Forecasting foreign exchange volatility: Why is implied volatility biased and inefficient? And does it matter?

Research has consistently found that implied volatility is a conditionally biased predictor of realized volatility across asset markets. This paper evaluates explanations for this bias in the market for options on foreign exchange futures. Several recently proposed solutions – including a model of priced volatility risk – fail to explain a significant portion of the conditional bias found in implied volatility. Further, while implied volatility fails to subsume econometric forecasts in encompassing regressions, these forecasts do not significantly improve delta-hedging performance. Thus this paper argues that statistical metrics are inappropriate measures of the information content of implied volatility. Implied volatility appears much more useful when measured by a more relevant, economic metric.     

VOLATILITY FORECASTS,TRADING VOLUME,AND THE ARCH VERSUS OPTION‐IMPLIED VOLATILITY TRADE‐OFF

We investigate empirically the role of trading volume (1) in predicting the relative informativeness of volatility forecasts produced by autoregressive conditional heteroskedasticity (ARCH) models versus the volatility forecasts derived from option prices, and (2) in improving volatility forecasts produced by ARCH and option models and combinations of models. Daily and monthly data are explored. We find that if trading volume was low during period t?1 relative to the recent past, ARCH is at least as important as options for forecasting future stock market volatility. Conversely, if volume was high during period t?1 relative to the recent past, option‐implied volatility is much more important than ARCH for forecasting future volatility. Considering relative trading volume as a proxy for changes in the set of information available to investors, our findings reveal an important switching role for trading volume between a volatility forecast that reflects relatively stale information (the historical ARCH estimate) and the option‐implied forward‐looking estimate.     

Financial crisis,Value-at-Risk forecasts and the puzzle of dependency modeling

Forecasting Value-at-Risk (VaR) for financial portfolios is a crucial task in applied financial risk management. In this paper, we compare VaR forecasts based on different models for return interdependencies: volatility spillover (Engle & Kroner, 1995), dynamic conditional correlations (Engle, 2002, 2009) and (elliptical) copulas (Embrechts et al., 2002). Moreover, competing models for marginal return distributions are applied. In particular, we apply extreme value theory (EVT) models to GARCH-filtered residuals to capture excess returns.Drawing on a sample of daily data covering both calm and turbulent market phases, we analyze portfolios consisting of German Stocks, national indices and FX-rates. VaR forecasts are evaluated using statistical backtesting and Basel II criteria. The extensive empirical application favors the elliptical copula approach combined with extreme value theory (EVT) models for individual returns. 99% VaR forecasts from the EVT-GARCH-copula model clearly outperform estimates from alternative models accounting for dynamic conditional correlations and volatility spillover for all asset classes in times of financial crisis.     

The incremental volatility information in one million foreign exchange quotations

The volatility information found in high-frequency exchange rate quotations and in implied volatilities is compared by estimating ARCH models for DM/$ returns. Reuters quotations are used to calculate five-minute returns and hence hourly and daily estimates of realised volatility that can be included in equations for the conditional variances of hourly and daily returns. The ARCH results show that there is a significant amount of information in five-minute returns that is incremental to options information when estimating hourly variances. The same conclusion is obtained by an out-of-sample comparison of forecasts of hourly realised volatility.     

‘Déjà vol’: Predictive regressions for aggregate stock market volatility using macroeconomic variables

Aggregate stock return volatility is both persistent and countercyclical. This paper tests whether it is possible to improve volatility forecasts at monthly and quarterly horizons by conditioning on additional macroeconomic variables. I find that several variables related to macroeconomic uncertainty, time-varying expected stock returns, and credit conditions Granger cause volatility. It is more difficult to find evidence that forecasts exploiting macroeconomic variables outperform a univariate benchmark out-of-sample. The most successful approaches involve simple combinations of individual forecasts. Predictive power associated with macroeconomic variables appears to concentrate around the onset of recessions.     

