Integrating Fundamental Metrics into Pairs Trading

Statistical arbitrage is a classic quantitative trading strategy. Its core premise relies on the assumption that two similar stocks should move broadly in tandem and, when their prices diverge, are expected to converge over time. To identify similar underlyings, statistical methods such as cointegration and correlation are commonly employed.

It is widely recognized by practitioners that similarity between two stocks is not purely statistical but also fundamental. However, relatively little research has focused on systematically incorporating fundamental similarity into pair selection. Reference [1] addresses this gap by developing a pairs trading selection framework that integrates fundamental characteristics alongside statistical measures.

To identify fundamentally similar pairs, the author utilizes variables such as return on equity, differences in log sales growth over the past five years, differences in the most recently available financial leverage ratios, geographic proximity (e.g., whether firms are headquartered in the same U.S. state), and industry alignment. Each fundamental metric is then assigned a score, and a composite indicator is constructed for pair selection.

The author pointed out,

This paper introduces a novel approach to stock pair selection in pairs trading, using a multi-factor scoring model based on significant variables identified through panel regression. These variables are weighted by their regression coefficients to construct a composite score. I benchmark this inter-pair characteristic-based method against the Gatev et al. (2006) SSD approach, analyzing both extended and updated trading strategies to account for repeated pair selections. The strategy’s robustness is tested with and without the COVID-19 crisis, and all results are reported net of transaction costs.

The key finding is that the multi-factor method consistently outperforms the benchmark with higher risk-adjusted returns. This performance edge persists beyond the COVID-19 period, with a narrower gap, highlighting the scoring model’s advantage in stressful markets. Though differences between updating and extending pairs are modest, the updated strategy delivers superior returns in volatile markets despite higher transaction costs. Nonetheless, the strategy underperforms relative to passive index investing. Even with rolling estimation windows, out-of-sample performance remains challenging.

In short, this paper introduces a novel stock pair selection framework that incorporates fundamental metrics. The results show that the proposed method delivers higher risk-adjusted returns than the benchmark, particularly in volatile markets, although it underperforms passive index investing and faces challenges in out-of-sample performance.

This article represents a meaningful step toward refining statistical arbitrage by incorporating fundamental characteristics into the pair selection process.

Let us know what you think in the comments below or in the discussion forum.

References

[1]  Lukas Reichmann, Pairs trading- Selection via scoring systems, Finance Research Letters Volume 93, March 2026, 109642

Post Source Here: Integrating Fundamental Metrics into Pairs Trading

source https://harbourfronts.com/integrating-fundamental-metrics-pairs-trading/

Extreme VIX: Regime Shifts and Return Predictability

The Volatility Index (VIX) is widely regarded as a forward-looking measure of market uncertainty and investor sentiment. Although it has been extensively studied, certain aspects remain insufficiently explored.

Reference [1] investigates the implications of extreme VIX values, defined as VIX > 45, and examines two key research questions in this context.

1. Extreme implied volatility (VIX > 45) is followed by significantly positive equity returns over three-month and one-year horizons, and
2. Negative returns, elevated volatility, and deteriorating sentiment increase the probability of a VIX spike (VIX > 45)

Using U.S. data from 2008 to 2025, the authors employ linear regression, logistic regression, and GARCH(1,1) models to conduct the analysis. They pointed out,

Results show that extreme VIX spikes offer contrarian signals, with significant positive returns over three-month horizons. Logistic regression confirms significance over both three-month and one-year periods. These findings remain robust after controlling for valuation, credit spreads, PMI, sentiment ratios, and interaction effects. While GARCH captures conditional variance, it lacks forward-looking predictive power in high-stress regimes. Overall, the evidence suggests that volatility timing merits consideration as part of a tactical allocation framework. Our findings also contribute to the market efficiency debate by integrating market expectations with economic indicators for risk-aware decision-making.

In short, the paper's conclusions are as follows,

• The evidence robustly supports the contrarian effect at the 3-month horizon. At the 1-year horizon, the predictive power comes primarily from continuous volatility measures rather than the binary VIX>45 indicator.
• The results clearly demonstrate that extreme volatility episodes are statistically predictable using valuation, sentiment, and medium-term volatility measures. The evidence shows regime transitions are preceded by identifiable market conditions rather than occurring randomly.

This study makes an important contribution by formally examining phenomena long observed by practitioners and by providing results with meaningful practical implications.

Let us know what you think in the comments below or in the discussion forum.

References

[1] Fazio, P., and Spohn, D. (2026). Predictive Signals from VIX Spikes: A Comparative Study of Linear, Logistic, and GARCH-Based Return Forecasting Models. Preprints.org.

