There’s another fantastic John Lanchester essay in the London Review of Books. This one reviews Flash Boys by Michael Lewis, an expose of high frequency trading that was followed by an FBI investigation into this practice the day after the book’s release (though the FBI deny the connection). I’d read about algorithmic trading in the past and found it very interesting but I hadn’t encountered ‘front running’ before: the indefensibility of which is, as Lanchester puts it, “IOTTMCO (Intuitively Obvious to the Most Casual Observer)”:
We want a market to be people buying and selling to and from each other, in a specific physical location, ideally with visible prices. In this new market, the principal actors are not human beings, but algorithms; the real action happens inside computers at the exchanges, and the old market is now nothing more than a stage set whose main function is to be a backdrop for news stories about the stock market. As for the prices, they move when you try to act on them, and anyway, as Lewis says, there’s the problem of the ‘dark pools’, which are in effect private stock markets, owned for the most part by big investment banks, whose entire function is to execute trades out of sight of the wider public: nobody knows who’s buying, nobody knows who’s selling, and nobody knows the prices paid. The man who did most to help Katsuyama understand this new market was an Irish telecoms engineer called Ronan Ryan. Ryan’s job involves the wiring inside stock exchanges, and he explained to Katsuyama just how crucial speed has become to the process of trading. All the exchanges now allow ‘co-location’, in which private firms install their own computer equipment alongside the exchanges’ own computers, in order to benefit from the tiny advantage this proximity gives in trading time. [….]
Speed matters so much because the new financial techniques involve exploiting minute discrepancies in price that exist only for fractions of a second. Exploiting them and, at times, creating them too. Aided by co-location, high-frequency traders ‘sniff’ incoming orders made by people like Katsuyama. In microseconds (that’s millionths of a second) their computer algorithms buy the share before the order has been executed, and then sell it on to the initial buyer for a tiny but guaranteed profit. That is the reason Katsuyama couldn’t buy shares for the price shown on the screen: he was being ‘front-run’. This is one of the main ways high-frequency traders make their money. The other two techniques singled out by Lewis are ‘rebate arbitrage’ and ‘slow market arbitrage’. Rebate arbitrage occurs when HFT firms profit from the fact that some exchanges now pay for traffic, by directing trades accordingly in order to make money from kickbacks.