The Volatility Risk Premium Around Macroeconomic Announcements

Markets are typically volatile, and price movements accelerate during macroeconomic announcements. We have discussed the macroeconomic announcement premium and the related beta arbitrage strategy.

Along this line of research, Reference [1] examined the returns of delta-neutral straddles around macroeconomic announcements. By analyzing these returns, one can draw conclusions about the volatility risk premium (VRP) on announcement dates.

The authors pointed out,

The study found negative returns across all windows across all weighting methods for all macroeconomic announcements. Therefore, there is no evidence that purchasing straddles ahead of announcements is profitable. However, there is still an important comparison to be conducted between the data for all days versus the data for macroeconomic announcements. This indicates if the straddles are priced differently on days of macroeconomic announcement. Equal weighting for macroeconomic announcements shows lower returns than for straddles on all days. Value weighting show inconclusive results as sometimes the straddles on macroeconomic announcements outperforms the entire sample, while sometimes they don’t. Volume weighted straddles on macroeconomic announcements outperforms straddles on all days for all windows in this study.

The other individual macroeconomic announcements such as FOMC, CPI and PPI in table 7 show lower returns compared to the equal weighted returns in table 6. This finding suggests that the risk for these options around the macroeconomic indicators could have been overestimated. The returns around individual macroeconomic announcements in this study are only calculated for equal weighted returns, it would be interesting to also use the other weighting methods. Additionally, it would be interesting to add and remove macroeconomic indicators to this analysis as there might be different effects for other events.

In short, straddle returns are negative, which indicates that the VRP persists even in the presence of macroeconomic announcements.

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

References

[1] Carl Windmar, Risk Premium around Macroeconomic Announcements: Evidence from Delta-Neutral Straddles, Lund University, 2025

Article Source Here: The Volatility Risk Premium Around Macroeconomic Announcements

source https://harbourfronts.com/volatility-risk-premium-around-macroeconomic-announcements/

Hedge Effectiveness Under a Four-State Regime Switching Model

Identifying market regimes is important for understanding shifts in risk, return, and volatility across financial assets. With the advancement of machine learning, many regime-switching and machine learning methods have been proposed. However, these methods, while promising, often face challenges of interpretability, overfitting, and a lack of robustness in real-world deployment.

Reference [1] proposed a more “classical” regime identification technique. The authors developed a four-state regime switching (PRS) model for FX hedging. Instead of using a simple constant hedge ratio, they classified the market into regimes and optimized hedge ratios accordingly.

The paper pointed out,

We develop a four-state regime-switching model using forward contracts to hedge foreign exchange positions. The hedging effectiveness results indicate that the PRS model reduces portfolio variance more effectively than other existing hedging strategies in dollar, euro, yen and lira markets. In the rupee market, the model shows the second-best performance. The findings suggest that, constructing the four-state regime-switching hedging with the optimized level of memory produces better results than employing a constant ratio obtained from the entire period. The findings are consistent with prior research that supports the use of a model that can be updated with more recent data over time (Kroner & Sultan, 1993; Myers & Thompson, 1989; Ricci, 2020).

The outperformance of the proposed model against the two other dynamic approaches means that it can capture asymmetry and fat-tail properties, which are frequently observed in FX returns. Importantly, the marked performance improvement in the case of lira suggests that the model might be able to offer more effective hedging for highly volatile currencies. This is because the model automatically adjusts the horizon to estimate the optimal hedge ratio based on the prevailing market conditions.

In short, the authors built a smarter hedging model by splitting markets into four conditions instead of two, adjusting hedge ratios and memory length depending on the volatility regime. This significantly improves hedge effectiveness, especially in volatile currencies.

We believe this is an efficient method that can also be applied to other asset classes, such as equities and cryptocurrencies.

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

References

[1] Taehyun Lee, Ioannis C. Moutzouris, Nikos C. Papapostolou, Mahmoud Fatouh, Foreign exchange hedging using regime-switching models: The case of pound sterling, Int J Fin Econ. 2024;29:4813–4835.

Post Source Here: Hedge Effectiveness Under a Four-State Regime Switching Model

source https://harbourfronts.com/hedge-effectiveness-four-state-regime-switching-model/

Comparative Analysis of Gold Forecasting Models: Statistical vs. Machine Learning Approaches

Gold is an important asset class, serving as both a store of value and a hedge against inflation and market uncertainty. Therefore, performing predictive analysis of gold prices is essential. Reference [1] evaluated several predictive methods for gold prices. It examined not only classical, statistical approaches but also newer machine learning techniques. The study used data from 2021 to 2025, with 80% as in-sample data and 20% as validation data.

