The Trades Are Free, the Data Will Cost You

I have written before about my model of stock-exchange fees. The basic idea is that big professional traders—bank trading desks, high-frequency trading firms—mostly get to choose which exchanges they trade at, but they don’t get to choose which exchanges they connect to. If you see 100 shares for sale at the same price at the same time at different exchanges, you can buy them at whichever exchange you like most, probably the one that charges you the lowest trading fee. But for various reasons—to maximize your arbitrage opportunities and minimize risks, to know what the market looks like, to comply with best-execution obligations—you pretty much need fast connections and direct data feeds at every exchange.

One implication of this is that traders will be price-sensitive to per-trade variable trading costs, but will have no choice but to pay whatever fixed data and connectivity charges the exchanges demand. Another implication is that a smart exchange will charge basically zero trading fees1 (in order to maximize its volume and importance2) and extremely high fees for fast connections and data (in order to maximize its revenue from captive customers). That is not quite what happens—trading costs are not zero3—but it’s a decent intuition; “revenues from ICE’s and Nasdaq’s various data-related businesses rival what the companies make from trading.”