A Game of Two Halves: Have Leeds United learned a lesson from last season?

In this article, Joe Morgan looks at Leeds United’s first- and second-half performances from last season to see if they have learned anything this season

We all know that, during the tenure of Marcelo Bielsa, Leeds United have tended to perform better in the second half than the first.

In the 2018/19 season, Leeds scored 73 Championship goals. However, only 26 of those came in the first half of matches which means Leeds scored almost two-thirds of their league goals after half time.

In fact, if last season’s matches ended at half time, Leeds would have accumulated only 58 points with just 11 victories and would have finished in 17th place. By the same token, if just the second half of matches counted then Leeds would have finished 2nd: 5 points clear of Sheffield United in 3rd place.

They didn’t. Instead, they finished 6 points adrift of the top two and their season ended in a devastating defeat at home to Derby County in the play-off semi final. But why were Leeds so much more effective in the second half of games?

From losing positions at half time in 10 games, Leeds were able to pick up 8 points. Of the 25 games that were level at half time, Leeds won 11, drew 6 and went on to lose 8 of them. In the 11 games Leeds were leading at half time, they won all 11.

Those numbers alone don’t necessarily suggest that Leeds’ second-half performances were anything special but Bielsa’s team did manage to pick up 47 of their points after improved second-half performances: over half of their total points.

The underlying numbers tell a slightly different story, though. In the first half of their games last season Leeds scored 0.56 goals and conceded 0.48 goals: a first-half goal difference of 0.08 per game. In the second half of games, however, Leeds scored 1.02 goals and conceded slightly more at 0.65 goals: a second-half goal difference of 0.37 goals per game.

Leeds scored 1.8 times as many goals in the second half of games than they did in the first half. As a result, it’s hardly surprising that they also conceded slightly more goals in the second half given that they were theoretical producing twice as much attacking output.

Comparing the actual goal-scoring numbers to expected goal data shows a similar pattern. First-half data shows Leeds with an xG of 0.7 and xGA of 0.42 which suggests they were maybe unfortunate to have such tightly contested first-half performances in some cases. When it comes to second halves, the data has Leeds with 0.96 xG and 0.54 xGA.

Last season, then, Leeds underperformed their xG in the first half of games and overperformed their xG in the second half. Their opponents, however, overperformed in front of goal in both halves. Had expected goal data been met, then Leeds would have finished with a goal difference of +32 — a goal difference 9 goals better than what they actually finished on — and would have presumably left them finishing on a higher points tally.

Leeds improved their attacking output greatly in the second half of games while managing to keep their defence about as solid as it was during the first half. So how did they manage it?

To begin with, Leeds allowed their opponents 0.5 shots fewer during the second half of games. Nevertheless, those second-half shots were more likely to hit the target by 10% than the first-half shots Leeds conceded. By contrast, Leeds themselves took almost 2 more shots in the second half than the first and hit the target over a third of the time instead of just over a quarter of the time in the first half.

On top of this, they also made an extra 2 ball recoveries and made almost 6 more challenges in the second half of games. Leeds’ PPDA (Passes allowed Per Defensive Action) was practically the same in both the first and second half, which suggests that there was no let-up in the intensity of Leeds’ pressing. This, coupled with the increased challenges and ball recoveries, shows how Leeds attempted to impose themselves on their opponents more in the second half of games.

Interestingly, opposition’s PPDA decreased in the second half of games, meaning that they allowed Leeds to play fewer passes before they attempted to make a tackle or intervention. Through the course of the season, it was clear that Leeds struggled to break down teams that defended deep and in a low block. This more proactive approach of opponents may have allowed Leeds more space to exploit and could be a key reason as to why Leeds were better in the second half.

They also found themselves susceptible to counter-attacks later in the game as they committed more players forward in an effort to break down defences. In the second half of matches last season, Leeds recorded slightly more counter-attacks resulting in a shot and they recorded just over one extra cross in the second period. Those crosses were marginally more accurate than those from the first half. Leeds were also involved in 3.5 more offensive duels during the second half of games and won a higher percentage of them. They took almost one more corner in the second half of games and a higher percentage of those corners ended with a shot on goal.