The authors pointed out,

The results highlight several key findings. First, descriptive and diagnostic analysis confirmed that gold remains a moderately volatile but consistently appreciating asset, with historical returns of 85% over the period. Among the forecasting models, Linear Regression and ETS outperformed ARIMA, KNN, and SVM, achieving the lowest error rates (RMSE 35.7) and the highest explanatory power (R² 0.986). Contrary to common assumptions, machine learning models such as KNN and SVM failed to surpass traditional statistical approaches, underlining the importance of model interpretability and stability in volatile markets. Forecasts for 2026 indicate a projected average price of $4,659, representing a potential 58.6% return, though results also highlight the necessity of caution given market uncertainties. Collectively, the findings demonstrate that simpler models can often provide more reliable forecasts than complex algorithms when applied to financial time series.

In short, the paper shows that,

• Linear Regression and ETS outperformed ARIMA, KNN, and SVM, delivering the lowest error and highest explanatory power,
• Machine learning models (KNN, SVM) did not outperform traditional statistical methods, emphasizing the value of interpretability and stability in volatile markets.

Overall, simpler models often provide more reliable forecasts than complex algorithms for gold time series.

Another notable aspect of the study is its autocorrelation analysis, which reveals that, unlike equities, gold does not exhibit clear autocorrelation patterns—its price behavior appears almost random. The paper also suggested improving the forecasting model by incorporating macroeconomic variables.

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

References

[1] Muhammad Ahmad, Shehzad Khan, Rana Waseem Ahmad, Ahmed Abdul Rehman, Roidar khan, Comparative analysis of statistical and machine learning models for gold price prediction, Journal of Media Horizons, Volume 6, Issue 4, 2025

Post Source Here: Comparative Analysis of Gold Forecasting Models: Statistical vs. Machine Learning Approaches

source https://harbourfronts.com/comparative-analysis-gold-forecasting-models-statistical-vs-machine-learning-approaches/

Volatility, Correlations, and Causal Links in Cryptocurrency Markets

Analyzing volatilities, correlations, and lead–lag relationships across financial assets is important for portfolio and risk management. As cryptocurrencies gain traction, research in this area is growing. Reference [1] studies the causal relationships, volatility, and correlations among major cryptocurrencies and the Crypto Volatility Index (CVI).

A distinctive aspect of this work is that, unlike prior time-series studies, it uses time–frequency domain methods, specifically wavelet analysis, to examine the relationships. The paper analyzes volatility as a driver of interlinkages between cryptocurrencies and the CVI across different investment horizons. It employs daily spot prices of Bitcoin, Ethereum, Tether, USD-Coin, BNB, and the CVI from April 2019 to August 2022.

The authors pointed out,

It can be inferred from their estimates that the CVI exhibits significant short-term, medium-term and long-term interlinkages with Bitcoin, Ethereum, Tether, USD-Coin and BNB. The time-frequency analysis using partial wavelet coherence depicts that CVI does not drive the perseverance of high correlations between cryptocurrencies. Specifically, there is no change in the degree of interlinkages between Bitcoin and other cryptocurrencies, Ethereum and other cryptocurrencies, Tether and other cryptocurrencies, USD-Coin and other cryptocurrencies, or BNB and other cryptocurrencies when CVI is included as a covariate. These results depict the prevalence of idiosyncratic shocks over the common interlinkages within individual cryptocurrency markets. As a result, CVI may not generally serve as a hedging proxy for these cryptocurrencies when included in the same portfolio. The estimates of wavelet bivariate correlations highlight that Bitcoin and Ethereum exhibit the highest degrees of co-movements across various time scales, followed by Bitcoin and BNB, and Ethereum and BNB. The co-movements in these pairs are positive from intraweek scales to quarterly scales, representing short-term and medium-term real economic transactions. These positive movements make diversification impracticable between these pairs, as they tend to exhibit behavior associated with speculative bubbles. The estimates from multiple wavelet correlations confirm that the degree of interlinkages is relatively high in the long term, with undulations or surges over the time horizon. This accentuates a higher convergence of cryptocurrencies’ returns in the long term, even in the presence of CVI. The wavelet multiple cross-correlations coefficients depict Ethereum as the most influential variable in the short term, Bitcoin as the most influential variable in the short to medium term, CVI as the most influential variable in the monthly to quarterly scale (i.e., in the medium term). The estimates also confirm that Ethereum leads or lags at the intraweek and weekly scales, Bitcoin leads at the fortnightly and monthly scales, and CVI lags at the monthly to quarterly scale.

