Crystal Palace: Season Preview 2021/22

First let’s take a moment to salute Roy Hodgson. From Palace youth player to retiring as Palace manager by way of a successful 45 year multi-continent managerial career, Hodgson has been a titan of the game and he will be missed, not least from the post-game interview circuit, on which he was ever good value. In truth, he leaves Palace at a good time for his legacy, as results have skewed ahead of expectation for a couple of seasons now and we can but speculate how much of a factor his veteran savvy was when placed next to landing on the right side of some statistical variance. What statistical variance would that be? That’s the expected goals for Hodgson’s (nearly) four seasons in charge and here are the seasonal outcomes: [table id=104 /] If i’m Roy Hodgson i’m pointing at the league table. Last year I wrote that ahead of the season were Palace given the option to take 17th, they would be wise to accept. The trending in metrics season to season was bad enough that it was a valid suggestion. The team had just scored 31 goals total in a season and seen their expected numbers crater by nearly 20 goals. In reality, Palace were never really in danger of relegation–by early March they had 30+ points and were ten points clear of 18th place–while their metrics got worse, in particular further declining as they coasted in to port late on. What was bad about 2020-21 Crystal Palace? Firstly, the defence conceded a lot of goals–66 was more than everyone bar Southampton and West Brom. Now that’s a top line red flag that doesn’t improve when peering beneath the surface, Vicente Guaita did have a less successful 2020-21 compared to 2019-20, but he basically kept goals out at expectation instead of buying his team an extra eight goals as he had done before. At the other end of the pitch, finally Christian Benteke scored ahead of xG (10 goals vs 8 xG) for the first time in a number of seasons and was backed up by ahead-of-xG finishing from Eberechi Eze (4 from 2) and Wilfred Zaha too (9 from 6). In attack, finally nobody bombed out. As we know, this kind of overperformance is nice to have, but impossible to rely on and technically, only Sheffield United’s attack was worse. Enter Patrick Vieira, but which one? The Patrick Vieira that managed New York City FC initially brought a possession game to the team before transitioning towards a high energy outfit stocked with younger talents, and was relatively successful, insofar as his team competed at the top of their Conference. Upon accepting the challenge to move to Nice in 2018, it could perhaps be presumed that Vieira might pursue a similar style, but what transpired wasn’t quite the same. Nice weren’t unsuccessful with Vieira at the helm but at least in his first season, 2018-19, goals were super hard to come by (they scored 30, one fewer than Palace in 2019-20) and finished seventh after a raft of low scoring games. At this point the style of play was somewhat disjointed, still prioritising a passing game from the back, but not necessarily very effective with it. Nice were a young team though and pointed upwards the following year in finishing fifth, albeit with warning signs. A season long ballooning over their non-penalty expected goal difference, which remained a steady -0.18 per game across both seasons helped but they remained a possession heavy (58% in 2018-19, 55% in 2019-20) unit with no real commitment to any kind of press and with very little functional cutting edge. His eventual dismissal in December 2020 was the result of a succession of stepped-on rakes in the form of five straight defeats in all competitions, four of which were at home, but actually coincided with metrics finally pointing upwards: There isn’t a ton to get excited about here, but sub-par metrics plus above par outcomes does speak quite loudly as to Hodgson’s last two seasons, albeit via a structurally different plan. Vieira is coming in here taking over a team that has struggled recently, and it does feel a risk if he looks to deploy a possession first outlook, mainly due to the transition that will be required in this Crystal Palace team. In recent seasons, it has escaped nobody’s attention that Palace have been an aging team and their 2020 transfer window started address that with the signings of Eberechi