Extract from a paper by Eric Budish, Peter Cramton, and John Shim proposing that frequent batch auctions — uniform-price double auctions conducted at frequent but discrete time intervals (e.g. every 1 second) – be employed to end the high-frequency trading arms race:
The fourth and final part of our argument shows that frequent batch auctions are an attractive market design response to the HFT arms race. Batching eliminates the arms race for two reasons. First, and most centrally, batching substantially reduces the value of a tiny speed advantage. In our model, if the batching interval is τ , then a δ speed advantage is only δ/τ as valuable as it is under continuous markets. So, for example, if the batching interval is 1 second, a 1 millisecond speed advantage is only 1/1000 as valuable as it is in the continuous limit order book market design. Second, and more subtly, batching changes the nature of competition among fast traders, encouraging competition on price instead of speed.
Intuitively, in the continuous limit order book market design, it is possible to earn a rent based on a piece of information that many fast traders observe at basically the same time – e.g., a jump in the price of ES – because continuous limit order books process orders in serial, and somebody is always first. In the batch market, by contrast, if multiple traders observe the same information at the same time, they are forced to compete on price instead of speed.