Passing The Latency Arbitrage Test

Miroslav Budimir, senior vice president, Deutsche Börse, offers an exchange operator’s view on ‘Flash Boys’ when speaking with The Trade this month.

Accusing high-frequency trading (HFT) of harming market integrity is not exactly hot news. Hardly bestseller material, one might think. But Michael Lewis’ new book is exactly that: not only was ‘Flash Boys’ able to top bestselling lists in the US, it also encouraged federal investigators and New York State’s attorney general to start looking
into the trading practices described in the book. The secret of the book’s success lies in the wholly new dimension Lewis added to the ongoing mistrust of HFT. According to the author, HFT firms systematically exploit retail and institutional investors alike by front-running nearly every investor’s order on a regular and industrywide basis. To do so, they exploit different execution mechanisms – open order book trading against dark trading, to be exact. This is made possible by a highly fragmented US equity market
with seemingly unlimited latency arbitrage opportunities and the overly complex market structure based on the Regulation National Market System. These preconditions
create fertile ground for an allegedly corrupt and dysfunctional system where brokers sell off their customers’ order flow to HFT firms instead of protecting their clients’ best interest. According to the author, even the exchanges contribute to the scam by setting dubious incentives for HFT and brokerage firms to engage as liquidity providers respectively takers, and by complicating the market by introducing a vast number of order types shaped for HFT firms.

Selective use of facts:

Picturing the described conspiracy scenario, it should not come as a surprise that reading this book “makes your blood boil”, as the New York Times wrote. Indeed, ‘Flash Boys’ is a compelling
and upsetting read, but this strength of the book is also its weakness: instead of offering a balanced account of what is going on in the US stock market, Lewis leaves
out all facts that do not fit into his scam theory. First of all, Lewis describes HFT as a single trading strategy, not as the mere use of technology that it actually is. Moreover, he depicts the fact that HFT firms update their quotes on one market when being hit on another as proof of his theory, even though this is just the usual modus operandi of liquidity providers.to manage their risk. By omitting this fact, Lewis pictures liquidity provision as a riskless business which it is clearly not. That is exactly the reason why the described – and selfexplaining – “Scalpers Inc” business model allegedly implemented by HFT firms is simply not possible. Lewis describes a case of latency arbitrage: An investor inserts a buy order into a dark pool, and submits a corresponding sell order to a public market. The investor would have expected to get a fill at the same price in both markets. However, Lewis explains this was not the case: Savvy traders have exploited this arbitrage situation by buying at a low price from the investor in the public market and nearly simultaneously selling at a higher price to the investor in the dark pool. Ultimately, the investor ended up paying a spread for this round trip transaction.

Could it happen here?

The key question arising from this situation is: Could the same thing have happened in Europe as well, and could it happen on Xetra? Let’s first take a look at a typical, pan-European reference price dark pool (on a separate note, let us not further consider unregulated broker crossing networks for equities as they are declared illegal by MiFID II, and anybody sending orders to those facilities demonstrates a large amount of starry-eyed idealism anyway). This dark pool typically imports prices from a range of markets, and executions are made at a so-called consolidated European Best Bid or Offer (EBBO) price. Typically, prices imported from geographically distant markets are latency sensitive and might be outdated once they have arrived. In the case of the aforementioned investor sending orders to both markets, his sell order sent to the public market would have had a lowering impact on the EBBO price. But due to latency, the more favorable EBBO price might not have arrived fast enough to the dark pool. As a consequence, speed-savvy traders would have observed the new sell order on the public market and swiftly executed the investor’s buy order on the dark pool at the old EBBO price: This is the ‘latency arbitrage’ situation that Lewis describes in his book. Is this situation also possible with Xetra’s Midpoint offering? No, it is not. And here’s why: Xetra Midpoint does not receive prices from other venues. It imports its midpoint price only from its own open book and thus renders latency arbitrage impossible. Had the investor sent both orders to Xetra, the midpoint price would have dropped below the price of the investor’s sell order on the visible market and the unfavorable executions would not have occurred. Therefore, Xetra MidPoint is the only European offering in German securities providing investors with an accurate real-time price for midpoint order execution. Xetra MidPoint is a fully regulated market with independent supervision by the Trading Surveillance Office ensuring that market participants play by the rules.