FT: Mt. Gox and the mother of all short squeezes

When Mt. Gox, the Japan-based bitcoin exchange run by Mark Karpeles, stopped honouring redemptions on February 7, 2014, the company initially blamed the affair on an obscure tech fail known as a malleability issue. Many, however, were unconvinced by the explanation, suspecting foul play, a hack or an inside job. When Mt. Gox filed for bankruptcy on February 28 it emerged 750,000 of customers’ bitcoins had been lost, plus 100,000 of Mt. Gox’s own stash — a sum collectively worth $473m at prevailing exchange rates. Karpeles himself, however, insisted the exchange had been the victim of external sabotage or fraud. Time went by. Customers put their complaints to the authorities. Alas, not much in the way of information came their way. At some point, rumours began to emerge that much of the run-up in bitcoin’s price to a record $1216.73 in November, 2013, had been driven by a bespoke algorithmic programme known as the Willy Bot, developed by Mt. Gox for its own profit, and that this punting bot may have been the cause of many of the losses. Without concrete evidence, however, this too remained a theory. The mystery prevailed. Customers began to accept the reality: the money was gone and they’d never get it back because that’s what happens when you punt on an unregulated exchange.

Source: Mt. Gox and the mother of all short squeezes — FT.com

Publication: Crossed and Locked Quotes in a Multi-market Simulation

Crossed and Locked Quotes in a Multi-market Simulation

Financial markets are often fragmented, introducing the possibility that quotes in identical securities may become crossed or locked. There are a number of theoretical explanations for the existence of crossed and locked quotes, including competition, simultaneous actions, inattentiveness, fee structure and market access. In this paper, we perform a simulation experiment designed to examine the effect of simple order routing procedures on the properties of a fragmented market consisting of a single security trading in two independent limit order books. The quotes in the two markets are connected solely by the routing decision of the market participants. We report on the health of the consolidated market as measured by the duration of crossed and locked states, as well as the spread and the volatility of transaction prices in the consolidated market. We aim to quantify exactly how the prevalence of order routing among a population of market participants affects properties of the consolidated market. Our model contributes to the zero-intelligence literature by treating order routing as an experimental variable.