The stats show slight improvements in most aspects when comparing the second half to the first, and whilst these improvements are only slight, they culminate in quite a significant improvement in terms of goal scoring and quality of goal scoring chances, with Leeds greatly increasing the number of goals scored in the second half of games.

The defensive numbers don’t suggest a significant difference between the two halves of the game. Leeds allowed slightly more shots on target against them during the second half which could go some way to explaining why the opposition scored more goals at an improved xG against Leeds in the second half.

Leeds also made slightly fewer interceptions and were forced to make more clearances, win more defensive and aerial duels and make more fouls. All of this suggests that Leeds come under more pressure during the second half of games and could mean that games were more open in the second half in general.

This is to be expected as teams would try to limit opportunities for Leeds in the first half so as to avoid having to chase the game in the second half. When teams held Leeds at the break, which was the case 25 times last season, they were more likely to come out more positively in the second half. This could explain why opposition attacking numbers improved slightly in the second half but also why Leeds’ did as well: they had more space to play in as their opponents looked for an opening or, alternatively, grew tired. This also might explain why Leeds’ defensive numbers stayed fairly consistent across the game but their attacking output significantly increased during the second half.

Is this trend likely to continue into the new season?

Obviously, it’s impossible to say with any real certainty with such a low sample size but in Leeds’ three games so far this season they are yet to concede a first-half goal during which time they have scored 0.67 goals. After half time, they have scored 1.67 goals and conceded 0.67 which suggests that maybe Leeds are set to be a second-half force once again.

Expected goal data, however, shows that Leeds’ first- and second-half performances are fairly similar. Leeds have xG numbers of 0.78 in the first half and 0.76 in the second half: it’s the xGA numbers that have much less parity. Leeds have given up an xGA of 0.39 in the first half but that rises to 0.82 in the second half. These figures, of course, come from a very select few games, but Leeds have allowed over double the number of shots in the second half and they have only 54% of possession compared to 64% in the first half.

Leeds also attempt 65 fewer passes in the second half of games. This could be down to the fact that Leeds have been ahead at half time in two of these games so the opposition may have come out in the second half with a more attacking mindset. On top of this, the numbers from the Forest game were the highest Leeds allowed in terms of xGA: 1.02 in the second half. This was greatly affected by their equaliser, though: Lewis Grabban’s somewhat freakish goal had an xG value of 0.8 which contributed to the majority of Forest’s threat.

Away from the expected goal data, though, Leeds still appear to have less dominance in the second half of games so far this season. These three attack momentum plots show how Leeds’ opponents have had longer periods of attacking threat in the second half than the first. The second one, in particular, shows how unlucky Leeds were to not get more than a point against Forest.

Of course, it could be the case that the change this season has come from the opposition taking a different approach to playing Leeds. Knowing that they will be taken apart if they play too expansively, it makes sense that Leeds’ opponents would sit deeper in the first half and look to snatch a goal in the second. As you can see from the threat momentum plots, each opposition has a period of sustained pressure late in the second half where they are looking for a goal.

It’s easy to grow frustrated watching Leeds having to break down a low-block and absorb counter-attacks. But it’s also important to remember that this shows that oppositions are treating Leeds with a huge amount of respect. And yes, there will be games where a lucky goal comes and Leeds drop points — such as the game against Forest — but in the course of the season, opponents are making Leeds’ defensive duties easier. All that remains is for them to unpick the lock.

It would be foolish to draw any conclusions about this season from what’s happened so far but it does appear that Leeds’ may not have to rely on the strength of their second-half performances so much this season.

Joe Morgan can be found on Twitter at White Spy.

If you would be interested to write a piece for All Stats Aren’t We, get in touch!

A Leeds United blog which focuses on the tactical and statistical aspects of the game