US Composite Leading Indicator Model

October 13, 2023


Tactical asset allocators and global macro traders routinely monitor macroeconomic indicators to assess the current and prospective state of economic growth. The rationale behind this practice is that, by gauging the economic landscape, they can also make informed predictions about the financial markets. Consequently, investors seek to position themselves more defensively or aggressively based on the prevailing economic conditions.

Systematic investors also frequently employ macroeconomic indicators to generate signals for tactical market positioning. There exists a multitude of approaches for crafting trading signals grounded in economic indicators. Some investors rely on composite recession signals, which we have previously discussed here. Additionally, certain investors adopt an economic trend methodology, capitalizing on the persistent influence of new information on asset prices by aligning their positions with trends in macroeconomic fundamentals, as highlighted in an intriguing paper by AQR available here.

In this article, we leverage the concept of an economic trend to formulate a systematic trading model that positions itself between the S&P 500 index and a risk-off asset such as T-bills or Treasuries. To this end, we utilize five leading macroeconomic indicators. It's important to note that there are numerous macroeconomic series that could potentially serve as the basis for constructing trading signals, making the selection of macroeconomic indicators somewhat arbitrary.

In this post, we focus on a select few indicators, primarily chosen due to their extended historical data availability dating back to 1960, their frequent monitoring by investors, and, in most cases (with one exception), their composite nature. A composite indicator is fashioned by aggregating individual indicators into a unified index, guided by an underlying model that captures the multidimensional concept it seeks to measure. The constituents of a composite indicator are often referred to as components or component indicators.


To construct our models, we used the following data sources:

  • The Global OECD-indicators encompass data from 39 countries: including Australia, Austria, Belgium, Canada, Chile, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Israel, Japan, Italy, Korea, Luxembourg, Mexico, Netherlands, New Zealand, Norway, Poland, Portugal, Slovakia, Slovenia, Spain, Sweden, Switzerland, Turkey, United Kingdom, United States, Brazil, China, India, Indonesia, Russia, and South Africa. These indicators are designed to furnish early warnings of shifts in business cycles, reflecting fluctuations in economic activity around its long-term potential level. Subcomponents within the CLI encompass a diverse range, including housing permits, manufacturing levels and surveys, consumer confidence, job vacancies, money supply, retail trade... You can find the complete list of subcomponents per country here.
  • The Conference Board Leading Economic Index (LEI): The LEI serves as an early indicator of significant inflection points within the business cycle, offering insights into the near-term trajectory of the economy. The ten components of The Conference Board Leading Economic Index® for the United States consist of factors such as average weekly hours in manufacturing, average weekly initial claims for unemployment insurance, manufacturers' new orders for consumer goods and materials, the ISM® Index of New Orders, manufacturers' new orders for nondefense capital goods (excluding aircraft orders), building permits for new private housing units, S&P 500® Index of Stock Prices, Leading Credit Index™, interest rate spread (10-year Treasury bonds less federal funds rate), and average consumer expectations for business conditions. You can learn more about the LEI-index here.
  • US Term spreads: They refer to the difference between interest rates on short- and long-dated government securities. Often regarded as a predictor of business cycles, an inverted treasury curve has been shown in numerous academic studies to herald future recessions. While investors frequently focus on spreads like the 3m-10y spread or the 2y-10y spread, term spreads are calculated for various points on the interest rate curve, including 3 months, 6 months, 1 year, 2 years, 5 years, 10 years, and 30 years, resulting in 21 distinct term spreads.
  • The US unemployment rate, Although not typically a leading indicator, the unemployment rate is included in the analysis due to its extended historical data availability, investor attention, and its use in the well-known recession model, the Sahm-rule, which will be employed further in our discussion.

Our Systematic Composite Leading Indicator Model is constructed based on five distinct sub-models. Each of these sub-models employs a specific macroeconomic signal derived from the aforementioned data. The following macro trading signals are computed:

  1. Global OECD Diffusion Momentum Signal: To capture the global economic trend impacting the S&P 500 index, we compute a global OECD-diffusion signal. This metric gauges the percentage of countries within our sample exhibiting positive economic momentum. Economic momentum for each country is defined as the monthly difference in their local OECD indicator. If the percentage of countries displaying positive economic momentum surpasses 50%, the signal is deemed positive and indicative of a risk-on stance. Otherwise, it signals a risk-off posture.
  2. US OECD Momentum Signal: This signal assesses the strength of the US economic cycle and is calculated as the 2-month change in the US OECD indicator. When the 2-month change is positive, the signal reflects a positive (risk-on) stance, and vice versa.
  3. US Conference Board Leading Economic Index (LEI) Momentum Signal: Economic practitioners often scrutinize the rate of change in the LEI index to evaluate economic momentum. Here, we define economic momentum as the 6-month rate of change in the LEI index. Additionally, we compute a second derivative, the 1-month change in the smoothed 6-month signal, to discern if economic momentum is on the rise or in decline. A positive signal (risk-on) is generated when the 6-month change exceeds 0.
  4. US Yield Curve Diffusion Signal: This signal quantifies the percentage of all term spreads within our sample that are negative or inverted. A positive (risk-on) signal is issued when this percentage falls below 20%.
  5. Sahm-Rule Signal: The Sahm Rule identifies signals linked to the onset of a recession when the three-month moving average of the national unemployment rate (U3) experiences a rise of 0.50 percentage points or more relative to its low over the previous 12 months. Named after Fed economist Claudia Sahm, this signal has demonstrated its effectiveness in recognizing historical recessions, although international evidence presents mixed results.


The rules for the five sub-models are crafted based on the aforementioned macro-signals. For illustrative purposes, we occasionally incorporate a trend/momentum overlay onto an economic signal. It's crucial to recognize that negative economic signals may suggest a weaker economic environment, but this doesn't necessarily translate to a negative market (price) environment. The S&P 500 index, for instance, could still perform well despite a strongly negative economic signal. To address this, we sometimes employ price trend filters, meaning that a model only shifts to a risk-off position when both the economic signal AND the trend/momentum signal are negative.

Another critical aspect is the selection of the risk-off asset. These assets are expected to perform reasonably well during challenging market conditions. Commonly used risk-off assets include Treasuries, T-bills, gold, the US dollar, managed futures, and high-quality investment-grade corporate bonds. This choice can have a significant impact, as is evident, for example, in the current market environment. For instance, if an economic signal turns negative, and an investor switches to Treasuries while long-term rates continue to rise, the chosen risk-off asset may suffer. In such cases, T-bills or gold might prove to be more prudent alternatives. In this context, we primarily focus on T-bills and Treasuries as they are the most prevalent risk-off assets. The decision between the two can be determined through momentum analysis, where the investor selects the risk-off asset with the most favorable current momentum.

For the sake of true comparative analysis, it may initially appear more appealing to maintain consistency across all models by using a standardized trend overlay. However, we also aim to emphasize the advantages of strategy diversification between sub-models. Since we cannot predict if a trend-overlay will perform better in the future, it may be advisable to apply a trend overlay to only a select few models. Strategy diversification has the potential to yield more favorable long-term results. The same can be said about signal diversification. For example, there are dozens of ways to measure trend and/or momentum. Since, it is hard to predict which trend parameter will work best, it could be advisable to use different measures of trend between sub-models. Next, we illustrate a couple of variations so that the reader can better comprehend the subtleties involved in designing systematic models.

The Global OECD Diffusion Momentum Model:

  • If the Global OECD Diffusion Momentum Signal is positive (more than 50% of countries have a positive economic momentum), the strategy invests in the S&P 500 index.
  • If the Global OECD Diffusion Momentum Signal is negative (less than 50% of countries have a positive economic momentum), the strategy is invested in T-bills.

The US OECD Momentum Model:

  • If the US OECD Momentum Signal is positive (the 2-month difference in the OECD-indicator is positive), the strategy invests in the S&P 500 index.
  • If the US OECD Momentum Signal is negative (the 2-month difference in the OECD-indicator in negative), the strategy switches to trend-modus. If the price trend (defined here as the 50-week moving average) of the S&P 500 is positive, the strategy remains invested in the S&P 500 despite the negative economic signal. If the price trend is negative, the strategy switches to T-bills.

The US Conference Board Leading Index Momentum Model:

  • If the economic momentum (6-month rate of change in the LEI) is positive, the strategy invests in the S&P 500 index.
  • If the economic momentum (6-month) is negative, but recovering (the 1-month change in the smoothed 6-month change is positive), the strategy invests in the S&P 500 index.
  • If the economic momentum (6-month) is negative, but worsening (the 1-month change in the smoothed 6-month change is negative), the strategy switches risk-off and invests in T-bills.

The US Yield Curve Diffusion Model:

  • If the Yield Curve Diffusion Signal is positive (less than 20% of term spreads are inverted/negative), the strategy invests in the S&P 500.
  • If the Yield Curve Diffusion Signal is negative (more than 20% of the term spreads are inverted/negative), the strategy switches to risk-off and invests in either T-bills or Treasuries depending on the composite momentum score of these 2 risk-off assets. Momentum is defined here as the average return over the past 3m, 6m and 12m.