FORECASTING VOLATILITY FOR PORTFOLIO SELECTION

The volatility of an asset is a primary input to the portfolio selection problem. Information about volatility is available from two sources, namely the share market and the option market. This paper examines the forecasting performance, over a three month investment horizon, of time series forecasts (from the share market) and option based implied volatilities. Three time series models, including GARCH, are used and twenty four implied volatility estimation models are employed. Using a data set of twelve UK companies, it is demonstrated that implied volatilities produce better individual forecasts than time series. However, more remarkably, forecasts combining implied volatilies and time series estimates significantly outperform both component forecasts.     

Using information quality for volatility model combinations

This paper proposes updated methodology for volatility model combinations which account for the informational content of innovations. An adaptive measure of information quality serves for the selection of model weights in order to improve daily volatility forecasts. The information quality proxy is related to the size of unexpected shocks in the volatility process. Our approach is illustrated in an empirical study with German stock market data.     

Evaluating the Forecast Accuracy of Emerging Market Stock Returns

This paper analyzes the forecast performance of emerging market stock returns using standard autoregressive moving average (ARMA) and more elaborated autoregressive conditional heteroskedasticity (ARCH) models. Our results indicate that the ARMA and ARCH specifications generally outperform random walk models. Models that allow for asymmetric shocks to volatility are better for in-sample estimation (threshold autoregressive conditional heteroskedasticity for daily returns and exponential generalized autoregressive conditional heteroskedasticity for longer periods), and ARMA models are better for out-of-sample forecasts. The results are valid using both U. S. dollar and domestic currencies. Overall, the forecast errors of each Latin American market can be explained by the forecasts of other Latin American markets and Asian markets. The forecast errors of each Asian market can be explained by the forecasts of other Asian markets, but not by Latin American markets. Our predictability results are economically significant and may be useful for portfolio managers to enter or leave the market.     

A note on the importance of overnight information in risk management models

This paper examines the economic value of overnight information to users of risk management models. In addition to the information revealed by overseas markets that trade during the (domestic) overnight period, this paper exploits information generated via recent innovations in the structure of financial markets. In particular, certain securities (and associated derivative products) can now be traded at any time over a 24-h period. As such, it is now possible to make use of information generated by trading, in (almost) identical securities, during the overnight period. Of the securities that are available over such time periods, S&P 500 related products are by far the most actively traded and are, therefore, the subject of this paper. Using a variety of conditional volatility models that allow time-dependent information flow within (and across) three different S&P 500 markets, the results show that overnight information flow has a significant impact on the conditional volatility of daytime traded S&P 500 securities. Moreover (time-consistent) forecasts from models that incorporate overnight information are shown to have economic value to risk managers. In particular, Value-at-Risk (VaR) models based on these conditional volatility models are shown to be more accurate than VaR models that ignore overnight information.     

Intraday dynamics of volatility and duration: Evidence from Chinese stocks

We propose a new joint model of intraday returns and durations to study the dynamics of several Chinese stocks. We include three U.S. stocks for comparison. Flexible innovation distributions are used for durations and returns, and the total variance of returns is decomposed into different volatility components associated with different transaction horizons. The new model provides strong improvements in density forecasts for duration and returns and only modest gains for points forecasts of the variance of returns. The conditional hazard functions are non-monotonic and there is strong evidence for different volatility components. Although diurnal patterns, volatility components, and market microstructure implications are similar across the markets, there are interesting differences. Durations for lightly traded Chinese stocks tend to carry more information than heavily traded stocks. Chinese investors usually have longer investment horizons, which may be explained by the specific trading rules in China.     

Forecasting aggregate stock market volatility with industry volatilities: The role of spillover index

In this paper, we aim to improve the predictability of aggregate stock market volatility with industry volatilities. The empirical results show that individual industry volatilities can provide useful predictive information, while the predictive contribution is limited. We further consider the spillover index between industry volatilities and find it displays strong predictive power for stock market volatility. Based on the portfolio exercise, we find that a mean-variance investor can achieve sizeable economic gains by using volatility forecasts of the spillover index. In addition, we conduct three extended analyses and further demonstrate the superior performance of the spillover index. Also, our results show robustness to a series of alternative settings. Finally, we investigate why the spillover index performs better and answer what information it contains. The results show that the spillover index can reflect and explain investor sentiments that are related to stock market volatility.     

