Correlation Risk

Introduction

In I-042, we learned the concept of portfolio concentration.

We saw that a portfolio is not necessarily diversified just because it contains many different instruments. If these instruments react similarly to the same market event, the same risk factor can be present through several positions.

One of the most important reasons for this is correlation.

Correlation helps us understand what historical relationship the price movements of two or more instruments have shown with one another.

This is not simply a statistical curiosity.

As traders, it matters to us because if several positions move in a similar way, a single market event can affect several of our positions at the same time.

For example, we may have:

five different stocks,

five different entry points,

five different stops,

but if all five are sensitive to the same market factor, the portfolio risk may be much more concentrated than we initially think.

The purpose of I-043 is therefore to move beyond the thinking of “different positions” toward the thinking of “different risk factors.”

1. What Is Correlation?

Correlation describes the direction and strength of the co-movement between two variables.

In trading, we can primarily examine how the returns of two instruments have moved in relation to one another over a given period.

The correlation value is typically between −1 and +1.

+1: perfect positive co-movement.

0: absence of linear co-movement.

−1: perfect negative co-movement.

In practice, most market relationships are not perfect.

Therefore, we are much more likely to examine whether the relationship is weak, moderate, or strong.

2. Positive Correlation

With positive correlation, two instruments typically move in a similar direction.

If one rises, the other has a greater probability of rising as well.

If one falls, the other also falls more frequently.

Important: positive correlation does not mean identical movement.

Two stocks, for example, may move in a similar direction while changing by different amounts, turning at different times, and having different volatility.

Therefore, we should not interpret correlation as if two instruments were drawing the same chart.

3. Negative Correlation

With negative correlation, instruments typically move in opposite directions.

If one strengthens, the other more frequently weakens.

In theory, a −1 correlation would mean perfect opposite co-movement.

In the real market, however, such a perfect relationship is rare and is not necessarily stable over time.

Positions with negative or lower correlation can under certain conditions reduce the overall movement of a portfolio.

But this does not mean that such a position is automatically “safe.”

4. What Does Zero Correlation Mean?

Correlation around 0 means that during the period examined, we did not find significant linear co-movement.

It does not mean that there is no relationship of any kind, that the two instruments are completely independent, or that there will be no relationship between them in the future.

This is particularly important.

Correlation is a relationship observed using a particular measurement method, on particular data, over a particular period.

It is not an eternal law of the market.

5. Correlation and Diversification

One purpose of diversification is to prevent the same risk factor from determining the result of our entire portfolio.

If two positions are strongly positively correlated, they diversify one another less in certain respects.

If we combine positions with lower or negative correlation, the total portfolio co-movement may decrease in certain market environments.

But precise wording is important: low correlation does not equal low risk.

A low-correlation instrument can itself be extremely volatile or risky.

6. Why Is Correlation Important to a Trader?

Suppose we have three positions.

Each trade is 1% Risk.

At first glance, this may seem appropriate.

However, if the three positions are strongly correlated, a common market event can affect all three trades at once, activate several stops, and cause a rapid drawdown.

Correlation can therefore help us recognize: “How many different risk bets do we actually have?”

These are not necessarily the same thing: 5 stocks, 5 trades, 5 strategies, and 5 risk factors.

7. Correlation Should Not Be Judged by the Stock Names

It is easy to say: “These are two different companies, therefore they are two different risks.”

That is not necessarily true.

Two companies may be in the same sector, sensitive to the same index, follow the same commodity prices, react to a similar interest-rate environment, or follow the same economic cycle.

Therefore, legal or business separation between companies does not automatically mean separation of market risk.

8. Correlation Between Stocks and Indices

A stock often has a relationship with the index or market to which it belongs.

For example, a stock associated with Nasdaq may be more or less sensitive to the general movement of Nasdaq.

This does not mean that it moves exactly the same way every day.

The stock’s own company news, earnings, and other factors also matter.

At portfolio level, however, it is important to recognize: individual stock risk + common index risk.

If we hold many similar stocks, index movement can become a common risk factor.

