We consider here a way to curtail the advantage that latency arbitrageurs have on public exchanges. Latency arbitrageurs outperform because they are the quickest to exploit mispriced assets as discussed earlier.

For the purpose of our explanation here we repeat our three step model of how information is transferred in financial markets:

  1. New information flows from the information source to the market participants;
  2. The market participants use the information to determine a new fair value;
  3. The market participants trade or adjust their bid or offer prices on the exchange to reflect this new fair value.

Latency arbitrageurs have invested in resources so that they can execute the above three successive steps faster than others.  They compete in the race to the top of the order book to exploit mispriced bids and offers

There will always be a winner in this race, because information is transmitted and processed sequentially, even by computers. As discussed before, it is likely that this race is and will be won consistently by one or a small number of arbitrageurs. An undesirable monopoly could result.

Attempting to prevent this by interfering with the information flow to market participants (step 1), would be impractical. Also, it would be difficult to limit participants’ ability to calculate the new fair value by, say, limiting allowed computer power (step 2). Practically it is only possible to level the playing field to some extent by interfering in step 3. Receiving new information faster (step 1) and calculating the fair value quicker (step 2) would not make any difference if one cannot execute on it timeously.

It is currently standard practice that trading occurs in, virtually, continuous time. So, in any trading period, there are an “infinite” number of races to the top of the order book. The relative advantage that latency arbitrageurs have can, therefore, be curtailed by reducing the number of “races” that can be won.

This can be achieved by changing from continuous time trading to discrete time trading as proposed by Budish et al. (2015) in the paper  “The high-frequency trading arms race: frequent batch auctions as a market design response”.

With batch auctions, trading happens only at discrete time instants as uniform price auctions, according to the following set of rules:

  1. Bids and offers are submitted in between auctions.
  2. Bids and offers are given preference only in relation to price and not on when they were submitted.
  3. Bids and offers need to be sealed, so no participant has an advantage just because they can act quicker in utilising this information.
  4. The market clears at the price where the largest volume of the security trades.

Even though the advantage which latency arbitrageurs enjoy is much reduced with batch auctions, the opportunity for latency arbitrageurs to profit remains. The frequency of the batch auctions will determine the magnitude of their advantage.

We do believe, however, that it is impossible to eliminate their speed advantage entirely, due to the sequential transmission of information, refer to a following post.