Eze and Nathan Ferguson. Unfortunately Ferguson missed the entire season injured while a steady opening season from Eze was  suffixed by an achilles injury which is thought to see him likely to miss a good portion of 2021-22. Not part of the plan, we go again! Weirdly my preview last season included a section on Conor Gallagher which detailed how his split season across Charlton and Swansea in 2019-20 saw different aspects of his play come to the fore, and was a useful instruction to factoring in team styles into player evaluation. He then rocked up at West Brom and found similar divergence playing first for Slaven Bilić and then for Big Sam Allardyce, each of which meant that his statistical profile once more varied: It paints a mixed picture, but is pretty positive for a young player in their first season in the league on a relegated team. The guy who logged 83% pass completion for Bilić will fit in nicely to Vieira’s plans, the 75% guy who played for Allardyce perhaps less so. Elsewhere, the proverbial chequebook has come out for some promising talent in Reading’s Michael Olise and Chelsea’s Marc Guéhi, who impressed on loan at Swansea last season and is likely ready for a step up in league. Both these signings look ideal, talented young players ready for the next step in their career, but the caution is in the entire squad profile. Palace’s rebuild appears to have come a little late, and as such the squad skews away from the age group it likely needs to–prime. The one signing that sits bang on the line here is that of Joachim Anderson, another player who spent 2020-21 on loan at a relegated side (this time Fulham) while performing not without credit. He was linked with Tottenham earlier in the summer and can viably be considered a coup for Palace. Guéhi and Anderson likely slot right in at left and right centre back especially give that the 1-2-3 2020-21 depth at left centre back (Gary Cahill, Scott Dann and Mamadou Sakho) have all departed and right centre back was mainly manned by Cheikhou Kouyaté, who may be a centre back now but certainly wasn’t always. A slew of contracts ended for Palace in the summer of 2021 and in particular, departures from long term veterans and regular performers such as Andros Townsend and Patrick van Aanholt mean the squad currently looks light overall, with a particular deficiency in midfield. Of course the transfer window is open a while yet so remedial action may be taken, but it appears likely that Palace will start the season without a strong bench. It’s possible that Palace are waiting for the music to stop elsewhere and planning to offer regular playing time to some of the top clubs less wanted stars, and that’s certainly a strategy that may bear fruit. Projection Squad turnover and thinness, a lack of prime age players, a new manager likely to change the style of play and long term sub-par metrics all point at risk factors ahead of 2021-22. Palace have rightly been praised for some of their moves in this window for addressing their long term future, but they still need to stay in this league in the present to benefit from those moves and that may prove tough. Last season’s projection was thus: The trending is wrong and it will take distinct and real change to the team’s metrics to give them a chance of steering clear of a relegation battle. Hodgson has been in the game long enough to see the warning signs and in most Premier League seasons there are double the amount of teams that perform at a level to put themselves into the relegation mix, so he may like the chances of winning a coin flip. But planning and execution of strategy are what limit risks such as this, and having been reactive rather than proactive in turning over their squad it could well be tough for the club. Offered seventeenth today, you’d have to take it. In truth very similar remarks apply now. The bookmakers have Palace, Brentford, Norwich, Watford, Burnley and Newcastle in the mix for sub-40 points at this juncture, meaning once more a likely six teams for three relegation spots, the same either/or dynamic as last season. This end of the table is a new challenge for Vieira and he will need all his man management skills to get the best from this squad, week in, week out and keep them competitive. Offered seventeenth today, they should take it.