9. Correlation Within a Sector

Sectors are particularly important.

Suppose we have five technology stocks.

Each is a separate company.

Yet they can share common sensitivities: technology-sector sentiment, interest-rate environment, Nasdaq, and repricing of growth stocks.

If the market suddenly reacts negatively to the technology sector, several positions may weaken at the same time.

Therefore, diversification within a sector may be limited.

10. Correlation Between Different Instruments

Correlation is not interpreted only between stocks.

We can examine, for example, a stock and an index, a stock and an ETF, a stock and the bond market, a currency and a stock index, gold and other market assets, or oil and energy stocks.

The goal is not to analyze every possible relationship at once.

It is to recognize the relationships that can materially affect the risk of our own portfolio.

11. Correlation Is Time-Dependent

This is one of its most important properties.

The correlation of two instruments can be high during one period and significantly lower later.

For example: 6 months: strong correlation; 1 month: moderate; 1 week: weak.

All three numbers can be mathematically correct.

We are simply measuring different periods.

Therefore, we must always know the time window on which we are discussing correlation.

12. Correlation and Timeframe

Timeframe also influences the relationship we see.

Two instruments may be strongly correlated on daily data but much less correlated on a shorter intraday timeframe.

This is because different factors can dominate over different time horizons.

Therefore, if we manage swing trades, a correlation analysis related to the time horizon of our swing positions may be more relevant to us.

The measurement horizon should therefore fit our trading system.

13. Return Correlation and Price Correlation

It matters what data we use to calculate correlation.

From a risk perspective, the co-movement of returns is often more informative than simple correlation of absolute prices.

For example, we do not merely want to know whether two stock prices move at similar levels.

We want to examine how their returns reacted to the same market movements.

This is closer to the real question of portfolio risk.

14. Correlation Is Not Causation

If two instruments are strongly correlated, this does not prove that the movement of one causes the movement of the other.

There may be a third common factor behind them.

Two stocks may move together because they belong to the same sector, are moved by the same index, or are sensitive to the same interest-rate environment.

Therefore, we should interpret correlation as a relationship, not automatically as a causal mechanism.

15. Correlation Under Stress

The correlation measurable in a normal market environment does not necessarily show precisely what happens in an extreme situation.

During a major market shock, the movements of several instruments can suddenly become more similar.

This can be particularly dangerous because the benefit of diversification may decrease exactly when we need it most.

Therefore, we should not only ask what may happen under normal conditions.

We should also ask: “What can happen under stress?”

16. Correlation and Volatility Together

Correlation and volatility are two different concepts.

Volatility describes how much an instrument moves.

Correlation describes the relationship between the movements of two instruments.

Two instruments can therefore be high-volatility and highly correlated, low-volatility and highly correlated, or high-volatility and low-correlated.

These represent completely different portfolio situations.

Therefore, we must not confuse the two concepts.

17. Three Positions, Three Different Risks

Suppose: A: 1R; B: 1R; C: 1R.

Total: 3R initial risk.

If the three positions have low correlation, the risk structure may be different from a situation in which all three move strongly together.

In the second case, a common market event can affect several positions at the same time.

Therefore, 3R does not necessarily represent the same thing in every portfolio.

18. Joint Sizing of Correlated Positions

If several positions are strongly correlated, the question may arise: “Do we use the same Risk % for each?”

There is no universal answer that applies to every system.

A portfolio system may decide that every trade is 1%, but the combined risk of trades connected to one common factor is limited.

Another system may reduce position sizes.

Another may allow fewer positions at the same time.

What matters is that the rule is defined in advance.

19. Simple Example of Correlated Positions

Suppose the account is 40,000 USD and maximum trade risk is 1% = 400 USD.

We find three strongly correlated technology stocks.

If we risk 400 USD on each: 3 × 400 = 1,200 USD.

This is 3% initial portfolio risk.

A portfolio-level rule might state, for example, that the combined risk of positions connected to the same factor may not exceed 1.5%.

Then we do not treat all three positions as completely independent 1% risks.

