Do Results Lie? (or The Introduction of ‘Tactistics’)

Do Results Lie? (or The Introduction of ‘Tactistics’)
August 4, 2015 Mihail Vladimirov

By Mihail Vladimirov.

Statistics and Tactics combined

The use of both statistics and tactics have been part of football to some degree since its inception. Both have also been valuable tools in different periods to aid sharp footballing minds to facilitate the next step of football evolution – people like Charles Reep, who used statistical data to invent the ‘long-ball’ strategy during the ’50s, and Rinus Michel and his ‘Total football’ tactical masterpiece designed as an answer to Helenio Herrera’s defensive ‘Catenaccio’.

In the past decade both spheres have gained greater importance and coverage. Clubs are now taking care of their own data needs and seek as much diverse analysis as possible, with the birth of sport science teams, multilateral analytics tools and divisions supporting the decision-making process, the creation of specific scouting departments tasked with various types of tactical espionage and research, etc. In recent years we have also seen the media and fans (so-called ‘Fanalysts’) covering what is happening on and off the pitch. Perhaps the reason why the statistical and tactical points of view have been isolated from each other until now is because few statistical analysts also possess an equally deep tactical understanding, which would give them the knowledge of how tactics influence the game. Likewise, not many tactically-minded people trust the numbers and still mainly prefer to back-up their beliefs and decisions with what they can see with their own eyes. I believe the next step for media and fans is for the worlds of statistical and tactical analysis to work in partnership (perhaps we can invent the term ‘Tactistics’!).

Neither approach is ‘wrong’ per se, but even if a person is a specialist in either department, by ‘disregarding’ one of the perspectives he may not be able to view the whole picture and so enjoy the maximum level of analysis. Obviously, certain information doesn’t always require both views. For example, how many tackles a player has made per game doesn’t need any tactical input. Equally, knowing what formation your team plays doesn’t need anything statistical. But understanding why exactly that player made so many tackles (or not) may require a deeper tactical understanding of his role within the side – which will be based on what type of player he is, his preferred style of play, what his role is within the formation, and the players around him, etc. Similarly, if you want to analyse the different formations your team is using, it may be helpful to have more than just a tactical understanding – some stats and figures will definitely help you gain a greater understanding and make a more detailed comparison.

Both sides of the coin are great to answer certain questions but are limited. Using various metrics and data to give you the numbers on everything that an individual player or the whole teams has done in a game, during a whole season, etc, statistical analysis is great to answer questions like ‘What happened?’ or ‘How often?’. But it can’t really explain ‘Why’ and ‘How’ all of this happened – which is where the tactical analysis comes into the play. Logically, if you’re interested in the full picture to be confident in your evaluation or to inform your decisions, you can’t do this with only a tactical or statistical analysis. To provide a superior analysis of what happened and why and how it happened, you will need both views to cooperate.

‘Truth’ of a result

There are many different ways in which a statistical and tactical analysis can work together to provide you with the full picture. What you wish to know will determine which metrics/stats/data and tactical aspects you’ll need to combine. For the purpose of this article, we set out to investigate Liverpool’s 13 games unbeaten run last season, and so the main need is to evaluate the ‘rightness’ of each result through the prism of how both teams performed.

Despite the way the season actually finished, let’s just go through the list, headline-like, of the facts of Liverpool’s 2014-15 season achievements. They finished sixth (which is only a place below the proverbial ‘par’ spot). They reached the semi-final stage of the two domestic cups. They participated in both the European tournaments. And they registered yet another remarkable run of games rarely seen in the modern game (13 league matches unbeaten). You could easily revise your view and say that last season was not such a bad one for Liverpool after all.

However, when you analyse precisely how and why Liverpool managed to achieve all these things you would probably have to revise that view again. Other articles and debates have already gone through the details and dissected the majority of the season-describing topics above. But arguably it’s Liverpool’s finest season achievement which is yet to receive the same level of scrutiny.

