This study introduces a time-varying parameter vector autoregressive (TVP-VAR) model to enhance dynamic connectedness measures originally developed by Diebold and Yilmaz (2012, 2014). The TVP-VAR model incorporates a time-varying variance-covariance structure, allowing for more flexible and robust estimation of dynamic connectedness. Unlike traditional rolling-window VAR models, the TVP-VAR approach does not require arbitrary window sizes or result in observation loss, and it is less sensitive to outliers. The study also emphasizes the merits of this approach through Monte Carlo simulations, demonstrating that TVP-VAR estimates are more accurate than rolling-window estimates in terms of parameter stability and sensitivity to outliers. The authors apply the TVP-VAR model to analyze dynamic connectedness among four major foreign exchange rates: USD/EUR, USD/GBP, USD/CHF, and USD/JPY. They compare the results with those from three different rolling-window settings and find that the TVP-VAR model provides more immediate and accurate adjustments to underlying events, while rolling-window estimates either overreact or smooth out effects depending on window size. The study also proposes uncertainty measures for both TVP-VAR and rolling-window VAR-based dynamic connectedness measures. The TVP-VAR model is shown to outperform rolling-window VAR models in terms of forecast accuracy, particularly for longer forecast horizons. The study also conducts a prior sensitivity analysis, allowing for different prior assumptions, and finds that results converge to similar outcomes after a certain number of parameter updates. Additionally, the study provides confidence intervals for rolling-window VAR connectedness measures, highlighting the significance of the results. The findings indicate that the EUR and CHF are significant net transmitters of shocks, while the GBP and JPY are significant net receivers. The EUR and CHF also dominate the GBP and JPY in terms of shock transmission. The study concludes that the TVP-VAR model provides a more accurate and flexible approach to measuring dynamic connectedness compared to traditional rolling-window methods.This study introduces a time-varying parameter vector autoregressive (TVP-VAR) model to enhance dynamic connectedness measures originally developed by Diebold and Yilmaz (2012, 2014). The TVP-VAR model incorporates a time-varying variance-covariance structure, allowing for more flexible and robust estimation of dynamic connectedness. Unlike traditional rolling-window VAR models, the TVP-VAR approach does not require arbitrary window sizes or result in observation loss, and it is less sensitive to outliers. The study also emphasizes the merits of this approach through Monte Carlo simulations, demonstrating that TVP-VAR estimates are more accurate than rolling-window estimates in terms of parameter stability and sensitivity to outliers. The authors apply the TVP-VAR model to analyze dynamic connectedness among four major foreign exchange rates: USD/EUR, USD/GBP, USD/CHF, and USD/JPY. They compare the results with those from three different rolling-window settings and find that the TVP-VAR model provides more immediate and accurate adjustments to underlying events, while rolling-window estimates either overreact or smooth out effects depending on window size. The study also proposes uncertainty measures for both TVP-VAR and rolling-window VAR-based dynamic connectedness measures. The TVP-VAR model is shown to outperform rolling-window VAR models in terms of forecast accuracy, particularly for longer forecast horizons. The study also conducts a prior sensitivity analysis, allowing for different prior assumptions, and finds that results converge to similar outcomes after a certain number of parameter updates. Additionally, the study provides confidence intervals for rolling-window VAR connectedness measures, highlighting the significance of the results. The findings indicate that the EUR and CHF are significant net transmitters of shocks, while the GBP and JPY are significant net receivers. The EUR and CHF also dominate the GBP and JPY in terms of shock transmission. The study concludes that the TVP-VAR model provides a more accurate and flexible approach to measuring dynamic connectedness compared to traditional rolling-window methods.