The exact limit is naturally part of our own system.

20. Successful Case Study

We work with a 60,000 USD account.

Our system allows maximum risk of 0.75% = 450 USD per trade.

We find three setups: a technology stock, a healthcare stock, and an energy stock.

All three setups are valid.

However, based on correlation analysis, we do not treat all three positions in exactly the same way.

Alongside the technology position, the energy and healthcare positions represent different risk factors.

The total portfolio risk remains controlled.

The success is not that we opened “as many trades as possible.”

The success is that we evaluated the selection and risk of the trades at portfolio level.

21. Failed Case Study

In a 40,000 USD account: Trade A: 1%; Trade B: 1%; Trade C: 1%; Trade D: 1%.

Total: 4% initial risk.

All four stocks are connected to the same technology segment.

The trader thinks: “I have four different stocks.”

After a negative sector-wide event, all four positions fall at the same time.

The problem was not necessarily that the individual trades were too large.

The problem was that essentially the same risk factor stood behind four supposedly separate trades.

22. Correlation and the Idea of a Hedge

Negative or lower correlation can in some circumstances provide an opportunity to reduce risk.

This is generally referred to as hedging.

However, no hedge is automatically a perfect protection.

The relationship between the two positions can change, sizing can differ, and the hedge has its own costs and risks.

Therefore, we should not interpret correlation as: “Negative correlation = guaranteed protection.”

23. Correlation Risk and the Stop Loss

The stop remains an important risk control for every individual trade.

But if several correlated positions hit their stops at the same time, the portfolio loss is added together.

Therefore, proper stop placement does not replace portfolio-level correlation analysis.

The two controls answer different questions: Stop: “When is our trade hypothesis invalid?” Correlation: “How related is the risk of several trades?”

We may need both.

24. Correlation and Position Sizing

Based on I-039–I-041, we already know: Setup → Hypothesis → Invalidation → Stop → Risk → Position Size.

Portfolio-level thinking adds: Setup → Hypothesis → Invalidation → Stop → Risk → Position Size → Correlation → Portfolio Risk.

Correlation therefore does not necessarily override the individual trade calculation.

Rather, it helps determine what overall risk structure results from the individual trades.

25. When Should We Reassess Correlation?

We do not need to recalculate correlation every minute.

It can be more important to reassess when the market environment changes significantly, when we enter a new sector, when we open several similar positions, when the strategy changes, when volatility increases, when a major macroeconomic event approaches, or when the portfolio structure changes materially.

The frequency of measurement should be defined by our own system.

26. Limitations of Correlation Measurement

Correlation is useful, but it is not perfect.

Important limitations include: it uses historical data; it depends on the time window; it depends on timeframe; it may not represent extreme moves well; it may not capture nonlinear relationships; and it does not establish causation.

Therefore, correlation should be used not as a standalone decision system but as risk-management information.

27. The Basic Idea of a Correlation Matrix

If we have several instruments, we can create a simple correlation matrix.

For example: A/B = 0.82; A/C = 0.15; A/D = −0.20; B/C = 0.18; B/D = −0.10; C/D = 0.35.

From the matrix we can quickly see that A and B move strongly together, C is less connected to A and B, and D may move differently in certain respects.

The matrix does not provide an automatic trading signal.

It can, however, be very useful for reviewing the risk structure of a portfolio.

28. Correlation and Strategy Diversification

We should not only examine instruments.

Strategies can also be correlated.

For example: breakout, momentum, and trend following.

All three strategies may work well in the same market environment.

If that environment disappears, the performance of all three may deteriorate.

Therefore, one question of strategy diversification is: “Are we using different rules, or are we actually managing different sources of risk?”

29. Correlation and Drawdown

If correlated positions become losses at the same time, drawdown can increase more quickly.

This can be especially problematic when individual Risk % is high, many positions are open, correlation is high, volatility is high, or leverage is used.

Correlation can therefore indirectly affect the speed and magnitude of the account drawdown.