The ‘headline’ says it was that three month run, between mid-December and mid-March, that saw the Reds miraculously revive a season which had increasingly looked like it was going into a coma. It certainly revived their challenge for the Top 4. The season before had seen a similar run of games achieved by Rodgers’ team, with 11 consecutive wins between February and mid-April, capturing fan’s imagination who proudly proclaimed it ‘poetry in motion’.The common thing between unarguably Rodgers’ two peak moments at Liverpool (so far) is the way the attention was cast exclusively on the mind-blowing results alone, with little interest shown at the time in how exactly they were achieved performance-wise (both tactically and statistically). Last summer, when things had calmed down, a two-part examination (links to Part 1 and Part 2) attempted to fill that gap in the coverage.

The purpose of this article is to offer a similar cold light of day analysis of the unbeaten run of 2014-15.

Arguably, to determine how a clash between two teams ‘should’ have resulted, two things are required. First of all, an analysis of how the game panned out tactically, meaning how both teams performed against each other in the actual game context. And secondly, a statistical evaluation of the chances created and a judgement of their quality. Goals win (and lose) you games and so, in the simplest terms, your game strategy is your plan to score and prevent goals. At its simplest, tactics is the detail of how you intend to achieve your game strategy, including organising your team based on the strengths and weaknesses of the opponent in the various phases of play to achieve the aim of scoring and preventing goals. In the meantime, statistical analysis is going through the numbers for various key aspects to illustrate which team achieved these aims best.

To improve on the type of information the statistical analysis is able to provide, in the past few years various analytical models have been developed. They combine certain stats/metrics to produce an insight into what actually happened. One such model is ‘Expected goals’ (xG). In simple terms, the method estimates the quality of chances a team creates and concedes which then produces an ‘expected goals’ value for both teams based on certain factors (a detailed explanation can be read here).

For the purpose of this article I’ve decided to use three different xG models to give greater credence to the rounded evaluation of the 13 game run. I’ve labelled the three models in the table below as v1 (author Michael Caley @MC_of_A), v2 (from Julien Assuncao @Birdace), and v3 (the work of Stephen McCarthy @SteMc74). To simplify the stats comparison, all xG estimates list Liverpool first (regardless of whether they were home or away).

The 13 game run

(click on the name of each game for a link to its full post-match tactical analysis)

Match #1 – Arsenal (H)
Actual result: DRAW 2-2

In a game that could have ended either way, a 10-men Liverpool eventually scraped the draw with a Skrtel goal from a corner in the dying minutes of added time. The Reds enjoyed overwhelming midfield dominance due to a 4-v-3 advantage, which allowed Coutinho and Lallana to run riot between the lines against the isolated and unsupported Flamini. But familiar issues robbed Liverpool once again, with shaky defending leaking two cheap goals, and with the lack of midfield runners and Arsenal’s defence staying deep to deny Sterling space to run in behind, Liverpool’s blunt attacking couldn’t convert any of their impressive spells of passing combinations into genuine goal-scoring opportunities. Meanwhile, for all their clinical finishing in front of goal, Arsenal were surprisingly sloppy in possession and couldn’t really launch more than a few dangerous breaks.

Skrtel v Arse

Statistically speaking, Liverpool’s dominance is shown in the 23 chances created. But their poor attacking display led to poor quality chances. Both teams only produced one ‘big chance’. Arsenal’s led to their second goal while Liverpool’s was a situation resulting directly following Arsenal’s sloppiness at the back when they couldn’t clear their lines. Surprisingly, given the above, all of three xG models favoured the Reds in this game and awarded them a definitive win.