Want to read about another team? The rest of our Premier League season previews can be found here If you’re a club, media or gambling entity and want to know more about what StatsBomb can do for you, please contact us at Sales@StatsBomb.com We also provide education in this area, so if this taste of football analytics sparked interest, check out our Introduction to Football Analytics course Follow us on Twitter in English and Spanish and also on LinkedIn

Chelsea: Season Preview 2021/22

Where do you go from here? In 2012, Chelsea sacked their concept manager, Andre Villas-Boas, handed the reigns to a club legend, Roberto Di Matteo, finished miles off the pace with 64 points then promptly went out and won the Champions League. In 2021, Chelsea sacked their club legend manager, Frank Lampard, handed the reigns to a concept manager, Thomas Tuchel, finished miles off the pace with 67 points then promptly went out and won the Champions League. It took Tuchel 123 days to hit that high point and he did it by focusing first and foremost on defence. In 19 Premier League games under Tuchel, Chelsea gave up more than one expected goal twice: firstly against West Brom in the freakish 5-2 defeat and latterly against Manchester City, two games from the end of the season in a match they won regardless. The difference that Tuchel made to the team was as stark as it was swift and they were clear second best behind Manchester City in the back half of 2020-21. In their first ten game phase under Tuchel the team closed everything right down. The attack was yet to gel together but the defence  gave up a remarkable four xG and conceded just two goals. By season’s end Chelsea’s Tuchel era metrics were in hailing distance of Manchester City with a +1 xG difference per game across his reign (for the same period City were +1.2) with the impressive side still mainly defence (0.56 xG against) while at least in terms of process, the attack had perked up and landed at a decent 1.56 xG per game. Only two teams have put up metrics of this ilk in recent seasons: Liverpool and Manchester City. It may be a short spell, but the strong indicator here is that Tuchel has managed to organise Chelsea’s high quality, high potential squad into a shape that is capable of contending. Lampard had the makings but not the execution, Tuchel had both. So what did he do to harness the talent in this squad? First up the obvious switch from Lampard’s general 4-3-3 and variations towards a distinct 3-4-3. This had the instant benefit that if you were a centre back, then welcome aboard, you now had a great chance of getting good minutes regardless of what had been happening so far in the season (hello Antonio Rüdiger, Andreas Christensen), ditto anyone who had experience of playing wing-back, let’s say in a league winning side playing a similar formation (come on down Marcos Alonso) and anyone called Mason Mount, mainly because he’s literally the ideal player for any team playing any formation, something on which both Lampard and Tuchel agreed, more later. Here’s the minutes splits across the squad: Chelsea had a relatively injury-light 2020-21 (and the squad to survive it too) so the balance between the Lampardis and Tuchelistas is fairly reliable to spot. Each manager loved Mount, Edouard Mendy, Timo Werner and Reece James–these were a core four and come rain or shine, their managers picked them. More Lampard favoured were Ben Chilwell (a bit), N’Golo Kanté, Kurt Zouma and Tammy Abraham while Tuchel was big on César Azpilicueta, Jorginho, Rüdiger, Christensen and Alonso. Kanté and Chilwell get a pass here too really as once the business end of the Champions League kicked in they were solid picks in that tournament, as was Kai Havertz. The risk players for this squad are quite clear: Abraham was losing minutes to Olivier Giroud, who has now left and has likely seen his potential position gazumped entirely by Romelu Lukaku (a transfer for him seems imminent too). Hakim Ziyech struggled to see pitch time for either manager last season and at the moment looks like the big miss/square peg among the high class attackers purchased in summer 2020, despite quite good statistical contribution when on the pitch. Christian Pulisic saw more game time at the back end of the season, enough to think he’s likely to stick around and feature, while Callum Hudson-Odoi remains a statistical marvel but repeatedly flits in and out of favour whoever is in charge. Seriously, Hudson-Odoi is fascinating. Snooping around earlier in the summer I spied this and tweeted it: More than 1000 minutes, more than 2.4 OPEN PLAY key passes per 90? 3 qualifiers in the PL, De Bruyne, Grealish and… Hudson-Odoi Then when I was poking at our new OBV model numbers for this article, he leapt off the page again, albeit predominantly in the Lampard era where he was clocking 0.56 OBV/90. This was right around better players in the league, albeit in pretty small minutes, which cautions against excitement, especially when you have The Real Deal Mason Mount in the mix:

People sometimes like to argue about model outputs, but I’m not interested in that here, as Mason Mount is constantly selected and great. In ways he reminds me of Italian midfielders of the late 1990s, an Angelo Di Livio or Massimo Ambrosini or Alessio Tacchinardi, players that were surrounded by flair and technical brilliance yet absolute team soldiers and vital cogs. That possibly denigrates everyone I’m talking about here, particularly Mount, who can contribute well at the business end of the pitch too, yet it’s really not meant to, Mount himself has had to semi-justify his position in the Chelsea and England team for the best part of two years, but Lampard knew his worth, and perhaps saw something of himself in him, succeeding through will and graft where more naturally gifted rivals faltered. He started him in 49/57 league games for Chelsea and only missed pitch time in two. Tuchel is under no illusions either, in May saying “He is crucial for our game, he is an absolutely key player” and the fans voted him Chelsea Player of the Year last season. And he’s still just 22! Unfortunately fellow midfield man Jorginho took the Champions League / Euros double, to hit his own high note, but let’s move beyond that for now and celebrate Mount, who also successfully boosted all of his expected metrics moving from Lampard’s team to  Tuchel:   Sorry, a sideline. So what did Tuchel do? We established that the defence improved, and it did so in the most desirable way: they reduced the volume of shots they were giving up and the shots they did give up were harder to score than the ones they had been giving up before, magic! Check out the annotation, one shot! For the record a Christian Benteke header, you can maybe understand how the scout report may have missed on that as Benteke’s long game of undershooting xG for multiple seasons paid off with decent goal returns in 2020-21. Overall, reducing the volume and quantity of shots allowed came across all facets of the game. They halved the volume of “clear” (just the keeper to beat) shots they allowed, reduced opposition set piece effectiveness and gave up fewer counter attacking shots. Part of how they got there was that Chelsea upped their tempo out of possession. Under Lampard they were fairly active and ranked fourth to sixth for all of PPDA (Passes Per Defensive Action), the average height of their defensive actions and their overall responsiveness to pressuring opposition ball receipts (our “aggression” metric). It took a little time under Tuchel–recall a handful of slower, possession heavy chance-lite fixtures in the early weeks–but during the second half of the season, they were second for PPDA (behind Leeds, of course), third for height of defensive actions (behind Man City and a resurgent Liverpool) and second for aggression (behind Leeds again), all notable moves in a more active direction. Just for raw pressure events and defensive actions, the workload went up, up, up before coming down, down, down just as they were all in on the Champions League, but we can see here more events, and a higher proportion of pressure events as Tuchel’s tenure settled in. They have the makings to play this way now and be effective with it: Some of the methodology also involved ball-hogging dominating possession: they played more passes than Manchester City in their 2-1 win and routinely outpassed weaker teams by large volumes. But they could win in other ways too, as a 1-0 victory over Liverpool showed. They played a fairly rigid and deep 3-4-[big gap]-3 and outworked their opponents while giving up next to nothing shot (7) or xG (0.28) wise. On the largest stage, the Champions League final was not dissimilar. How do you improve a Champions League winning team that has shown it can outwork, out pass, out shoot and out defend its opponents, seemingly at will? Chelsea are eternally in “win-now” mode and “win-now” mode means buying Serie A’s best striker for close to £100m to End Instantly debates around how you can man your front three, whether a false 9 will work best or if your very-good-but-to-my-mind-not-quite-good-enough-to-lead-a-Chelsea-front-line-perhaps-a-tier-lower-23-year-old-academy-graduate is the right choice. And why have they done this? Finishing. The divergence between Chelsea’s stellar Tuchel era metrics and the reality of clambering into the top four slots can be blamed on one thing: the attack fired well under expected values under Tuchel. Analytics 101 remains get the process right and let the variance work itself out. In truth this was a team effort. Thirteen Chelsea players took ten or more shots in the Premier League under Tuchel and of those thirteen, a meagre two of them exceeded their goal expectation: Kurt Zouma (one goal from 13 shots and 0.85 xG) and Marcos Alonso (two goals from 20 shots and 1.62 xG). This is frankly ridiculous, and not really a reflection of the quality of Chelsea’s players. Werner and Havertz may have had personally less impressive seasons than they may have hoped but they’re not obviously sub-par finishers, nor historically was Olivier Giroud. One specific aspect of play they did not get from their forward corps was as follows: Yes; a guy crashing the six yard box. It’s not that Lukaku is a noted plus finisher, more that he is a reliable scorer–he will take up the positions you need from your main striker and score goals. To this regard Lukaku does have the vibe of a missing piece and seven years on from departing for a decent fee is returning. It will be hoped that as a focal point, Lukaku can enable his equally talented colleagues to thrive around him, and regardless he’s surely as close to guaranteed as any Chelsea striker to chip in handily. More broadly on personnel, while clued up fans may bemoan the loss of Abraham and Fikayo Tomori, they’re getting very good fees for these players and the supply line from their youth teams to the first team is doing just great, with Mount, Christensen and James all hitting big recently. Projection There’s a disconnect between stellar Tuchel-era metrics and winning the Champions League contrasting with 2020-21 points accrual (76 point pace) and bookmaker expectations (around 76-78 points). This plants Chelsea firmly alongside Liverpool in a “best-of-the-rest” category, a couple of clips ahead of Manchester United yet trailing Manchester City. In real terms they will likely need to bounce forward another ten points on top of that and make a +20 points season jump to season to contend for the title. With a summer of planning for Tuchel, a stable squad that isn’t undergoing much remedial work and the natural belief of a successful team, it’s possibly easier imagining Chelsea steering upwards once more than not. With Pep Guardiola, Jürgen Klopp and now Tuchel in the Premier League mix, it would be thrilling to see these three teams battling hard on all fronts and this season is well set for these coaching titans to tee off. Evidently Tuchel has not found long term stability in his managerial career, but with this squad and the big prize already in his back pocket, he may be primed to stay a good while yet. The league is better for that.


