Perfect Recession Predictors
Why popular treasury spread benchmarks do not work & what indicators work instead.
TLDR
A groundbreaking study, Perfect Recession Predictors (2024), identifies 83 flawless recession signals within the U.S. Treasury yield curve, outperforming traditional indicators like the 10-year minus 2-year spread and the Near-Term Forward Spread (NTFS).
These perfect predictors, primarily longer-term forward spreads (e.g., 53-month forward rate minus 1-month rate) with a 1-year moving average, have predicted every U.S. recession since 1962 without errors.
As of August 2023, these indices signal a 95.5%–99.9% probability of a recession by August 2025, offering traders a robust framework for navigating economic cycles.
Introduction
Predicting recessions is one of the most challenging yet rewarding tasks for professional traders.
Traditional indicators like the 10-year minus 2-year Treasury spread or the Near-Term Forward Spread (NTFS) have been widely used for decades, but they are far from perfect.
A groundbreaking study, Perfect Recession Predictors (2024), revolutionizes this field by identifying 83 flawless recession signals hidden within the U.S. Treasury yield curve.
This blog dives deep into the paper's findings, offering actionable insights and detailed takeaways for traders to refine their strategies and gain a competitive edge.
The Problem with Traditional Indicators
The study begins by highlighting the shortcomings of popular yield curve spreads.
The 10-year minus 2-year spread, for example, has generated 26 false positives since 1962, while the 10-year minus 3-month spread has produced 32 false positives. The NTFS, which measures the 18-month forward spread, has an even worse track record with 59 false positives.
These spreads often signal recessions that never materialize, leading to costly missteps for traders.
The authors argue that these indicators are too simplistic and fail to account for the full complexity of the yield curve.
The Quest for Perfection
The study defines a perfect recession predictor as one that generates no false positives and no false negatives.
Using daily yield curve data from 1961–2023, the authors tested 645 million combinations of Treasury spreads (term and forward) with varying maturities and moving averages.
Their supercomputer-powered analysis revealed 83 perfect spreads: 58 forward spreads and 25 term spreads.
These predictors successfully identified all 8 U.S. recessions since 1962 without any errors.
The perfect spreads share several common features.
Most are forward spreads with rates starting approximately four years into the future, such as the 53-month forward rate minus the 1-month rate.
Additionally, the short rates used in these spreads almost always have a 1-month maturity.
The study also found that applying a 1-year moving average (258–268 business days) to these spreads significantly reduces noise and improves accuracy.
In terms of performance metrics, the perfect predictors scored a flawless 1.0 on the Area Under the Receiver Operating Characteristic (AUROC) curve, compared to scores of 0.63–0.69 for common spreads.
Bootstrap validation further confirmed their superiority, with perfect spreads averaging only 2.8 mispredictions versus 14–34 for traditional benchmarks.
Why Forward Spreads Outperform
The perfect predictors are predominantly forward spreads, such as the 53-month forward rate minus the 1-month rate.
These longer-term forward spreads outperform traditional term spreads for several reasons.
First, the yield curve is generally upward-sloping, meaning longer-term rates embed higher term premiums.
This reduces the likelihood of false positives because inversions (where long-term rates fall below short-term rates) are rarer.
For example, a 4-year forward rate is less volatile and more resistant to transient shocks than near-term spreads.
Second, the study emphasizes the importance of moving averages.
Smoothing daily data with a 1-year moving average (250–260 trading days) filters out short-lived inversions caused by noise, such as monetary policy shocks.
This approach aligns with the empirical finding that a 1-year moving average window minimizes mispredictions.
Third, the economic mechanism behind these perfect predictors is rooted in the relationship between yield spreads and economic growth.
Persistent productivity shocks drive a positive correlation between yield spreads and growth, while transitory shocks (e.g., monetary policy, labor supply) create false signals.
Moving averages help mitigate the impact of these transitory shocks, strengthening the overall predictive power of the spreads.
Practical Takeaways for Traders
Traders should focus on monitoring longer-term forward spreads, such as the 53-month forward rate minus the 1-month rate (FS(53,1,1)) or the 52-month forward rate minus the 1-month rate (FS(52,2,1)).
These spreads have demonstrated unparalleled accuracy in predicting recessions.
To implement these strategies, traders should use the Liu-Wu (2021) yield dataset, which provides accurate forward rate calculations.
Applying a 1-year moving average (250–260 trading days) to these spreads is crucial for reducing false signals.
Unlike ad-hoc windows such as quarterly or monthly averages, the 1-year moving average has been empirically validated as the optimal smoothing period.
The study also proposes recession-predicting indices that aggregate all 83 perfect spreads.
These indices provide a comprehensive measure of recession risk and have shown remarkable accuracy.
As of August 2023, the indices signaled a high probability of a recession by August 2025, with values ranging from 95.5% to 99.9%.
Traders can use these indices as a reliable gauge of upcoming economic downturns.
However, it's important to recognize the limitations of these predictors.
While they have performed flawlessly in historical data, structural shifts such as changes in Federal Reserve policy or market reforms could alter their dynamics.
Additionally, implementing these strategies requires access to high-frequency yield curve data and significant computational resources.
Case Study: 2020 vs. 2023
The 2020 recession provides a compelling case study.
The NTFS, a common spread, failed with a false positive, while the perfect spreads accurately signaled the recession.
This highlights the superiority of the new predictors over traditional indicators.
Looking ahead, the authors' indices (weighted τ=10) show a 95.5%–99.9% probability of a recession by August 2025.
This strong signal suggests that traders should prepare for potential economic turbulence in the coming years.
Conclusion
The era of relying on flawed yield curve spreads is over.
By shifting focus to longer-term forward spreads (4+ years ahead) and applying rigorous moving averages, traders can gain a decisive edge in anticipating recessions.
While no indicator is future-proof, this research provides the most robust framework yet for navigating economic cycles.
For a holistic view, traders should pair these signals with complementary indicators such as the Sahm Rule or unemployment trends.
By combining these tools, traders can build a more comprehensive and accurate recession forecasting system.
For the full analysis and methodology, refer to the original paper: Perfect Recession Predictors.
Keywords: Recession prediction, yield curve, forward spreads, Treasury bonds, trading strategies, economic indicators.