If markets are efficient, you will never make profitable trades as a result of reading the Financial Times. Efficient markets move quickly and respond to any new headlines – disappointing earnings, a cut in interest rates, a fraud or a safety incident. Markets will sometimes overreact, drifting backwards after a lurch, or underreact, taking time to digest the true impact of the new information – but overreactions and underreactions should balance out. And when no news is available, the prices of an efficient market won’t change much.
But do markets really react efficiently to news? It would be easy to tell if it were easy to identify all genuine news. Sadly, it is not. Yet two inventive new academic papers claim to have solved the problem of identifying news, in two very different contexts. The studies could not be more unalike. One looks second-by-second at trading data from one of the world’s most active financial exchanges. The other analyses market information that is more than two centuries old.
Karen Croxson and J. James Reade of Oxford University studied the Betfair exchange, a sports betting site that supports many more trades than the London Stock Exchange. Betfair allows punters to bet on football games, and the market stays open throughout the match. Croxson and Reade studied how the price of different bets varied as goals were scored during English league games.
This is an excellent test of the market’s response to news: the bets have a clear value at the end of the game, goals are scarce and important events – and (unless one is a referee) they are easy to spot. And the stakes are not trivial: hundreds of pounds a second are wagered during the match.
The idea of using sports betting to test market efficiency came from Steven Levitt (the co-author of Freakonomics) and Ricard Gil. Levitt and Gil had conducted an earlier study in rather thinner betting markets, and found that prices jumped immediately after a goal, but they then drifted further in the same direction. Was that because the traders were sluggishly digesting news of the goal? Or was it because the clock was ticking down, no news being good news for the team in front? Croxson and Reade offer a clever answer, by looking at those goals scored just before half time. Relevant news hardly ever emerges during half time and the pair find that, although trading is active during the break, prices barely move at all. This shows that the market traders instantly absorb the news of a goal. After the second half begins, prices start to drift again, just as Gil and Levitt found.
That suggests an efficient response both to news and to the absence of news, in sports betting markets at least.
But Peter Koudijs of Barcelona’s Universitat Pompeu Fabra has a different perspective. He looked at prices of three English stocks (the East India Company, the Bank of England and the South Sea Company) on a secondary market in Amsterdam from 1771 to 1777.
Koudijs realised, and proved, that relevant news flowed almost exclusively from London to Amsterdam – and always through the same channel, a boat sailing across the North Sea bringing market data to Amsterdam. Depending on wind speed and direction, the “packet boat” might arrive promptly or after a delay of more than a week, occasionally starving the Amsterdam market of news for days on end.
Koudijs discovered that when the wind was unfavourable and no news was available, Dutch prices for these English companies were highly volatile anyway. That is not an efficient market.
So, have we discovered something uniquely inefficient about Dutch markets, or something uniquely efficient about sports betting? I am not sure. An analysis of Dutch football games is the logical research extension.
Also published at ft.com, subscription free.