Want to read about another team? The rest of our Premier League season previews can be found here If you’re a club, media or gambling entity and want to know more about what StatsBomb can do for you, please contact us at Sales@StatsBomb.com We also provide education in this area, so if this taste of football analytics sparked interest, check out our Introduction to Football Analytics course Follow us on Twitter in English and Spanish and also on LinkedIn

Tottenham Hotspur: Season Preview 2021/22

It’s the end of an era over at Tottenham, but which era? When Mauricio Pochettino left the club in November 2019, a Champions League final was fresh in the memory yet team metrics had declined precipitously. A week before he left he commented thus:

“We are in the process to [re]build and we will see if we have the time to build what we want”

With a team largely powered by a core of players remaining from Pochettino’s era, a season and a half of Jose Mourinho followed. Talks of rebuilding continued to surround the team but despite vigorous work in the 2020 transfer window, the innate core of the team remained similar to before, and results and metrics continued to plateau. When Mourinho himself was jettisoned in April 2021, it came as no real surprise. In particular, Tottenham waned against the better teams in the league, and a 4-3-9 WDL record against teams in the top half under Mourinho in 2020-21 simply didn’t cut it. When your entire schtick as a manager is that you’re a winner, you really need to win games at a faster clip than that. Now Mourinho certainly chose a reactive style of play but we can see here that he failed to fundamentally improve the team’s metrics to even the level of Pochettino’s declined 2018-19 outfit: With Mourinho’s departure entirely justifiable, an extended recruitment process eventually landed on Nuno Espírito Santo, fresh away from Wolves after a tough season saw them part ways. What Espírito Santo will bring to the team is somewhat open to interpretation; his Wolves teams often played with apparent caution in a 3-5-2 and he had a small squad. Tottenham also saw time with three centre backs during Mourinho’s time, and played in a similarly reactive style, but are awash with talented attackers, to the extent that it seems feasible that Espírito Santo may deploy a more offensively charged system, much as he has in former jobs. Indeed, pre-season has seen more of a 4-3-3. But who will form the core of the team? That question is harder to answer and the arrival of Juventus’ former Chief Football Officer Fabio Paratici has accelerated early summer transfer activity both in and out of the club; one thing is for sure now: it is rebuild time. The good news is that two of the team’s main attacking stars appear likely to be retained for 2021-22 albeit under slightly different circumstances. Son Heung-Min had another great season and recently signed a new four year contract. His shot map shows quite clearly what Mourinho tried to get from him: open shots off throughballs, and he exceeded his expected goals by a decent margin for the fifth straight season: There’s an interesting question around Son’s new contract–he’s 29 years old, somewhat of a speed merchant and signed up until he’s 33. The next four years may not be his best four years, but his elite finishing ability may keep him as a significant net plus contributor ahead of what may normally be expected from a player with his style. Balancing these decisions is never easy–see Liverpool’s trio of late-peak forwards as a case in point–but at the very top of the sport, I suspect a player as an asset to the team persists longer than general age curve work suggests. Similar comments could apply to Harry Kane, who has never been a speed guy, but has always been a plus finisher. However, there’s enough noise in the room to suggest that Kane is more inclined to finish his footballing days outside of North London than within, albeit his contract situation (3 more years), likely asking price (a lot) and Manchester City already dumping £100m on Jack Grealish all look like reasons that may preclude his departure, at least this summer. That said, those of us who recall the departure of Gareth Bale back in 2013 will not be wagering on any specific outcome. The big analytics story on Kane is that in 2020-21, he was very much back and there will always be the suspicion that the enforced pandemic break was the best thing that could happen to him. Prior to that a series of injuries and quick returns had apparently derailed his overall effectiveness and seen his underlying metrics drop off, albeit without the goals ever really disappearing.   Mourinho certainly set up teams to get the best out of Kane, and 2020-21 saw his highest shots per 90 (3.6), xG per 90 (0.48) rates since 2017-18 and the added boost of high volume creation, in particular towards forward partner Son. From a metric perspective, it’s fascinating to note that essentially what Kane is doing on the pitch is broadly the same as before–apart from the key aspects of well, shooting and creating shots. Back when he declined from nearly five shots per game in 2017-18 to around half that in early 2019-20, I attempted to investigate what he was doing differently and found scant evidence outside the shot decline, his average touch was slightly deeper, but little more. It seems that Kane is resilient enough to endeavour to play a similar game to that which he has found great success with, but how or if a manager uses