In short, the study finds that CVI shows short-, medium-, and long-term interlinkages with major cryptocurrencies such as Bitcoin, Ethereum, Tether, USD-Coin, and BNB, but it does not drive high correlations across them. Wavelet correlation estimates reveal strong co-movements, especially between Bitcoin–Ethereum, Bitcoin–BNB, and Ethereum–BNB, which remain positive across short- to medium-term horizons. Long-term interlinkages remain high, with Ethereum most influential in the short term, Bitcoin in the short to medium term, and CVI in the medium term, confirming varying lead–lag roles across time scales.

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

References

[1] Vandana Dangi, Cryptocurrency Implied Volatility as a Driver of the Interlinkages Across Cryptocurrencies’ Returns: A Wavelet Analysis, VIKALPA The Journal for Decision Makers 1 –29, 2025

Originally Published Here: Volatility, Correlations, and Causal Links in Cryptocurrency Markets

source https://harbourfronts.com/volatility-correlations-causal-links-cryptocurrency-markets/

Measuring Concentration Risk in Equity Markets Through HHI and Gini Metrics

Concentration risk arises when too much exposure is tied to a single firm, sector, or asset. If one company, or a handful of large players, becomes too big in a portfolio or market, their failure can drive outsized losses. In essence, concentration risk is the danger of being undiversified: the success or failure of just one exposure can disproportionately affect overall performance.

In the banking sector, credit portfolio managers often use indicators such as the Herfindahl–Hirschman Index (HHI) or the Gini coefficient to measure and manage concentration risk. Reference [1] extends this approach to equities, examining how equity index concentration affects stock market performance in the S&P 500, NASDAQ 100, and STOXX Europe 600. The authors pointed out,

By extending the research to the factor analysis using the Carhart four-factor model, we can confirm that industry-level concentration also negatively affects alpha and predicts elevated idiosyncratic volatility beyond systematic factors. We also conclude that the excess returns of the S&P 500 are affected by firm-level concentration through factor exposures, and possibly temporary overvaluation which reverses in the long-term. Moreover, the finding of short-term positive effect on returns of the NASDAQ 100 on the firm-level is explained by factor exposures, and the long-term effect could be explained by the relatively different index composition consisting of tech giants. While the STOXX 600 exhibits immediate negative effects on alpha, this could potentially be explained by the relatively smaller role of passive investment in Europe, differing investor sentiment, and the larger and more diversified nature of the index. Additionally, we find that while some increased volatility especially in the U.S. indices could be explained by factor exposures, concentration increases idiosyncratic volatility beyond systematic factors, possibly due to temporary overvaluations, increased price pressure resulting from passive flows, and lower effective number of stocks.

In short, the study finds that industry concentration leads to lower alpha and higher idiosyncratic volatility beyond systematic factors.

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

References

[1] Timi Oksanen, Niko Ollila, Concentration in Equity Indices: Implications for Returns and Volatility, Aalto University, 2025

Post Source Here: Measuring Concentration Risk in Equity Markets Through HHI and Gini Metrics

source https://harbourfronts.com/measuring-concentration-risk-equity-markets-hhi-gini-metrics/

Volatility of Volatility as a Risk Factor in Crypto Options

As cryptocurrencies become mainstream, liquidity in crypto derivatives such as perpetual futures and options is improving. They also attract more attention from researchers. Reference [1] contributes to the growing literature on crypto derivatives by studying Bitcoin options volatility. Specifically, it shows that BTC options share both the “usual” and “unusual” volatility components found in traditional finance.

• The usual component reflects volatility evolving gradually in response to macroeconomic developments or regulatory changes.
• The unusual component represents sharp increases in volatility, highlighting heightened volatility of volatility (VOV).