The US Sahm-Rule Model:

  • If the Sahm-rule Signal is positive (the difference between the 3-month average unemployment rate and the 12-month prior low is smaller than 0.5%), the strategy invests in the S&P 500 index.
  • If the Sahm-rule Signal is negative, the strategy switches to momentum-modus. It allocates to the asset with the best composite momentum (3m, 6m and 12m average return). The model can choose between 3 assets: S&P500, Treasuries or T-bills.

The US Composite Leading Indicator model:

  • The strategy allocates 20% of capital to each of the 5 models above. Rebalancing occurs monthly.
  • If for example four models are risk-on and one model is risk-off in T-bills, the model invests 80% in the S&P 500 index and 20% in T-bills


We offer an overview of the outcomes derived from the five sub-models. For clarity and illustration, we occasionally draw comparisons with the S&P 500 index and a more stringent 60/40 portfolio. Additionally, the ultimate Composite Leading Indicator Model is assessed against a 60/40 portfolio for reference.

1. Global Diffusion Momentum Model

Approximately 70% of the countries within our sample exhibit positive economic momentum, as indicated by the red line chart below. Given that the global economic momentum signal surpasses the 50% threshold, the strategy is presently allocated to the S&P 500 index.

The strategy has delivered a commendable performance relative to the S&P 500, showcasing superior returns and reduced risk. Notably, drawdowns have been considerably more contained in comparison to the S&P 500 index. Moreover, the strategy exhibited resilience during the turbulent past two years. While the strategy outperformed across the entire dataset, it's important to acknowledge that there were periods of underperformance extending beyond a decade. Nevertheless, even during these phases, drawdowns and volatility remained comparatively restrained.


The current two-month rate of change for the US OECD indicator is in positive territory. Consequently, the strategy is presently allocated to the S&P 500 index.

The strategy has surpassed the S&P 500 benchmark in both risk-adjusted returns and overall returns. It effectively sidestepped substantial drawdowns during critical periods, such as the recessions in the 1970s and the 2000s. Moreover, it exhibited resilience over the past two years and managed to steer clear of the market turmoil experienced in 2022.

3. US Conference Board LEI Model

The current six-month change in the US leading index, as depicted in the red line chart below, registers below 0%, indicating a negative economic momentum. Nevertheless, there are signs of an economic momentum recovery, with the one-month change now in positive territory. Consequently, the strategy is currently allocated to the S&P 500 index.

The sub-strategy consistently outperforms our S&P 500 benchmark, both in terms of returns and risk-adjusted performance. While it experienced comparatively higher drawdowns in comparison to the preceding two models, this was offset by achieving superior returns. It's noteworthy that the strategy hasn't outpaced the S&P 500 benchmark in terms of pure returns since 2008, but it has excelled in preserving capital by avoiding the significant crash of that year, resulting in a more favorable risk-reward profile.

However, the strategy did encounter some challenges, albeit to a lesser extent than the S&P 500, during the bear market of 2000-2003, which was primarily a consequence of the technology bubble's burst. Economic indicators often face difficulty in promptly detecting such events, as the technology crash commenced before the actual economic downturn.

4. US Yield Curve Diffusion Model

Approximately 80% of the yield curve is currently exhibiting an inverted trading pattern, surpassing the 20% threshold, indicating a risk-off environment. In such conditions, the model has the option to allocate funds to either Treasuries or T-Bills, guided by momentum. Given that the momentum score favors T-Bills, the strategy is presently invested in T-Bills.

For illustrative purposes, we are now comparing the model with a more stringent benchmark, the 60/40 allocation (60% equity, 40% treasuries). This benchmark provides a heightened emphasis on risk-adjusted performance. While the sub-model outperforms both in terms of returns and risk-adjusted returns, there are several noteworthy considerations:

  1. The strategy is susceptible to significant drawdowns in comparison to the 60/40 benchmark.
  2. The strategy exhibited subpar performance relative to the benchmark between 1983 and 2009, spanning almost three decades.
  3. A notable resurgence in the strategy's performance is evident from 2010 onward.

The primary concern here revolves around the duration of the signal. Inverted yield curves often serve as a precursor to a looming recession, but they occasionally revert to positive territory just as the recession becomes apparent to all, leading the Federal Reserve to initiate rate cuts. Consequently, the strategy at times switches back to the S&P 500 during the initial stages of a recession, as observed in 2008. This period coincides with the market's most challenging phase. In a forthcoming blog post, we may consider an adjustment to the model by introducing a lag in the signal to account for this economic phenomenon. To maintain the integrity of our models and avoid the introduction of hindsight, we are currently presenting the results in accordance with the logical rules outlined previously.

