Back to Blog
Metrics

Trade Clustering & Correlation Analysis

Practice—Process·Trading Analytics
January 24, 20258 min read

Most traders think diversification means trading different instruments. In reality, two seemingly unrelated trades can be highly correlated—and a cluster of correlated positions can amplify your risk far beyond what you intended.

What Is Trade Clustering?

Trade clustering analysis groups your trades by how similarly they behave. Two trades "cluster" together when they tend to win or lose at the same time, regardless of the instruments involved. A long position in gold and a short position in the US dollar might cluster tightly because they respond to the same macro forces.

Practice—Process analyses your historical trades to identify these hidden relationships and presents them in an interactive correlation heatmap.

The Correlation Heatmap

The heatmap displays pairwise correlations between all your traded instruments (or strategies). Each cell ranges from -1 to +1:

| Correlation | Meaning | |-------------|---------| | +0.7 to +1.0 | Strongly correlated—wins and losses together | | +0.3 to +0.7 | Moderately correlated | | -0.3 to +0.3 | Weakly correlated—largely independent | | -0.7 to -0.3 | Moderately inversely correlated | | -1.0 to -0.7 | Strongly inversely correlated—natural hedge |

Reading the Heatmap

Bright warm cells indicate high positive correlation. Cool cells indicate negative correlation. The diagonal is always 1.0 (every instrument is perfectly correlated with itself).

Look for blocks of warm colour—these reveal clusters of instruments that move together. If you are simultaneously long on several instruments in the same cluster, your effective position is much larger than you might realise.

Concentration Risk

Concentration risk occurs when multiple open positions share the same underlying driver. Examples:

  • Long AAPL, MSFT, and GOOGL simultaneously (all tech, all correlated to NASDAQ)
  • Long crude oil futures while also long CAD/USD (both driven by energy prices)
  • Multiple momentum trades entered on the same euphoric market day
  • The clustering dashboard highlights when your open positions exceed a configurable correlation threshold, alerting you before concentration becomes dangerous.

    Diversification Score

    Practice—Process calculates a diversification score from 0 to 100 based on the average pairwise correlation of your open positions:

    | Score | Rating | Interpretation | |-------|--------|---------------| | 80–100 | Excellent | Truly independent positions | | 60–79 | Good | Reasonable diversification | | 40–59 | Fair | Some hidden correlation | | 20–39 | Poor | Significant concentration risk | | 0–19 | Critical | Positions behave as one large trade |

    Strategy-Level Clustering

    Beyond individual instruments, Practice—Process clusters your strategies. You might discover that your "breakout" and "momentum" strategies are essentially the same trade expressed differently. When both fail, they fail together—meaning your strategy diversification is illusory.

    The strategy cluster chart groups your approaches by outcome correlation, helping you identify which strategies provide genuine diversification and which are redundant.

    Temporal Clustering

    Trades can also cluster in time. If you tend to enter several positions within a short window (often driven by the same catalyst), those trades share timing risk. Temporal clustering analysis shows:

  • Average number of trades entered per session
  • How often multiple entries occur within 30 minutes of each other
  • Whether clustered entries perform differently from isolated entries
  • Using Correlation Data

    Position Sizing

    When two positions have a correlation above 0.6, treat them as a combined position for risk purposes. If your max risk per trade is 1% of capital, two correlated 1% positions represent roughly 1.7% total risk, not 2%.

    Hedging

    Identify instruments with strong negative correlation in your history. These serve as natural hedges during uncertain periods.

    Portfolio Construction

    Aim for a portfolio where the average pairwise correlation stays below 0.4. This ensures that drawdowns in one position are less likely to coincide with drawdowns in others.

    Understanding correlation is the difference between thinking you are diversified and actually being diversified. Let the data reveal what your intuition might miss.

    P
    Practice—Process
    Trading Analytics

    More from the Blog

    Metrics

    Equity Decomposition: Understanding Your P&L Drivers

    8 min read
    Psychology

    Rule Adherence: The Real Cost of Breaking Your Own Rules

    9 min read