30. The Professional Approach

In professional portfolio management, the goal is not necessarily to reduce every correlation to zero.

That is not always possible or even necessary.

The goal is rather to identify important common factors, measure the relationships, understand their limitations, define portfolio-level risk limits, and regulate positions accordingly.

Correlation is therefore not an independent trading signal, but one information layer in risk decision-making.

31. Common Mistakes

1. “Different stock = different risk.” Not necessarily.

2. Treating a single correlation value as permanently valid. Correlation changes.

3. Not defining the time window. A 20-day and a 2-year relationship are not the same.

4. Not matching the analysis to the timeframe. Different relationships may be relevant for swing and intraday trading.

5. Confusing correlation with volatility. They measure different things.

6. Treating correlation as causation. Co-movement alone does not prove causality.

7. Treating negative correlation as a guaranteed hedge. The relationship can change.

8. Looking only at individual trade risk and not the combined correlated risk.

9. Relying too heavily on statistics. Correlation comes from historical data and does not replace analysis of the market environment.

32. The Essence of Correlation Risk

Correlation helps us understand what relationship the returns of several instruments have shown with one another.

For us, however, the most important question is not the correlation number itself.

It is: “How much common risk do our positions carry?”

A portfolio may contain many instruments, many trades, and many strategies while having only a few genuinely different risk factors.

Professional thinking therefore does not stop at individual position sizing.

In I-039–I-041 we learned how to size a trade.

In I-042 we learned to recognize portfolio concentration.

In I-043 we examined how these positions can be connected and how the same risk factor can appear in several trades at once.

The goal is not to eliminate every correlation.

The goal is to know when several trades are actually built on the same market risk.

Practical Exercise

Select at least 10 instruments from an actual or hypothetical portfolio.

For each instrument, collect: instrument name, timeframe, sector, position direction, position size, Risk %, 1R, nominal exposure, related index, related sector, volatility, and the correlation measurement window.

Then:

Correlation groups — Find at least three instruments between which we assume a strong positive relationship.

Correlation matrix — Create a simple correlation matrix for the selected instruments.

Common factor — For every strongly correlated group, identify the possible common risk factor.

Stress test — Assume that the most strongly correlated positions all hit their stops because of a common market event. Calculate the USD loss, account-percentage loss, and total R loss.

Portfolio-level decision — Decide whether to reduce position size, skip a trade, use a lower Risk %, or change the rule for correlated exposure.

Time-window comparison — If possible, compare the correlation of the same instruments over two different windows and observe how much the picture changes.

The goal is not to find a “perfect” correlation value. The goal is to recognize how many positions in the portfolio are actually backed by the same risk factor.

Key Terms

Correlation Risk; Correlation; Positive Correlation; Negative Correlation; Portfolio Concentration; Diversification; Portfolio Risk; Exposure; Risk per Trade; Position Sizing; Volatility; Drawdown; R-multiple; 1R; Correlation Matrix; Time Window; Timeframe; Hedging; Common Risk Factor.

Knowledge Check

1. What is correlation, and what are we trying to measure with it?

2. What do positive and negative correlation mean?

3. Why does correlation around 0 not mean that two instruments are completely independent?

4. Why is correlation important when examining portfolio diversification?

5. Why do two different companies not automatically represent different risks?

6. Why can it be important to examine the relationship between stocks and indices?

7. Why can correlation change over time?

8. Why do timeframe and the selected time window matter when interpreting correlation?

9. What is the difference between correlation and volatility?

10. Why does high correlation not automatically mean causation?

11. Why can it be dangerous if correlation changes during stress?

12. How can a correlation matrix help analyze portfolio risk?

13. Why might a portfolio-level risk limit be necessary for strongly correlated positions?

14. Why does negative correlation not automatically mean perfect hedging?

15. What is the main goal of managing correlation risk in a professional trading system?

Az Akadémia szöveges tartalmának részben, vagy egészben történő másolása, többszörözése, átdolgozása, terjesztése, valamint nyilvános bemutatása a szerző előzetes, írásos hozzájárulása nélkül tilos.