A new World Cup predictive model was released by Five Thirty Eight. It projects Brazil as massive favorites to win the World Cup and gives much smaller chances to Spain, Germany and Argentina. Nate Silver even specifically addressed my World Cup underdogs, Portugal. He wrote:
Portugal? SPI is more down on the Team of Five than it seems it should be. In SPI’s defense, Portugal was a little underwhelming in World Cup qualifying, drawing twice with Israel and once with Northern Ireland. And the team isn’t deep: While Cristiano Ronaldo is one of the best two or three footballers in the world, Portugal has no other player who clearly belongs in the top 100.
The key point here is that Portugal’s World Cup qualifying campaign was unimpressive, in particular because of draws with weaker sides Northern Ireland and Israel. The underlying “Soccer Power Index” rating uses goals scored and conceded as its primary inputs. Portugal scored five goals and conceded five these draws, and so for SPI, those results project that Portugal will play poorly in the future.
Expected goals, by contrast, is based on the insight of analysts like James Grayson that finishing rates are extremely variable. In a three-match sample, a team’s underlying ability is much better reflected by the quality of chances that they create rather than the rate at which they convert these chances. Hot and cold streaks in shot conversion are common, and they rarely reflect an underlying strength or weakness in a soccer team.
This chart displays the shots in Portugal’s three qualifying draws, weighted for expected goals value. An xG method looks at these games and concludes that Portugal dominated in the creation of chances, and thus is likely to play well in the future. Likewise, the rich vein of finishing form that Israel and Northern Ireland found in the qualifiers should not reflect poorly on Portugal.
My World Cup predictions will be based on expected goals. This method produces some unexpected results, like the high rating for Portugal. But when I look at that shot chart, I have trouble believing that the Portuguese should be rated poorly because of their performance in those matches. They did not finish well, and unexpectedly Israel went on a fantastic run of shooting form. But in the games to come, those finishing rates are likely to regress to the mean. That leaves chance creation and expected goals as better markers of Portugal’s skill, and that’s why my model likes Portugal much more than Nate Silver and Five Thirty Eight.