Tactics and performance verdict: DRAW
xG models verdict:
v1 = WIN (1.8 – 0.9)
v2 = WIN (1.83 – 0.99)
v3 = WIN (1.82 – 0.69)

Match #2 – Burnley (A)
Actual result: WIN 1-0

This game was the first in the run when the Reds had enormous troubles. The resistance they got from the opponent was exclusively based on tactical intelligence and physical effort. Burnley dominated the whole of the first half thanks to overloading the midfield zone, allowing Liverpool time on the ball at the back while restricting their forward passing options and then pouncing with an energetic press whenever a deep-lying Liverpool player on the ball looked in trouble. While Dyche’s team were first to every ball and succeeded in making the game scrappy, ultimately they lacked any real quality in attack. Just like Liverpool in the Arsenal game, Burnley lacked any real quality in their chance creation, resorting to crosses into the box and long shots.

Liverpool only came into the game in the second half when Burnley tired and could no longer maintain their pressing. Their individual quality eventually shone through when Coutinho put Sterling through on goal to score the winning goal. Thereafter the Reds showed their superior quality and intelligence by smartly shutting down the game, keeping the ball well, and taking the sting out of the game.

Liverpool were economical going forward, with only one ‘big chance’ (Sterling’s goal) but were also efficient as that chance came from only nine chances created. In terms of the xG models it is interesting that they contradict the tactical analysis again with none of them giving Liverpool a solid enough advantage to ‘earn’ the win here.

Tactics and performance verdict: WIN
xG models verdict:
v1 = DRAW (1.3 – 1.0)
v2 = DRAW (0.57 – 0.80)
v3 = DRAW (0.64 – 0.76)

Match #3 Swansea (H)
Actual result: WIN 4-1

This was definitely the best Liverpool performance in that run of 13 games (and arguably of the season). Rodgers’ team totally dominated their opponent offensively and defensively.

Tactically, the game’s highlights were all within the first half. With their reactive strategy, Swansea made it easy for Liverpool from the off (standing-off, getting men behind the play and looking to defend zonally inside their half). The Swans’ two banks of four not only surrendered the initiative to Liverpool too easily, but it didn’t even help in terms of providing the intended positional solidity it was presumably in place designed for. The midfield two of Britton and Shelvey were outnumbered 4-v-2, leaving them isolated and struggling to perform their defensive roles.

The home team used their 4-v-2 midfield supremacy to simultaneously bypass Swansea’s deep sitting and compact  front pair, overload their struggling midfield duo and then continuously involve Coutinho and Lallana in dangerous pockets of space to facilitate threatening attacking moves. With the added help of Henderson splitting his roles with Lucas to provide the required movement diversity and so providing an extra attacker joining from deep in midfield, the home side offered a more ‘complete package’ of genuine attacking threats.

Even when Swansea had spells of dominance, Liverpool showed tactical versatility to adapt and provide both the required defensive solidity and counter-attacking threat.

Statistically speaking this was a real thrashing. Liverpool registered 16 chances with four being clear-cut. (two of these were not actually converted, otherwise the result would’ve been even more daunting for the visitors). Given the huge disparity in both teams’ performances, it is only logical to see all of the xG models gave Liverpool an explicit win.

Tactics and performance verdict: WIN
xG models verdict:
v1 = WIN (2.3 – 1.1)
v2 = WIN (2.57 – 1.20)
v3 = WIN (2.75 – 1.11)

Match #4 – Leicester (H)
Actual result: DRAW 2-2

For the second time in the run an opponent tried to pressurise Liverpool and gave Rodgers’ players huge problems, and largely ‘levelling the playing field’. Even using a slightly different formation to Burnley, Leicester hassled the Reds’ build-up play. Leicester found much more joy going forward than the Clarets though, mainly thanks to their pacy wingers Schlupp and Mahrez.

Liverpool’s whole build-up play was slow and ponderous, often followed by sloppy forward passes which gave the ball back to the opposition. After the opening penalty, Leicester’s intensity dropped and the home team became more settled in possession, creating a couple of decent opportunities. Towards the end of the half though, following an enforced substitution and a change in the formation, Pearson’s team once again ‘levelled the playing field’ and began causing the Reds big problems, despite going 2-0 down from another Gerrard penalty.