Combining conditional volatility forecasts using neural networks: an application to the EMS exchange rates

The present paper examines the out-of-sample forecasting performance of four conditional volatility models applied to the European Monetary System (EMS) exchange rates. In order to provide improved volatility forecasts, the four models’ forecasts are combined through simple averaging, an ordinary least squares model, and an artificial neural network. The results support the EGARCH specification especially after the foreign exchange crisis of August 1993. The superiority of the EGARCH model is consistent with the nature of the EMS as a managed float regime. The ANN model performed better during the August 1993 crisis especially in terms of root mean absolute prediction error.     

Volatility Forecasting in the Hang Seng Index using the GARCH Approach

The aim of this paper is to add to the literature on volatility forecasting using data from the Hong Kong stock market to determine if forecasts from GARCH based models can outperform simple historical averaging models. Overall, unlike previous studies we find that the GARCH models with non-Normal distributions show a robust volatility forecasting performance in comparison to the historical models. The results indicate that although not all models outperform simple historical averaging, the EGARCH based models, with non-normal conditional volatility, tend to produce more accurate out-of-sample forecasts using both standard measures of forecast accuracy and financial loss functions. In addition we test for asymmetric adjustment in the Hang Seng, finding strong evidence of asymmetries due to the domination of financial and property firms in this market.     

Forecasting Commodity Market Returns Volatility: A Hybrid Ensemble Learning GARCH-LSTM based Approach

This study investigates the advantage of combining the forecasting abilities of multiple generalized autoregressive conditional heteroscedasticity (GARCH)-type models, such as the standard GARCH (GARCH), exponential GARCH (eGARCH), and threshold GARCH (tGARCH) models with advanced deep learning methods to predict the volatility of five important metals (nickel, copper, tin, lead, and gold) in the Indian commodity market. This paper proposes integrating the forecasts of one to three GARCH-type models into an ensemble learning-based hybrid long short-term memory (LSTM) model to forecast commodity price volatility. We further evaluate the forecasting performance of these models for standalone LSTM and GARCH-type models using the root mean squared error, mean absolute error, and mean fundamental percentage error. The results highlight that combining the information from the forecasts of multiple GARCH types into a hybrid LSTM model leads to superior volatility forecasting capability. The SET-LSTM, which represents the model that combines forecasts of the GARCH, eGARCH, and tGARCH into the LSTM hybrid, has shown the best overall results for all metals, barring a few exceptions. Moreover, the equivalence of forecasting accuracy is tested using the Diebold–Mariano and Wilcoxon signed-rank tests.     

Forecasting daily variability of the S&P 100 stock index using historical,realised and implied volatility measurements

The increasing availability of financial market data at intraday frequencies has not only led to the development of improved volatility measurements but has also inspired research into their potential value as an information source for volatility forecasting. In this paper, we explore the forecasting value of historical volatility (extracted from daily return series), of implied volatility (extracted from option pricing data) and of realised volatility (computed as the sum of squared high frequency returns within a day). First, we consider unobserved components (UC-RV) and long memory models for realised volatility which is regarded as an accurate estimator of volatility. The predictive abilities of realised volatility models are compared with those of stochastic volatility (SV) models and generalised autoregressive conditional heteroskedasticity (GARCH) models for daily return series. These historical volatility models are extended to include realised and implied volatility measures as explanatory variables for volatility. The main focus is on forecasting the daily variability of the Standard & Poor's 100 (S&P 100) stock index series for which trading data (tick by tick) of almost 7 years is analysed. The forecast assessment is based on the hypothesis of whether a forecast model is outperformed by alternative models. In particular, we will use superior predictive ability tests to investigate the relative forecast performances of some models. Since volatilities are not observed, realised volatility is taken as a proxy for actual volatility and is used for computing the forecast error. A stationary bootstrap procedure is required for computing the test statistic and its p-value. The empirical results show convincingly that realised volatility models produce far more accurate volatility forecasts compared to models based on daily returns. Long memory models seem to provide the most accurate forecasts.     

