Correlation table for selected strategies available at Algohubb

At Algohubb, we believe that calculating the correlation based on monthly loss is the most practical approach. Our published research reflect this method:

We are only interested in the correlation for Loss, because if our profits are correlated (occur at the same time), we have no problem with that! However, we definitely want to avoid accumulating losses.

Correlation by Loss is calculated on StrategyQuant v140 or higher. If you are interested in the correlation for Profit/Loss, different time intervals, or correlation of seasonal strategies, you will find details below.

All strategies were tested:

• for the same period - 2010.01.01-2024.05.29

• using MM$ (the same ammount invested consistently in each strategy)
  

What is correlation?

Correlation is a statistical measure that determines how two sets of data move in relation to each other over a certain period of time. It can be positive (both variables increase together), negative (one variable increases as the other decreases), or zero (no relationship between the variables). Correlation is typically measured on a scale from -1 to 1, where:

• 1 indicates perfect positive correlation,

• -1 indicates perfect negative correlation,

• 0 indicates no correlation.
  

The most commonly used measure of correlation in finance and statistics is the Pearson method.

Interpretation of correlation map

On the above correlation map, you can see how different investment strategies are correlated based on monthly losses. The colors help visually distinguish the levels of correlation:

• Green (low correlation): Better diversification, as strategies have different drivers and do not move similarly.

• Orange (moderate correlation): Moderate diversification, some similarities in strategy movements.

• Red (high correlation): Poor diversification, strategies move very similarly.
  

Correlation values are presented in a table, where each cell shows the correlation between a pair of investment strategies. For example, the correlation between "Emerging Monthly Mover" and "Gap Jumper" is -0.30, indicating a moderate negative correlation.

The map assists in pinpointing strategies that can be smartly paired in a portfolio to minimize risk and boost the likelihood of consistent profits.

What is the best correlation between strategies?

The ideal correlation between investment strategies depends on the goal, but the leading expectation is usually Diversification and Risk Reduction.

To achieve this, we typically look for low or negative correlation between strategies.

• Low Correlation (close to zero): A correlation ranging from -0.2 to 0.2 suggests that two strategies are almost independent. Changes in the results of one strategy do not significantly affect the other. This correlation is ideal for diversification because it reduces the volatility of the entire portfolio.
  

• Negative Correlation: When one strategy makes money, the other might lose it, and the other way around. For example, a -0.5 correlation suggests this kind of relationship. This can be good during market ups and downs because gains from one strategy can balance out losses from another.
  

Factors we can influence in strategy creation process

The most important factors we can control in managing correlation are:

1. Instrument

Financial instruments on which the strategy is based are a key factor influencing correlations. Different instruments, such as stocks, bonds, commodities, or currencies, have different characteristics and react differently to changing market conditions.

• Stocks and their sectors: Fluctuating stock prices can influence correlations depending on the sector in which they are listed.

• Bonds: Mainly react to changes in interest rates and economic conditions.

• Commodities: Sensitive to changes in supply and demand, geopolitical events, and seasonality.

• Currencies: React to monetary policy, differences in interest rates, and macroeconomic indicators.
  

Fortunately, most of these asset classes can be found directly or indirectly in the form of ETFs, which can be scaled almost indefinitely. Even better diversification at the instrument level can be achieved using futures contracts, but they require a large capital and therefore also a lot of experience.

Also, remember that most instruments will be inversely correlated to the USD if it is quoted in it.

2. Strategy Type

This is an underrated but crucial factor that we can manage in building a low-correlation portfolio with ETFs and stocks. Different strategies, even on the same instrument, can complement each other well in changing market conditions.

• Trend Following Strategies: Focus on following dominant market trends.

• Mean Reversion Strategies: Based on the assumption that asset prices return to average values.

• Seasonality Strategies: Utilize market seasons in which they operate.

• Bias Strategies: Focus on identifying repeatable patterns or trends in asset price movements.

• Breakout Strategies: Explore short-term breakouts from key levels.

  
  

3. Trading Direction (Long/Short)

Trading direction, or the choice of a strategy that works on Long or Short positions, can be an important aspect in portfolio diversification. It affects how we make money and how we protect our portfolios from risk.

Inverse ETFs, like SQQQ, offer a practical option instead of shorting stocks. I'll cover the complexities of short strategies for stocks and ETFs in a separate document.

4. Time Interval

As algorithmic traders, we can choose the time interval our data is based on, which is crucial. For instance, correlations might change if our strategy works on an intraday, daily, or weekly basis.

• Short-term (e.g. intraday): May exhibit higher volatility and less stable patterns.

• Medium-term (e.g. daily): May show more stable patterns.

• Long-term (e.g. weekly): May reveal long-term trends.
  

Change in correlation over time

Remember that correlation can change over time. It's a good idea to review the correlation annually and adjust your portfolio as needed.

Summary

Understanding correlation and skillfully utilizing it can significantly improve investment results and reduce portfolio risk. Whether you aim for diversification, protection against downturns, or maximizing profits in specific market conditions, analyzing correlation is key to conscious and effective investing.

Remember that markets are dynamic and ever-changing, so regular analysis and adjusting strategies in response to current market conditions are essential for success. Investing is not only a science but also an art, and understanding correlation is one of its most important tools.

The module for building portfolio strategies and testing their correlation is included in StrategyQuant. Correlation by Loss can be properly calculated in SQL 140 or higher.

Other correlation results for Algohubb strategies

At Algohubb, we calculate the correlation based on monthly loss as shown. We focus on loss correlation since simultaneous profits aren't an issue for us. Our aim is to prevent accumulating losses. Monthly intervals offer a clearer view of correlations than daily or weekly data in our opinion.

To review the remaining results, here they are:

Correlation on Loss by Day

Correlation on Loss by Week

Correlation on Profit/Loss by Day

Correlation on Profit/Loss by Month

Correlation of seasonal strategies

In our view, research on correlation in Seasonal Strategies is most effective when done yearly. Here's the breakdown:

Correlation on Loss by Year