The presence of a second natural forward helped Leicester once again swarm forward at the start of the second half, leading to panic and uncertainty creeping into Liverpool’s defensive play. The visitors began to focus all their attention down Liverpool’s left side, with right winger Mahrez regularly moving infield to further exploit the zone between the lines where the immobile duo of Lucas and Gerrard couldn’t adequately protect the backline. Eventually, and deservedly, Leicester equalised with two goals in quick succession around the hour mark.

In the remaining 30 mins Leicester logically adopted a more defensive outlook which handed the initiative back to the home side. But despite the his substitutions and formation changes, Rodgers’ team couldn’t really open them up.

Stats-wise this game was as big a disappointment as the tactical analysis suggested. Liverpool had two big chances (from 13 created) but both of them came from penalties which Gerrard converted. Meanwhile Leicester had the only open-play clear-cut opportunity in the game (interestingly it wasn’t any of their goals but an opportunity they failed to capitalise on with the game still at 1-0). The xG models are unanimous in also being quite critical of Liverpool’s play, with M. Caley’s model being very close to even awarding Leicester the win.

Tactics and performance verdict: DRAW

xG models verdict: 

v1 = DRAW (1.1 – 1.5)

v2 = DRAW (0.92 – 0.83)

v3 = DRAW (0.94 – 1.08)

Match #5 – Sunderland (A)
Actual result: WIN 1:0

In this game Liverpool were truly deserving winners, even if they didn’t shine as brightly as in the Swansea game. In what could be described as a low-key ‘professional’ win, the Reds’ tactics once again provided the platform for undisputed supremacy right from the start. The main talking point was the role of Markovic as a runner off the flank serving to compensate for Henderson (who stayed deep to provide an extra layer of defensive stability and playmaking abilities from behind the play) and Gerrard (now asked to perform an advanced midfield role) lacking the mobility to burst into goal-scoring positions. In addition Borini also caught the eye with the way he usefully spearheaded the attacks and worked the channels to open up the required space for others to play into.

Marko v Sund

Sunderland were poor throughout the whole game and when they were reduced to 10-men at the start of the second half it was ‘game over’. The only disappointment in Liverpool’s play was that they couldn’t take maximum advantage of their man-advantage and create more threatening attacking moves and goal-scoring chances having played the whole of the second half 11-v-10.

The stats reflect the nature of a game with Liverpool quietly dominating but being efficient enough to win the game without too much fuss. They had 15 chances, with two being clear-cut (Markovic’s opening goal, with the other being Borini’s missed chance to make it 2-0 towards the end of the first half). Sunderland were so poor they barely registering any goal-scoring threat, and all three xG models had Liverpool as convincing winners.

Tactics and performance verdict: WIN

xG models verdict:
v1 = WIN (2.1 – 0.5)
v2 = WIN (1.21 – 0.17)
v3 = WIN (1.68 – 0.26)

Match #6 – Aston Villa (A)
Actual result: WIN 2-0

Villa started the game very promisingly by being aggressive and man-oriented to deny Liverpool space and time on the ball in the midfield third. But the home side had two clear weaknesses that eventually caught up with them and made the game so much easier for Liverpool than it might otherwise have been.

In spite of all the positives gained by their aggressive pressing, Lambert’s side were incredibly shaky defensively due to how open they were in front of their back four. Not only did Liverpool have the 4-v-3 advantage based on how both teams lined-up formation wise, but the hosts had a deep-lying playmaker and not a proper holding midfield as their deepest midfielder working as the main shield between the lines. This meant that, without Liverpool even playing that well as a whole (they were too ‘boxy’, lacked midfield fluidity and were defensively sluggish), Coutinho and Sterling could exploit the ample space left between the lines and complement Borini.