him as a focal point within the team may well define outcomes. Mourinho recognised that and built a team to get the best from Kane, and he got that. Did other players suffer by way of comparison? Perhaps? Either way, this season will be informative once more. The under-the-radar story for this team is the second departure of Gareth Bale. In a fitful season, when he was on the pitch (around one third of available minutes) he was lights out good. He scored eleven goals at around double his xG, added a couple of assists and was the only player in the league with above one goal contribution per 90. For shot contribution, only Kevin De Bruyne (6.7) and Bruno Fernandes (5.5) exceeded his 5.2 per 90. He also ranked second in the division behind Grealish in our OBV/90 metric. Any way you slice the Bale pie, it came up tasty. Sure he wasn’t a key starter and in the aggregate trailed Kane and Son, but having this kind of weapon as a bench option will be missed. Personnel So far the rebuild has been as follows: 1. New goalkeeper Pierluigi Gollini, initially on loan from Atalanta. This makes some sense, as Joe Hart was the back-up to Hugo Lloris and with Lloris entering the last year of his contract, a degree of succession would be logical, possibly Gollini represents that. 2. Left footed left winger 20 year old Bryan Gil swapped for a chunk of cash plus club legend Erik Lamela with Sevilla. Lamela at 29 was good to go, and this kind of one for one “past out, future in” equation is nice and tidy from a squad management perspective. Dissecting a stats case for Gil is curious as playing for a stylistically distinct and struggling Eibar team in 2020-21 wasn’t the passport to ballooning metrics as we can see here: Okay, so what’s the story here? Much of what we see here grades out at around league average, and the combination of team style and his age enables us to be perhaps more positive than may seem at first glance. If you’re 20 years old and clocking league average metrics, the prospect of what you can do down the line is pretty positive. People who watch football, which couldn’t be me, also appeared happy with this deal. 3. Probably Cristian Romero No Tottenham fan is likely to complain about defensive reinforcements and the departure of Toby Alderweireld makes that a pertinent detail. Given how readily media talk around moves has turned into reality this summer, it very much appears that the Argentinian international centre back Cristian Romero is likely to arrive shortly from Atalanta via Juventus. Fresh from being crowned Serie A Defender of the Year and a Copa America triumph with his national team, it’s easy to see why Romero is coveted and could attract a large fee. A twin effect jumps out from our metrics: during both his time at Atalanta and Genoa he has recorded extremely high pressure volumes and extremely high foul volumes. The centre back archetype in the modern Premier League is somewhat opposite to this, for example Virgil van Dijk pressures and fouls infrequently, and it’s a logical detail. If you enact pressure, you may well get beaten and leave space behind and be forced to foul. Atalanta routinely play high up the pitch and need robust and active defenders, Genoa in 2019-20 less so, but Romero was notably active in both teams. How Romero slots into Espírito Santo’s presumably more cautious set-ups and how active he is will be a story to follow, as the handful of Argentinian Copa America games are less conclusive here. What else occurs personnel wise is hard to know. At Juventus Paratici frequently oversaw large scale squad turnover on a season to season basis, and having already sold Alderweireld to Al-Duhail in Qatar (as a natural replacement for Mehdi Benatia who he sold there in 2018 from Juventus), it’s clear he is well capable of finding clubs where others may not. However should Romero follow Gil in though the door, that would be around £75m invested in the squad, and it’s hard to see the club allocating significantly more resources for player purchases without some balancing of the books. We shall see. Projection In 2016-17 Tottenham finished second, the following year they finished third, then fourth, then sixth and in 2020-21 seventh. With Leicester apparently well in the mix for top four places these days, the big six appear to have either shrunk to four (Man City, Liverpool, Chelsea, Man Utd) or extended out to seven (add in Arsenal). Either way, Tottenham’s desire is to be firmly in the mix with the smaller group, and the Mourinho experiment did not slow a decline that was straightforward to forecast via metrics even as they finished fourth in 2018-19. As such, bookmaker predictions have them in that second tier alongside Leicester and Arsenal and an estimation of around 60 points doesn’t deviate far from either of the last two season totals (59, 62). None of this is easy to counter in either direction. The squad remains fairly deep and talented while appearing to lack some of the cohesion and clear ethos that we see from the league’s best sides. Last season’s top four will be a hard nut to crack for all teams this year and a deal of hope has to surround further recruits and whether Espírito Santo can take a more talented squad than he had at Wolves (two seventh placed finishes) and land them higher up the table. Fifth to seventh remains by far the most likely outcome and anything over 65 points would represent a step in the right direction. Rome wasn’t built in a day, nor just one transfer window.


