To Cross or Not to Cross the Spread: That Is the Question

Abstract

Traders have always empirically estimated the short-term dynamic of the market. Contrary to popular belief, building a quantitative estimate of the next trade is not some sort of “Holy Grail” only accessible to the darkest high-frequency traders. In fact, this knowledge is easily understandable by anyone who observes the data attentively. Thanks to our tick database, we find that order book imbalances allow us to make reliable forecasts regarding the side (bid/ask) of the next trade; in short, on average, aggressive trades will mainly target the smallest limit (in size). We also find that aggressive trades will consume a larger share when the smaller limit is hit rather than the larger limit. Using this insight regarding the next trade enables us to estimate the risk induced by a passive posting and thereby helps to decide whether or not to cross the spread using objective criteria based on the Sharpe ratio. Similar rules can be applied to trading algorithms to help eliminate many unnecessary aggressive trades and thus significantly increase trading performances.