At 1-0 Liverpool retreated into counter-attacking mode, which helped them stabilise their game defensively while further decreasing their overall threat going forward. This gave Villa the green light to start the second half on the front foot. The tactical tweaks resulted in the home side having a period of putting Liverpool under pressure using their newly found movement fluidity and extra support to Benteke. Despite all this Lambert’s side struggled to create any truly dangerous attacking moves and eventually their pressure faded with Liverpool finishing the game as the team in control (with Rodgers’ in-game adaptations being the main reasons for the Reds doing so with efficiency).

Once again Liverpool benefited from an opponent playing rather poorly, and performed just well enough to earn a victory. The visitors had 12 chances with two being big ones (coming close together – one scored and the other missed by Sterling at 1-0 which would have killed the game off before half-time). The narrowness between both teams’ performance is shown in how differently the xG models saw the game.

Tactics and performance verdict: WIN

xG models verdict:
v1 = DRAW (1.2 – 1.3)
v2 = WIN (2.00 – 1.32)
v3 = DRAW (1.50 – 0.94)

Match #7 – West Ham (H)
Actual result: WIN 2-0

This was another game when a pressing team gave Liverpool troubles early on. West Ham started proactively with Allardyce ordering his midfielders to get tight and ‘match’ Liverpool’s. As was now seeming to become the norm, the visitors also chose to target the Reds’ right-hand side creating another batch of problems for the home side to deal with. In addition, with Lallana and Coutinho both forced into helping out defensively and continually drop deep and/or wide, Sterling became isolated up front.

When West Ham’s initial pressing intensity faded, Liverpool gradually started to see more of the ball, but despite the increased possession the home team struggled to open up their opponents. The only attacking danger Rodgers’ team managed to produce was the Markovic chance at the end of the first half.

A rare West Ham defensive error and good Liverpool combination play at the start of the second half saw Sterling open the scoring. With the onus now on the visitors to get back into the game, their manager quickly reacted by reshuffling his team tactically. But the Hammers continued to look sluggish both on the ball and especially without it, so it was easy for Liverpool to defend and then organise dangerous counter-attacks. Following such a situation, Sturridge made it 2-0. When West Ham were reduced to 10-men when Reid was forced to depart through injury with all Allardyce’s substitutions used, it meant it was ‘game over’.

Despite their initial defensive and offensive struggles, this was a deserved win for Liverpool mainly thanks to how poorly their opponent played after the bright opening period. All of this was reflected in the chance creation stats with the Reds registering three big chances (out of 10 created) and West Ham failing to seriously threaten Mignolet. Unsurprisingly the xG models are similarly categorical in their appraisal of the game.

Tactics and performance verdict: WIN

xG models verdict:
v1 = WIN (2.5 – 0.6)
v2 = WIN (2.02 – 0.39)
v3 = WIN (2.06 – 0.34)

Match #8 – Everton (A)
Actual result: DRAW 0-0

In what was a dull and largely uneventful game Liverpool struggled to break down an Everton team set up by their manager to mainly nullify the attacking threats of their Merseyside rivals. The Toffees succeeded by preventing Liverpool’s left-sided trio of Moreno, Coutinho and Sterling linking-up play dangerously and building any positional threat. As in the earlier derby at Anfield the key defensive role for Everton was perfectly executed by McCarthy. Meanwhile Naismith, playing centrally, worked tirelessly to make his team compact and hard to penetrate through the middle by continually dropping deep to make it 4-v-4 in midfield.

For all of Everton’s great defensive work, Liverpool simply lacked any kind of creativity and movement fluidity to even stand a chance of working around Martinez’s reactive tactics. Rodgers’ team was once again too ‘boxy’ in terms of shape and movement off the ball, and lacked runners getting ahead of the ball to try and stretch the play in diverse ways.

Even when Sturridge was brought in in attempt to increase Liverpool’s direct attacking threat, Everton were quick to react, nullifying the danger by simply dropping ten yards deeper and minimising the space for him and Sterling to operate in. This further exposed the Reds’ lack of creativity and complementing movement fluidity.