These two characteristics must be incorporated into an options pricing model. The paper proposed a so-called affine realized volatility of volatility (ARVOV) model, which explicitly accounts for VOV risk by introducing a distinct latent variable to capture its dynamics. This latent factor serves as a critical risk driver in the return-generating process and strongly influences volatility behavior.

The authors pointed out,

This paper introduces a novel option pricing framework, the ARVOV model, that explicitly separates the dynamics of volatility and VOV. Distinct from conventional option pricing models, our model treats VOV as an independent source of risk, reflecting uncertainty about future volatility itself, rather than as a mere extension or function of volatility. By incorporating realized VOV measures into the modeling of conditional variance, the ARVOV model offers a more flexible and precise characterization of volatility dynamics in cryptocurrency markets. Utilizing the exponential affine structure of the MGF, we derive a closed‐form European option pricing formula through Fourier inversion. Empirical analyses are conducted using high‐frequency Bitcoin price data and historical Bitcoin options data sourced from Deribit.

Model performance is rigorously evaluated by assessing the RMSE between model‐filtered volatilities and their corresponding realized measures. Additionally, we evaluate option pricing accuracy via the RMSE of implied volatility, defined as the discrepancy between market‐based and model‐based implied volatilities. Empirical results indicate that our ARVOV model reduces implied volatility RMSE by 8.55% compared to the second‐best benchmark. Further robustness tests across various moneyness categories, maturities, and market volatility levels consistently demonstrate the superior performance of the ARVOV model, particularly under conditions of extreme moneyness, shortest and longest maturities, and heightened market volatility.

In short, the study highlights the unique nature of BTC volatility and develops a pricing model that explicitly incorporates VOV risk.

An interesting result is that explicitly modeling VOV allows the variance risk premium (VRP) to take time-varying signs, unlike the predominantly positive VRP found in other studies. This significantly improves the ability to capture the complex behavior of VRP and option prices in highly volatile cryptocurrency markets.

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

References

[1] Lingshan Du, Ji Shen, Pricing Cryptocurrency Options With Volatility of Volatility, Journal of Futures Markets, 2025; 1–26

Originally Published Here: Volatility of Volatility as a Risk Factor in Crypto Options

source https://harbourfronts.com/volatility-volatility-risk-factor-crypto-options/

Integrating Structured and Unstructured Data with LLMs and RAG

Traditional quantitative methods often rely on structured data, such as time series. With the emergence of Large Language Models (LLMs), it is now possible to process unstructured data. A new line of research focuses on integrating unstructured data analysis into traditional frameworks.

Along this line, Reference [1] proposed the use of LLMs together with retrieval-augmented generation (RAG) to process both structured and unstructured data concurrently. Specifically, the authors developed a system that first applies LLMs to detect regime shifts using time-series techniques, then employs RAG to integrate external knowledge into the model’s decision-making process. By retrieving relevant information from a vector database and combining it with the model’s capabilities, RAG improves both the interpretability and effectiveness of trading strategies.

The authors pointed out,

This study demonstrates the integration of fine-tuned LLMs with RAG for adaptive trading strategies and portfolio management. By combining numerical time series data and textual insights from news and macroeconomic indicators, the proposed framework addresses the challenges of multimodal data integration in financial markets.

The experimental results highlight the value of fine-tuning smaller LLMs, such as GPT-4o Mini, which improves regime shift detection and trading decision accuracy while maintaining computational efficiency. The application of SAX enhances the compatibility of time series data with LLMs, while the CoT framework ensures transparency and robustness in decision-making. This proof-of-concept establishes a solid foundation for integrating advanced LLMs in quantitative finance.

In short, incorporating RAG into the framework enhances the model’s ability to understand complex macroeconomic environments and adapt trading strategies as conditions evolve. Experimental results show significant gains in predictive accuracy and risk-adjusted returns, demonstrating the practical value of these fine-tuning methods in finance.

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

References

[1] Li, C., Chan, C.H.R., Huang, S.H., Choi, P.M.S. (2025). Integrating LLM-Based Time Series and Regime Detection with RAG for Adaptive Trading Strategies and Portfolio Management. In: Choi, P.M.S., Huang, S.H. (eds) Finance and Large Language Models. Blockchain Technologies. Springer, Singapore.

Post Source Here: Integrating Structured and Unstructured Data with LLMs and RAG

source https://harbourfronts.com/integrating-structured-unstructured-data-llms-rag/