We examine whether low-volatility factor investing can benefit from using time series models that employ intraday data.
Low-volatility factor investing requires an investor to identify the next period’s low-volatility stocks. Existing approaches assume that each stock’s volatility follows a random walk based on daily return data. This assumption does not concur with current econometric literature that documents the benefits of employing intraday data for predicting future volatility. Therefore, we ask whether recent econometric volatility models are useful for low-volatility investing. By constructing model-based forecasts for all S&P 500 stocks, we find that econometric models are superior to the random walk in identifying low-volatility stocks, also lead to more stable portfolios with less rebalancing and, thus, to higher returns after transaction costs.