That neither team did enough in attack to be a worthy winner is reflected by the stats, with neither registering a clear-cut chance. Liverpool had 12 chances but they were all hopeful pot-shots rather than well-worked penetrative combinations. Despite all this, all of the xG models (even while rating their efforts differently) make it very clear that Liverpool had the upper hand and deserved to be winners.

Tactics and performance verdict: DRAW

xG models verdict;
v1 = WIN (1.9 – 0.4)
v2 = WIN (1.08 – 0.37)
v3 = WIN (0.94 – 0.23)

Match #9 – Tottenham (H)
Actual result: WIN 3-2

Tactically, this game revolved around Spurs having a 5-v-4 overload in the middle and, given how both teams lined-up formation-wise, the fascinating 1-v-1 duels on the flanks. With Kane dropping deep joining Dembele, Eriksen, and Lamela creating overloads and dragging the three man defensive unit all over the place, and with Mason also pushing forward, along with Liverpool’s lack of a natural holding midfielder, it was clear why they struggled to cope with Tottenham’s attacking threat.

Due to Spurs’ dominance, the Reds had to adapt and play on the break, successfully using their wing-backs (especially Ibe) with Markovic moving across into pockets of space, helping cause overloads themselves and link up the play.

The second half started how the first one ended, with the home team enjoying more possession but unable to create anything meaningful with it because the visitors now sat deeper and the Reds had no-one and no space to target behind the away team’s defensive line. Meanwhile Pochettino’s team were happy to either retain ‘sterile’ possession or hit on the break as and when the situation allowed for it.

In a game full of defensive struggles from both teams, the stats show a minor advantage to Liverpool with three big chances to Spurs’ two. But it should be noted that the visitors directly played a ‘helping hand’ in most of Reds’ genuinely promising goal-scoring situations with their incredibly sloppy errors and unforced cheap surrendering of possession. This somewhat diminishes Liverpool’s claim for a deserved win, particularly in the light of the way they were technically and tactically outplayed for large parts of the game. Despite this, based on the number of high quality chances registered, it’s not surprising that all of the xG models are making Rodgers’ team deserved winners.

Tactics and performance verdict: DRAW

xG models verdict;
v1 = WIN (2.3 – 1.5)
v2 = WIN (1.95 – 1.17)
v3 = WIN (1.87 – 1.16)

Balo v Spurs

Match #10 – Southampton (A)
Actual result: WIN 2-0

The way both teams shaped-up at the start suggested there was going to be an interesting tactical battle. But Coutinho’s wonder goal in the first few minutes completely changed this, and largely made the Saints’ presumed approach redundant.

Thereafter, it was a case of Koeman’s team lacking the creativity, guile and fluid movement to open the Reds up. Southampton’s best chances in the game happened when they had space to catch Liverpool on the break and also involved a number of debatable refereeing decisions.

Meanwhile, having gone 1-0 up, Rodgers’ team quickly shut up shop and concentrated on making the Saints’ already struggling attack even less of a threat. Koeman’s ineffective second half changes and questionable in-game adaptations continued the theme for the second half. Even when Rodgers inadvertently gave them a huge tactical advantage with his change midway through the second half, the home team couldn’t capitalise and it was easy for the visitors to eventually wrap up the game scoring a second goal.

In a game where it was all about their sheer determination, team work and defensive work-rate, it was unsurprising to see Liverpool create so little going forward. They had five chances with one being a clear-cut (their second goal). Southampton also had one big chance but they missed theirs when already a goal down in the first half. This is one of the rare games when the xG models are portraying a variety of outcomes with none awarding Liverpool the win and one actually giving the win to the Saints.

Tactics and performance verdict: DRAW

xG models verdict;
v1 = LOSS (0.5 – 1.2)
v2 = DRAW (0.59 – 0.65)
v3 = DRAW (0.59 – 1.09)

Match #11 – Man City (H)
Actual result: WIN 2-1

This was a game that was all about both teams being not so good defensively, Liverpool enjoyed a huge midfield advantage but struggled to capitalise on it, while City created the better chances.

