One of my current interest in microstructure research is to understand what drives intraday volatility in stock prices. Why? From an economic point of view, it is not so obvious why intraday volatility matters beside relating intraday volatility to information arrival. But from a practical point of view, it matters. Take for example a transition portfolio manager that has for objective to lower its transaction cost. If price volatility is high, we then expect a volatile bid-ask spread and market depth, which in turn increase transaction cost. For this blog post, I simply want to highlight a new research by Cont et al. (2014) in the Journal of Financial Econometrics that attempts to measure intraday volatility in a very clever way. But first, let me introduce you to the VPIN measure developed by Easley, Lopez de Prado and O’hara to forecast intraday volatility.
The recent academic article by Easley, Lopez de Prado and O’hara (2013) show that market order imbalance can anticipate moments of intraday turbulence in stock prices if we “deform” calendar time and compute the market order imbalance in volume time. By deforming calendar time, it means to move away from regular time tick like one minute or second time interval. Easley et al. (2012) suggest a volume-clock where the time interval length will be a function of the trading volume activity. Let say we start our trading day at 9:30am and there is a high amount of volume at the opening of the market then we can set our first time interval from 9:30:00 to 9:30:44 and the next one from 9:30:44 to 9:30:54 if volume increases. If the amount of trading volume decreases from our last interval, then the end of the next interval may be at 9:33am and hence longer. For each of these intervals we compute the market order imbalance. The authors then suggest to sum the absolute value of the market order imbalance over multiple intervals. Higher is the sum of order imbalance, higher the likelihood of high market volatility. I have skipped a lot of details on the measure but you get the essence.
Following the series of paper on the VPIN by Easley et al., Bondarenko and Bollerslev (2013) published a paper in the Journal of Financial Markets rebutting the ability of the VPIN to anticipate market volatility. I won’t go much in the details, but it is worth the read. The debate between Easley et al. and Bodarenko and Bollerslev continues here and here.
Then this brings the work of Cont et al. (2014). The author suggest a very clever way to “potentially” forecast intraday volatility. They don’t forecast volatility per se but show that there is a high correlation between their measure of order flow imbalance and market volatility (or price changes). What I like about their measure is that they combine both market order imbalance and liquidity provision imbalance. It is computed as follow:
where , , and stands for a limit order, cancel of an order, and market order execution of a order on the buy or sell side of the book.
As you can see, they incorporate both the the imbalance of the market and limit side. They argue that when the limit order imbalance is high, the market order imbalance impact on stock prices becomes very noisy. How liquidity is provided in markets matters to forecast and anticipate market intraday volatility. It may actually be more important than market order imbalance simply because it is liquidity providers that set prices afterall.
What is yet to be done is to use their measure to forecast volatility both in calendar and volume time… which I hope to take the time and test it out.