We first investigate the evolution of opening and closing auctions volumes of US equities along the years. We then report dynamical properties of pre-auction periods: the indicative match price is strongly mean-reverting because the imbalance is; the final auction price reacts to a single auction order placement or cancellation in markedly different ways in the opening and closing auctions when computed conditionally on imbalance improving or worsening events; the indicative price reverts towards the mid price of the regular limit order book but is not especially bound to the spread. 0:56 WSPC/INSTRUCTION FILE auctions 2 and their dependence on special days (month end, quarter end, etc.) and on the capitalization of the assets of US equities. Gu et al. (2010) report that the average density of the auction limit order book on both sides of the final auction price is well fitted by an exponential distribution and compute the persistence fluctuation properties of the order size in the Shenzhen Stock Exchange. More recently, and more in line with our paper, Boussetta et al. (2016) study the French Stock exchange and find that different kind of market participants enter the pre-auction periods at markedly different times, the slow brokers acting first, while high-frequency traders tend to be active nearer the end of auctions. In the same vein, Bellia et al. (2016) show how and when low-latency traders (identified as high frequency traders) add or remove liquidity in the pre-opening auction of the Tokyo Stock Exchange. Lehalle and Laruelle (2018) devote part of a chapter to auctions in a spirit close to ours, in particular regarding typical daily activity patterns. Finally, Challet (2019) shows that in Paris Stock Exchange, the antagonistic effects of accelerating event rate near the auction time and the decrease of the typical indicative price volatility cannot fully explain the observed diffusion properties of the indicative price, which is likely due to strategic behavior. This paper is organized as follows: we first determine the distribution of the volumes of matched orders at both auction ending times. We then find rules of thumb to estimate opening and closing volumes from daily data. Finally, we examine the dynamics of the indicative price, imbalance, and matched volume. We show in particular that the reaction of the final auction price to the placement or cancellation of an order is markedly different in the opening and closing auctions, and that the indicative auction prices are mostly under-diffusive, especially during the pre-closing auction period. Finally, we relate the dynamics of the indicative price to that of the limit order book: although the mid price does attract the indicative price, the latter fluctuates rather much and is not specially bound to be in the spread, which, we argue, is due to the sparseness of the auction order book.
DataEach exchange follows its own auction rules and pre-auction information dissemination procedure. The basic principle of all auctions however is the same: traders ma...