Pellegrini’s 4-2-2-2 formation left plenty of gaps for Liverpool to exploit through the middle and down City’s right flank. But for all their supremacy in these zones, Rodgers’ team once again disappointed with their offensive play and chance creation quality, and relied heavily on individual magic to secure the three points.

Stats-wise this was a game full of contradictions. Liverpool had zero big chances from their nine created while Man City registered two clear-cut opportunties (one which was missed at the start of the second half at 1-1 that potentially would have put the visitors ahead). The xG models portray a very close game, despite the difference in the big chances created, with all of them agreeing it was another underserved actual win for Liverpool.

Tactics and performance verdict: DRAW

xG models verdict;
v1 = DRAW (1.4 – 0.9)
v2 = DRAW (1.11 – 0.90)
v3 = DRAW (0.69 – 0.98)

Match #12 – Burnley (H)
Actual result: WIN 2-0

This game was a huge contrast to the meeting at Turf Moor earlier in the run. Here Dyche opted for a cautious tactical approach, looking to sit deep and compact early on, perhaps conserving energy for a late surge (an approach his team often adopted during in this period).

Meanwhile Liverpool showed some interesting tactical features and, thanks to their well-balanced and fluid shape when moving forward, managed to earn a comfortable 2-0 lead before the hour mark. The main highlight of the Reds’ play was the way in which they stretched Burnley on both flanks with plenty of runners from deep (Can, Allen and Henderson), supporting the threat by providing plenty of positional overloads.

Once Rodgers’ team had a two-goal lead they changed their approach to combining dangerous counter-attacks with sitting a bit deeper, being solid enough defensively, and comfortably absorbing any half-hearted attempt from Burnley to get back into the game.

Statistically this was another low-key and professional win for Rodgers’ team. Liverpool were controlled and efficient, creating 16 chances with one big one (missed at 1-0 in the first half). The xG models were in full agreement in giving a ‘clean’ victory to Rodgers’ men.

Tactics and performance verdict: WIN

xG models verdict;
v1 = WIN (1.6 – 0.4)
v2 = WIN (1.21 – 0.21)
v3 = WIN (1.27 – 0.41).

Match #13 – Swansea (A)
Actual result: WIN 1-0

The last match of the run was a text-book game of two halves. The first half saw Swansea being brilliant with their all-round play. The main reason for this was Monk’s decision to start the game with a diamond formation that had the side midfielders sitting behind the play, which provided extra defensive stability and assurance in possession. This enabled his team to both nullify the main threat of Liverpool’s front three while also minimising the possibility of the Reds dominating in the midfield third.

However, in doing all this the home side sacrificed some attacking verve and penetration, so it wasn’t until their midfielders started to roam forward more thet they caused the Reds any troubles. While the Swans didn’t create a succession of big chances, they did have their opportunities and Liverpool had to count on Mignolet once again to keep them in the game.

At half-time Rodgers made some decisive adjustments, introducing yet another previously unused formation – the ‘3-diamond-3’. This was not only suitable for the game context, but its introduction was so ‘out of the blue’ that it caught the home team by surprise and completely turned the game in Liverpool’s favour as Swansea’s first half dominance quickly disappeared. After Liverpool took the lead, Monk’s team fell apart and never really recovered enough to mount any serious fight back. This was expertly used by the Reds to see out the game, with Rodgers’ team showing tactical intelligence and savvy to manage the rest of the game.

If viewed as a whole, the numbers show this game as a deserved (even if hard-fought) win, which is reflected in all three xG models. However, it’s important to note that effectively the game was all about Mignolet keeping Liverpool in it during the first half, before Henderson snatched the winner in bizarre fashion as soon as the Reds had gained an upper hand at the start of the second half. Neither team had a big chance in that period, with the only two that were created coming from Liverpool late in the game after they went a goal up and Swansea tried to push on (which served to boost the Reds xG models ratings).