An invaluable lesson investors can draw from this model, is to exercise continued vigilance during the initial six months following the restoration to a positive interest rate curve.

5. Sahm-rule Model

The present 3-month rolling average of the United States unemployment rate stands at 3.63%, in comparison to the 12-month low of 3.4%. Given that the spread between the two figures falls below 0.5%, the Sahm-rule indicator does not signal a recession, and the strategy is presently maintaining full investment exposure in the S&P 500 index.

Once more, we subject this model to a more rigorous evaluation by comparing it with the 60/40 benchmark. While the sub-strategy exhibited superior performance over the entire data set, several noteworthy considerations emerge:

  1. The strategy demonstrated relatively weaker performance between 1975 and 1994, spanning nearly two decades.
  2. Subsequently, the strategy consistently outperformed.
  3. It is noteworthy that the strategy's risk-adjusted performance, as indicated by the Sharpe and Sortino ratios, closely aligns with the benchmark. Drawdowns, on average, do not exhibit greater containment.
  4. The strategy primarily operates as a return enhancer, shifting towards equities during favorable economic conditions.

6. Final Composite Leading Indicator Model

The composite model allocates 20% of the capital to each of its constituent models, with monthly rebalancing. It is compared against a stringent 60/40 portfolio. The current portfolio allocation, based on the composite model's signals, is 80% invested in the S&P 500 index and 20% in T-bills, guided by the following sub-models:

  • Global OECD Diffusion Model: Risk-on allocation of 20% to the S&P 500 index.
  • US OECD Model: Risk-on allocation of 20% to the S&P 500 index.
  • US Conference Board Leading Index Model: Risk-on allocation of 20% to the S&P 500 index.
  • US Yield Curve Model: Risk-off allocation of 20% to T-bills.
  • US Sahm-Rule Model: Risk-on allocation of 20% to the S&P 500 index.

In our assessment of performance, the following observations are made:

  • The Composite Strategy outperformed the 60/40 benchmark with a 8.62% return, compared to 6.8% for the benchmark, resulting in a difference of 2% compounded annually.
  • The Composite Strategy exhibited superior risk-adjusted performance, boasting a Sortino ratio of 0.66 compared to 0.38 for the benchmark.
  • The composite model maintained a lower maximum drawdown of 21.39%, in contrast to the benchmark's 30.6% drawdown.
  • The Composite Strategy effectively navigated through significant bear markets, such as those in 1971, 1974, 2008, and 2022. However, it did not avoid the 1987 crash and experienced a drawdown similar to the benchmark during the 2000-2003 bear market. This is often attributed to the challenges economic indicators face in capturing short-term market crashes and bubble bursts.
  • The Composite Strategy showcases more consistent relative performance compared to its individual sub-models.

Overall, we are content with the outcomes of this initial model driven by economic reasoning. Patient investors may derive benefits from monitoring critical leading economic indicators and aligning their portfolios accordingly. Nonetheless, it is essential to maintain vigilance and a measure of humility when interpreting these results for the future.

It's important to note that this model primarily relies on macro factors as decision signals. We believe a comprehensive systematic portfolio should encompass strategies based on various factors and signals. In this context, this composite model could function as a sub-strategy within a well-diversified portfolio.

In hindsight, this model holds potential for improvement, but we have presented these results based on available data and our original methodology:

  • We focused on widely recognized composite leading indicators, except for the Sahm-Rule. Other indicators could potentially yield better performance.
  • For a more timely insight into the labor market, investors could consider incorporating more leading labor indicators, such as the Conference Board Employment Index.
  • It appears that the yield curve diffusion signal might be too early in switching back to a risk-on stance when the curve normalizes after inversion. Systematic investors might consider introducing a lag for re-entry purposes.
  • Systematic investors may experiment with different lookback periods for indicators.
  • The introduction of trend or momentum filters overlaid on economic signals is an option to consider for systematic investors.
  • Constructing rules across all models to determine the choice of risk-off assets (e.g., Treasuries, T-bills, gold, defensive equities, the dollar) could be explored more deeply by systematic investors.
  • Systematic investors could play around with several portfolio optimizers to determine the percentage allocation to each sub-model. For example, certain optimizers would allocate less to sub-strategies that show heightened volatility and/or correlation. Or an optimizer could allocate more weight to those strategies with strong relative momentum.