Forecasting stock market volatility with various geopolitical risks categories: New evidence from machine learning models

This paper investigates how geopolitical risks influence the prediction performance on the US stock market volatility with machine learning models. Further, it compares the predictive performance of individual and combination forecast methods. With SHAP algorithm, it could identify which factor has a great impact and fully extract the information of geopolitical risks in predicting. Empirical results show that military build-ups and escalation of war have great importance on predicting realized volatility among various geopolitical risks. The research further emphasizes the superior performance of machine learning and forecast combination methods, especially SVR method and trimmed mean combination. In addition, by allocating portfolio according to the machine learning-based volatility forecasts, particularly elastic net and random forest, a mean-variance investor can achieve sizeable financial benefits. Our paper provides substantial implications for political risk management and volatility forecasting.     

Improving Volatility Forecasts Using Market‐Elicited Ambiguity Aversion Information

Distinguishing between risk and uncertainty, this paper proposes a volatility forecasting framework that incorporates asymmetric ambiguity shocks in the (exponential) generalized autoregressive conditional heteroskedasticity‐in‐mean conditional volatility process. Spanning 25 years of daily data and considering the differential role of investors' ambiguity attitudes in the gain and loss domains, our models capture a rich set of information and provide more accurate volatility forecasts both in‐sample and out‐of‐sample when compared to ambiguity‐free or risk‐based counterparts. Ambiguity‐based volatility‐timing trading strategies confirm the economic significance of our proposed framework and indicate that an annualized excess return of 3.2% over the benchmark could be earned from 1995 to 2014.     

Information criteria for GARCH model selection

In this paper, a set of appropriately modified information criteria for selection of models from the AR-GARCH class is derived. It is argued that unmodified or naively modified traditional information criteria cannot be used for order determination in the context of conditionally heteroscedastic models. The models selected using the modified criteria are then used to forecast both the conditional mean and the conditional variance of two high frequency exchange rate series. The analysis indicates that although the use of such model selection methods does lead to significantly improved forecasting accuracies for the conditional variance in some instances, these improvements are by no means universal. The use of these criteria to jointly select conditional mean and conditional variance model orders leads to performance degradation for the conditional mean forecasts compared to models which do not allow for the heteroscedasticity.     

The jump component of S&P 500 volatility and the VIX index

Much research has investigated the differences between option implied volatilities and econometric model-based forecasts. Implied volatility is a market determined forecast, in contrast to model-based forecasts that employ some degree of smoothing of past volatility to generate forecasts. Implied volatility has the potential to reflect information that a model-based forecast could not. This paper considers two issues relating to the informational content of the S&P 500 VIX implied volatility index. First, whether it subsumes information on how historical jump activity contributed to the price volatility, followed by whether the VIX reflects any incremental information pertaining to future jump activity relative to model-based forecasts. It is found that the VIX index both subsumes information relating to past jump contributions to total volatility and reflects incremental information pertaining to future jump activity. This issue has not been examined previously and expands our understanding of how option markets form their volatility forecasts.     

Stock and bond market interactions with level and asymmetry dynamics: An out-of-sample application

We model the dynamic interaction between stock and bond returns using a multivariate model with level effects and asymmetries in conditional volatility. We examine the out-of-sample performance using daily returns on the S&P 500 index and 10 year Treasury bond. We find evidence for significant (cross-) asymmetries in the conditional volatility and level effects in bond returns. The out-of-sample covariance matrix forecasts of the model imply that an investor is willing to pay between 129 and 820 basis points per year for using a dynamic trading strategy instead of a passive strategy.