Tactics and performance verdict: WIN (just about)

xG models verdict;
v1 = WIN (1.9 – 0.6)
v2 = WIN (2.03 – 0.39)
v3 = WIN (1.67 – 0.51)

Here is all the above information summarized in a table-format for simple reading and comparison (click on the image to see in full size):



What this case study revealed was that more often than not tactics and statistics go hand-to-hand in how they appraise the games on an individual basis. From these 13 games, seven saw the same verdict from both types of analysis. In other cases, slight numerical disparity caused the overall difference in the verdict (for example the Arsenal home, Burnley away, Villa home, and Southampton away games). The Everton and Swansea away games saw a bigger disagreement between how both sides saw the games.

They also arguably showed the flaw of statistical analysis in general and the xG models in particular. They will assess games as one whole, without considering the specific contexts and so not appraising the game in periods based on what happened, and especially how and why it happened, and the way the game played itself out. Which is why for greater objectivity and assessment the statistical review needs the help of tactical analysis in the same way the tactical analysis benefits from the stats and chalkboards providing greater depth and ability to explain by contextualising and visualising things better.

Anyway, in terms of total points attributed it’s curious that while very close to each other, the three xG models all reached different conclusions for this run of 13 games. They had complete agreement for 11 games, which explains the really minor difference in the total points. Of these 11 games, seven agreed and four disagreed with the actual result. From these four games Liverpool were +/- 0 in terms of points total, which is a fascinating classic example of different calculations, same result.

What is also interesting is that neither of the models declared Liverpool deserved the full 33 points they actually acquired from these games. Two of the models are close to how the tactics and performance analysis saw the games, with a point or two difference easily explained given the technicalities around the numerical ratings of each model.

All of this is not intended to suggest that Liverpool were lucky to win 33 points or stay unbeaten in that run of 13 games. It’s intended as an exercise to look at how using combined detailed analysis means the way such runs of results, and how their media and fan narratives portray them, can be seen somewhat differently.

Fans and media could perhaps be excused if their growing excitement as the run developed clouded their judgement or views, especially after a bad period like that seen between September and mid-December. But for the professionals working in football clubs, who are actually responsible for objectivity and constantly seeking ways to progress and avoid stagnation, to be sharing a belief similar to the one of the media and fans narrative should be seen as inappropriate given the associated risks of possibly overlooking hidden or subtle potential causes for future regressions, even while the actual results are fine in the moment. Such valuable information is often invisible on the surface at first glance. This is why extensive analysis combining tactics and statistics is needed to dig deeper and then combine the pieces of a puzzle that may help you to anticipate and prepare for the next storm when it arrives.

Based on the results in 2014/15 the question could be asked as to whether such an analysis was carried out and lessons learned from the previous 2013/14 season. Given the way that Liverpool’s peak spell in that season was the 11-game winning run, which actually involved a high portion of good fortune and a large dependency on set-plays (two factors that cannot be relied on), there was a risk that it set the scene for Rodgers’ team to be seen as an unstoppable force being lavished with ‘poetry in motion’ praise. 2014/15’s 13 game unbeaten run gives all concerned a second chance, and another opportunity to show lessons are being learnt and suitable actions taken to ensure similar mistakes and underperformance won’t be repeated in the future.

Perfection is clearly a utopian state and something that never really exists beyond the world of dreams and ideal theoretical scenarios. But at the same time in practical terms a ‘perfection’ can be achieved, though when the word is used to describe the never-ending direction of improvement rather than a final destination and resting place.

Liverpool don’t need to be perfect per se, but they need to be aspiring and trying their best to be identifying and eliminating faults or flaws. Now they are in the role of the underdog, preventing periods of weaknesses and breakdowns is arguably an essential path for Liverpool to take. It is one of the few ways for them to chase and surprise their competitors who may possess greater resources, but also be more prone to complacency and losing focus.