New Delhi: Canada’s draw with Bosnia & Herzegovina in Toronto last night appeared to be a series of disorganised clashes in which one team enjoyed the lion’s share of the ball but failed to convert until late in the game.
Much like Czechia’s 1-2 loss to South Korea the previous day, it was a classic stylistic battle between one team maximising open play, while the opposition sat back and relied on set-pieces.
This is a tactical debate that has dominated much of the discourse around club football this past season. A big reason for this is the underlying Expected Goals (xG) data behind Arsenal’s Premier League title victory and Champions League runners-up finish. In 38 Premier League matches, Arsenal created 66.13 xG, of which just 55% came from open play. The trend held for the Gunners in their 15 Champions League games as well, as roughly 57% of their xG of 28.48 featured open play chances.
Here’s how the Canada-Bosnia encounter played out in numbers. From 61% possession, Canada created 1.23 xG, while Bosnia managed 0.96 xG from the remaining 39%. Cyle Larin—the substitute striker who scored Canada’s equaliser—made up for one-sixth of Canada’s xG by himself, and he was only brought on in the 76th minute.
On the surface, it doesn’t quite read as the same contrast as the 1-1 score from the Champions League final, in which PSG generated 1.72 xG from 75% possession over Arsenal’s mere 0.51 xG across 120 minutes of laborious football.
However, the stylistic differences from the encounter in Toronto are starkly apparent. Roughly 79% of Canada’s xG was generated by their chance creation in open play, while around 98% of Bosnia’s xG involved goal threat from set-pieces.
This was evident from the game’s early stages itself. Despite Bosnia’s star striker Edin Dzeko missing the game amid fitness concerns, Canada seemed uncharacteristically panicked in defence.
That panic was especially amplified in set-pieces and any other occasion when the ball was played in the air. Bosnia were offered at least four shots on goal from free headers, including Jovo Lukic’s goal from a corner in the 21st minute. Overall, the Bosnians won 43 aerial duels compared to Canada’s 21—mostly within the first half.
The underlying data behind USA’s 4-1 win over Paraguay matched the one-way traffic on the pitch, but the Canada-Bosnia game in Toronto reflected similar stylistic differences to the Czechia-South Korea match in Guadalajara.
Just like Bosnia, Czechia too attempted to play to their strengths in physicality and set-pieces and took the lead. The major difference was that unlike Canada, South Korea were far more creative in chance creation and clinical in converting them into goals.
At the World Cup, set-piece significance is far from a new phenomenon. Of the 169 goals scored at the 2018 edition in Russia, 73—or 43%—resulted from set-pieces. But the 2022 edition in Qatar, which was the highest scoring ever, saw a significant downturn—less than a quarter of the 172 goals were from set pieces.
Of the 12 goals scored at this World Cup so far, just three goals in three different games have come from set plays. One from Czechia, one from Bosnia and a consolation effort from Paraguay after USA were already 3-0 up. While it’s a tiny sample size with 100 more games to go, it doesn’t reflect a major shift away from the last World Cup just yet.
The tournament’s expansion to 48 teams and the aforementioned data around ‘set-piece kings’ Arsenal’s style of play had raised pre-tournament questions of whether more teams will prioritise set-pieces the way teams like Czechia and Bosnia already have.
But during a media roundtable last month, FIFA’s Technical Study Group (TSG)—which analyses all World Cup games with the help of ex-players and data specialists—downplayed the possibility of a similar ‘set-piece boom’ as Russia 2018.
According to TSG member Gilberto Silva, who won the World Cup with Brazil in 2002: “[Y]ou don’t have much time to prepare a team for these tournaments. Of course, [set-pieces] can be a weapon and teams will use it, but not as the main one.” The data so far looks to be supporting his point.
