Spurred on to Attack?

In our post yesterday we shared some recent work undertaken by Onside Analysis investigating who have been the most important players for Spurs this season. Today’s post is the second part of that analysis, turning our attention to Spurs’ attacking options and how they have performed this season.

Rather than repeating material from yesterday, if you missed that post we recommend you start there, to understand the methodology that we have used to produce today’s results.

While Spurs have had a pretty good 2012-2013, one area that has been particularly debated, sometimes quite vociferously, is what combination offers Spurs best attacking option.

With Defoe and Adebayor being the only two out-and-out strikers with Premier League playing time this season, it might appear that Spurs are not blessed with a multitude of options. However, PFA and Football Writers Player of the Year Gareth Bale has contributed more than his fair share of goals this season, often from advanced positions. Add Aaron Lennon and the attack-minded beginning-of-season signings Clint Dempsey and Gylfi Sigurdsson into the equation, and exactly who to include and who to leave out is far from a trivial question.

To illustrate the debate, we’ve included the following graphic, demonstrating the Average Difference in Actual and Expected Goal Difference, for each of the attacking players available to Spurs.

This graph is a subset of the graphic from yesterday’s post, restricted to the attacking players. When ordered according to our suggested ranking (taking into account Spurs’ performance when each player is in the team and when they are out, adjusting for the number of games in each case), Gareth Bale is the highest ranked attacking player. Given his contribution in terms of goals, many of then vital in securing the points for Spurs, this may not be surprising.

What may be more of a surprise amongst Spurs’ fans, is that Adebayor is the second ranked player. In fact, in the games where he is in the team Spurs’ have achieved an average difference of Actual and Expected Goal Difference of 0.35 goals. This is well above the Spurs’ overall average value (0.11), and in fact is the highest amongst all the attacking players. (The reason this doesn’t make Adebayor the highest ranked player by our measure, is that Spurs’ performance in his absence, under achieving Expected Goal Difference by 0.07 goals, is not any way near as poor as the average performance when other players are missing).

This graph is a subset of the graphic from yesterday’s post, restricted to the attacking players. When ordered according to our suggested ranking (taking into account Spurs’ performance when each player is in the team and when they are out, adjusting for the number of games in each case), Gareth Bale is the highest ranked attacking player. Given his contribution in terms of goals, many of then vital in securing the points for Spurs, this may not be surprising.

What may be more of a surprise amongst Spurs’ fans, is that Adebayor is the second ranked player. In fact, in the games where he is in the team Spurs’ have achieved an average difference of Actual and Expected Goal Difference of 0.35 goals. This is well above the Spurs’ overall average value (0.11), and in fact is the highest amongst all the attacking players. (The reason this doesn’t make Adebayor the highest ranked player by our measure, is that Spurs’ performance in his absence, under achieving Expected Goal Difference by 0.07 goals, is not any way near as poor as the average performance when other players are missing).

Defoe’s excellent form at the start of the season initially limited Adebayor to substitute appearances. In fact, when we investigated what would happen if we “redefined” a player being “in” the team for a particular match, setting a threshold of only having played 25% (23 minutes) or more of the game, Adebayor’s ranking dropped significantly. Given that player appearances lasting between 23 and 45 minutes predominantly occur when the player starts on the bench, this might indicate that Adebayor is not a good impact substitute, or that he is only brought on when the team is already underperforming relative to expectation. Either way, we felt that if a player had played less than half the match, this was too little time to attribute the result to that player, so feel more comfortable with our original results.

While looking at individuals is interesting, clearly football is not a solitary game, and separating out a player’s effects from his team-mates is a difficult but important question. (This, in our opinion, is one of the reasons why it is not straightforward to translate the Moneyball approach successfully applied in baseball to football).

To take the first steps towards understanding how players combine, we replicated the same analysis that we performed for individuals, but looked at pairs of attacking players. Specifically, we considered Spurs’ performance (the average difference of Actual and Expected Goal Difference) for each pair, looking at the games that the pair was “in” the team (using the original 50% threshold) and those when it was not in the team. Importantly, the numbers for any particular partnership make no assumption about whether players outside that pairing were playing or not. So, for example, the results for the Adebayor / Bale pairing, will include games when Defoe has played and games when he has not.

The top ranked pair is Bale and Dempsey. When they are in the team together Tottenham have exceeded Expected Goal Difference by 0.47 goals. This is greater than the average difference in Actual and Expected Goal difference for either individual player, suggesting that their contribution as a pair is greater than as individuals.  On the other hand, in the games where they have not been in the team together Spurs have underachieved by a goal difference of 0.2 goals.

Gareth Bale features in the 3 top ranked pairings, again highlighting his obvious importance to Spurs. However, if the comparison is between Defoe and Adebayor, the two out-and-out strikers, Adebayor appears to come out on top. His partnerships with Bale and Lennon have been more effective than Defoe’s partnerships with the same players.

One tactic that Spurs’ have employed a handful of times this season is playing Defoe and Adebayor together. When played together, Spurs have exceeded Expected Goal Difference by 0.37, significantly above the Tottenham average of 0.11. In the games where the two haven’t played together, Spurs have only exceeded Expectation by 0.05. This suggests that a large part of Adebayor’s success has come through the games when he has played with Defoe, rather than without.

Interestingly, the greyer colour of the bars featuring Adebayor indicates that the difference between Actual and Expected Total Goals (from both teams) has been fewer in Adebayor’s games than the average across all Tottenham games. As Spurs’ goal difference is more than expected in his games, this implies that the number of goals conceded by Spurs has been reduced. Given Adebayor’s height at 191cm, his value at defending set pieces may also be a factor in his higher ranking in our data.

With one game to go, the numbers are unlikely to change considerably. And with Defoe lacking game-time since his recent injuries, it is likely that Adebayor will start the crucial last game of the season. There might not be a real choice in this case, but perhaps these numbers might provide some reassurance that Spurs’ can perform at least as well with Adebayor in the team.

For a tactical review of Adebayor’s performance this season see this post from one of the Analysts.

This article first appeared on Onside Analysis – check out their excellent work here.