Article Source Here: Extreme VIX: Regime Shifts and Return Predictability

source https://harbourfronts.com/extreme-vix-regime-shifts-return-predictability/

Volume Effects in Pairs Trading Performance

Volume is an important factor that has not been sufficiently studied in the literature, although increasing attention is now being devoted to its role. For example, we recently discussed the link between intraday volume and volatility.

Continuing this line of inquiry, Reference [1] revisits the age-old quantitative strategy of pairs trading and incorporates volume analysis into the framework. The author studied pairs formed from the S&P500 constituent stocks using the conitegration method. The data spans from 2005 to 2024. They pointed out,

This study investigated pairs trading strategies across market segments, parameters, weighting methods, and cost structures using U.S. equity data (2005-2024). Volume emerged as the dominant performance factor, with high-volume pairs consistently outperforming low-volume counterparts across multiple metrics, including superior returns when converging towards the historical relationship. The success of volume-based segmentation may stem partly from grouping stocks with similar trading characteristics, creating more stable price relationships. Notably, optimal parameter configurations were remarkably similar between volume deciles, demonstrating consistent underlying market dynamics across liquidity environments.

Contrary to efficient market predictions, performance improved in the testing period (2015-2024), suggesting structural market changes or institutional constraints have preserved the strategy’s effectiveness despite widespread documentation. Different weighting methodologies produced similar results, with equal weighting combining simplicity and strong performance.

In short, the paper finds that trading volume is a dominant performance driver, with high-volume pairs consistently outperforming low-volume counterparts. Contrary to efficient market predictions, strategy performance improved in the later sample period, while results remained robust across parameter choices and weighting schemes. Overall, pairs trading still remains profitable.

This represents an important contribution that may reinvigorate research on pairs trading and trading volume.

Let us know what you think in the comments below or in the discussion forum.

References

[1] Mustafa Hussein, Can Safety and Profitability Coexist? Performance Analysis of Pairs Trading among S&P 500 Stocks, 2025, University of Gothenburg

Post Source Here: Volume Effects in Pairs Trading Performance

source https://harbourfronts.com/volume-effects-pairs-trading-performance/

State-Dependent Correlation Between the S&P 500 and the VIX

It is well known that the correlation between the S&P 500 and the VIX index is negative. In fact, arbitrage and hedging strategies have been designed around this relationship, for example, hedging vega risks with delta.

However, practitioners also recognize that this correlation can break down and even turn positive. Reference [1] addresses this issue by examining the time-varying nature of the correlation. The paper first posits and demonstrates that volatility-of-volatility (VoV) risk is distinct from volatility (Vol) risk, and then classifies the market into four regimes:

1. Low VOL risk and low VOV risk (LVOL–LVOV)
2. High VOL risk and low VOV risk (HVOL–LVOV)
3. Low VOL risk and high VOV risk (LVOL–HVOV)
4. High VOL risk and high VOV risk (HVOL–HVOV).

The authors pointed out,

We develop theoretical hypotheses on the correlations between the S&P 500 and VIX based on the VFE and two types of risk in the S&P 500–VIX pair: (1) VOL risk in the S&P 500 (i.e., the underlying asset) and (2) VOV risk in VIX (i.e., its VOL product). According to our hypothesis, the correlations between the S&P 500 and VIX do not remain constant and change across various VOL and VOV risk states. We developed a regime‐switching model with state‐contingent correlations to test our hypotheses.

Our empirical results are consistent with the following notions. First, shocks that drive high VOL risk are not necessarily identical to those that drive VOV risk and thus VOL and VOV do not respond to shocks in the same way. These results imply a four‐state VOL system: high/low VOL risk and high/low VOV risk pairs. Second, the minimum inverse correlation between the S&P 500 and VIX is observed under a high VOL–low VOV risk state. In contrast, the maximum inverse correlation between the S&P 500 and VIX occurs under a high VOL–high VOV risk state. Third, the proposed state‐contingent correlations prove more effective in portfolio construction than conventional time‐dependent correlations.

In short, the paper develops a regime-switching model to show that the relationship between the S&P 500 and the VIX varies across distinct volatility (VOL) and volatility-of-volatility (VOV) risk states. It identifies a four-state system and demonstrates that correlation strength depends on the joint VOL–VOV regime.

This is an important contribution, as it quantifies the dynamic relationship between the S&P 500 and the VIX, thereby providing useful guidance for refining hedging and arbitrage strategies.

Let us know what you think in the comments below or in the discussion forum.

References

[1] Leon Li, Carl R. Chen, Volatility Risk and Volatility-of-Volatility Risk: State-Dependent Correlations Between VIX and the S&P 500 Stock Index and Hedging Effectiveness, Journal of Futures Markets, 2025; 1–20

Post Source Here: State-Dependent Correlation Between the S&P 500 and the VIX

source https://harbourfronts.com/state-dependent-correlation-between-the-sp-500-and-the-vix/

Multifractality and Its Underlying Drivers in Cryptocurrency Markets

Cryptocurrencies, like other financial time series, can be analyzed using traditional time series and econometric methods. However, they present additional challenges due to their distinctive characteristics, including extreme volatility, heavy-tailed distributions, and long-range temporal correlations, which [glossary_exclude]warrant [/glossary_exclude]specialized examination.

