asset relationships change in real time.

While you explore, visit the Diversification Strategies and Risk Management Basics pages to build a solid foundation.

Finally, set up simple alerts on MT4/MT5 so you know immediately when key correlations move beyond the levels you care about, and don’t hesitate to contact our support team if you need assistance.

FAQ

• What does it mean if two assets have a positive correlation?

  It means they tend to move in the same direction—when one rises, the other usually rises too.

   
• Why would correlations change over time?

  Market conditions shift, so relationships that held last month may not hold today, especially during big events.

   
• How do I use a rolling window for correlation?

  A rolling window simply recalculates correlation over a recent time frame (for example, the last 30 days) so you see how the relationship evolves.

   
• Is zero correlation good or bad?

  Zero correlation means two assets don’t move together in a predictable way; it can help with diversification, but doesn’t guarantee safety.