Reference [1] analyzes the multifractality characteristics of major cryptocurrencies and investigates their underlying sources.  The authors analyzed data from January 1, 2018, to December 31, 2024, and pointed out,

Taken together, these results reinforce the importance of applying multifractal frameworks to digital markets. They offer practical insights for volatility forecasting, systemic risk monitoring, and understanding the maturation of blockchain-based financial ecosystems. For example, the strong correlations between BTC and ETH on various time scales can be used in optimal portfolio construction [80]. The decay of the autocorrelation function, which indicates the average size of the volatility cluster [37], may also be used in risk management. The consistency of the log-return tail distributions with the inverse cubic power-law allows for the selection of appropriate distributions in various risk mitigation methods, such as Value-at-Risk. Moreover, the greater hierarchical dependence of correlations at the level of larger fluctuations compared to smaller ones, documented by left-sided asymmetry of the multifractal spectra, may potentially allow for extending risk management strategies to include fluctuation-specific correlation patterns. The direct link between multifractality and risk management is an interesting subject for further studies.

Briefly, the article finds that digital asset markets, including BTC, ETH, DEX trading, and NFTs, exhibit clear multifractal scaling, but this multifractality is driven primarily by temporal correlations, not by heavy-tailed return distributions alone.  It also shows strong multifractal cross-correlations between BTC and ETH, especially during large fluctuations. Overall, the study argues that genuine market complexity in crypto stems from persistent correlation structure rather than from fat tails alone.

This study provides important insights into cryptocurrency dynamics, enabling portfolio and risk managers to design more appropriate trading and risk management strategies.

Let us know what you think in the comments below or in the discussion forum.

References

[1] Drozdz, S.; Kluszczynski. R., Kwapien. J.: Watorek. M., Multifractality and its sources in the digital currency market. Future Internet, 2022, 1, 0.

Article Source Here: Multifractality and Its Underlying Drivers in Cryptocurrency Markets

source https://harbourfronts.com/multifractality-underlying-drivers-cryptocurrency-markets/

Improving Momentum Strategies with Machine Learning

Machine learning (ML) is increasingly prominent in modern finance and is being adopted across a wide range of applications. However, using ML to extract alpha remains nontrivial. Reference [1] proposes a novel approach to improving the risk-adjusted returns of momentum strategies using machine learning.

It employs three ML methods—linear regression, random forest, and neural networks—to predict next month’s information coefficients (ICs) using information available at the end of the previous month. These predicted ICs are then combined with momentum factors to construct momentum and reversal strategies.

The authors pointed out,

Machine learning helps improve momentum strategies by capturing strong correlations between stocks and their returns using information coefficients (IC). Shallow machine learning enables us to trace the sources of advantage and adapt to non-linear correlations that traditional methods might overlook. We confirm that a unidirectional momentum strategy can yield significant returns. Furthermore, we observe that implementing a bidirectional approach leads to even more pronounced gains. We find that shorting weak stocks outperforms longing strong stocks, indicating the persistence of the "winners keep winning, losers keep losing" phenomenon. Additionally, we observe that price reversals do not occur within a month. Optimizing the number of stocks in the portfolio leads to the highest excess returns within a certain threshold range. However, implementing inverse trades using machine learning models does not yield positive returns, indicating that the predictive ability of the models may be inadequate to identify profitable opportunities in these situations. Finally, we discover that momentum factors with longer holding periods are more powerful predictors, accurately capturing the correlation between stocks and their returns. We assume that momentum factors with shorter observation periods are less effective, possibly due to the effect of mean reversion. Our research underscores the importance of accurately predicting IC values and the effectiveness of momentum strategies in generating positive returns.

In short, the paper applies machine learning models to momentum investing by predicting future information coefficients and using them to construct momentum and reversal portfolios. It finds that bidirectional momentum strategies outperform unidirectional ones, longer-horizon momentum signals are more informative, and portfolio performance is maximized when the number of selected stocks is optimized within a threshold.

This represents an interesting twist on applying machine learning to investing, using predicted information coefficients to guide portfolio construction rather than directly forecasting returns.

Let us know what you think in the comments below or in the discussion forum.

References

[1] Chang, H.-L., Cheng, H.-W., Lan, Y.-M., & Yu, J.-P. (2025). Machine Learning in Momentum Strategies. EFMA

Post Source Here: Improving Momentum Strategies with Machine Learning

source https://harbourfronts.com/improving-momentum-strategies-machine-learning/