Want to read about another team? The rest of our Premier League season previews can be found here If you’re a club, media or gambling entity and want to know more about what StatsBomb can do for you, please contact us at Sales@StatsBomb.com We also provide education in this area, so if this taste of football analytics sparked interest, check out our Introduction to Football Analytics course Follow us on Twitter in English and Spanish and also on LinkedIn

West Ham United: Season Preview 2021/22

To move West Ham from 16th place in 2019-20 to 6th in 2020-21 has to rank alongside any of David Moyes’ career achievements. His return to the club 18 months after leaving in the summer of 2018 was not greeted with great positivity, and it appeared to take the enforced pandemic shutdown to get the ship steering in the right direction. However, few teams can be seen to have benefited so greatly from playing in the current climate, and the way the club has evolved and improved in Moyes’ tenure is a ringing endorsement towards the elusive attribute every manager would like–time. That 26 point season-on-season improvement ranks third behind Leicester 2015-16 and Chelsea 2016-17 across the Premier League era: It was no fluke either and is reflected in the shot metrics that powered the team:

  • non-penalty xG rose from 1.14 to 1.34 per game
  • non-penalty xG conceded declined from 1.49 to 1.17 per game
  • …causing xG to go from -0.35 per game to +0.17; half a goal gains are not to be undervalued, this team improved substantially
  • they took more clear shots (1.5 up to 1.9 per game)
  • they conceded fewer clear shots (2.6 down to 1.9 per game)
  • the chopped a load off their set piece shot concessions (down from 3.4 to 2.6 per game, xG here halved from 0.31 to 0.17 per game)

…while stylistic play features were clear too:

  • 32% of box entries were via a cross (most in league)
  • led the league for metrics relating to pace and directness
  • second lowest passes per defensive action
  • least counterpressures and resultant regains
  • closest shot distance (both non-penalty and open play)

This all paints a picture of an organised, somewhat reactive and patient team. They also knew which games represented their best chance of success. Eventual top four? A 0-1-7 WDL record. Rest of the league?  A stellar19-7-4 record. We discussed the role of long carries and shots and goals from defensive regains in last year’s preview and these features generally persisted. West Ham kept their shape and waited for opportunities to break forward. Once they did they created chances at a decent volume for their key contributors, once more Michail Antonio benefited but also a resurgent Jesse Lingard: The boost that Lingard brought to the back half of West Ham’s campaign was both large and perhaps unexpected. Lingard, a player who has been noted to steer under his expected goals values in the past, scored eight non-penalty goals from an xG of around four to force his way all the way back into the England squad.It appears likely that he will be staying in Manchester this season and for West Ham losing not just the player but one in the form of his life is a significant blow. As yet, the Hammers have been slow to enter the transfer market, with the main change from the start of last season the removal of their highly priced bench men. Sebastian Haller left for Ajax in the winter and Felipe Anderson was practically given away in a cheap deal to take him to Lazio. Neither convinced for West Ham in the aggregate and quickly fell down the pecking order under Moyes, but their removal, alongside that of Lingard, does leave something of a lack of depth in attacking positions. In general Moyes operated a small squad in 2020-21, and he talked about how he instilled an effective group ethos during difficult times in a column for the Times ahead of the Euros:

“At West Ham, during the club season, I made a conscious effort to make training more fun…to not keep it so heavy. I thought about the world the players were having to live in, there was no opportunity for them to leave the bubble (so) we had to make sure they came into work and felt they were with their friends. Footballers were challenged by lockdowns and the lack of “normal” life and found they were happy to get out of the house, pleased to be in training, and craved it from the point of view of structure and routine. So there was actually more training — but we made sure that many days there were lighter training loads.  (In general) unity and spirit helped “smaller” clubs navigate the difficult conditions and do well in many of the domestic leagues.”

What has been ahead of and during this summer done is limited; Craig Dawson’s loan was converted to a permanent deal back in April and Alphonse Areola recently arrived on loan from PSG after a solid season in London at Fulham: With Fabiański now 36 years old, it makes sense to start thinking about succession and we will see who Moyes puts his faith in from week one. Areola is a more active and “modern” goalkeeper than Fabiański, but the Polish veteran has done little wrong in his West Ham tenure and may well see the position still as his own. Warning Signs? One of the hallmarks of Moyes’ second spell has been his consistency in selection. In 2020-21, eight players played in more than 70% of the team’s minutes and the core group he has relied upon is small. When Antonio missed time in the autumn of 2020, Haller stepped in, but there are a further string of players who look core–perhaps Declan Rice, Aaron Cresswell, Tomáš Souček and Vladimír Coufal. The team avoided injury problems last season and that consistency in selection looked to work to West Ham’s benefit. The main selection choices week to week look likely to be in the attacking midfield band. Jarrod Bowen, Manuel Lanzini, Pablo Fornals are all in the mix but could Saïd Benrahma’s second season see better returns and omit the need to replace Lingard in the market? He saw a lot of bench time in Premier League season one, and the variation of his profile compared to his time at Brentford was stark. Gone was the shot-happy drifting left-sided attacker, replaced by a more shot-shy, creative player fitting in across the attacking band. He was in competition with Pablo Fornals for a left-sided slot then Lingard became the key man in the centre and Benrahma never quite nailed down the starting space. It would be no surprise if Moyes values positional discipline and Benrahma doesn’t quite offer exactly what he wants from his wide attackers, but he’s certainly talented enough to come forward and contribute more, if empowered. Setting realistic expectations for West Ham’s season means without further investment, a replication of anything approaching last season’s 65 points would represent the absolute best outcome this group could expect, and doing so, even from similarly small net positive metrics is fairly unlikely. There’s also a hidden drift in the end of season form to note: Throughout Moyes’ tenure, West Ham’s xG and goal differences have tracked fairly reliably. This is no given, and we can see divergence at the end of 2020-21, in which goal difference persisted significantly ahead of xG. Now, that was great to keep the results coming as the summer beckoned, but is less encouraging ahead of the new season; West Ham were at their worst in 2020-21 at season’s end. Where will West Ham end up? Bookmakers are currently projecting a firm mid-table finish and around 50 points and while that may seem slightly ungenerous, it’s not hard to toss around their metrics, quietness in the transfer market and risk profile towards squad depth versus key injuries and feel that is in range. The coming month and transfers could move the needle here though and I’d probably have them good for a couple more wins than that, particularly if they acquire a useful back-up or alternative to Antonio. 2020-21 was a season in which the unique circumstances of low pressing pandemic-ball and a stable first team helped Moyes extract the best from a group of players that played with a clear identity. With crowds back in stadiums, it’s hard to imagine that the slightly neutral aspect of 2020-21 will persist and some of the benefits accrued from a more passive strategic style may well decline. That does not mean that West Ham’s future prognosis is particularly worrying, just that a small detour down into mid-table is the most likely outcome. This would still represent a huge overall improvement on what Moyes inherited and justify his retention, with a view to continuing to build and sustain within the league comfortably. To move upwards or match 2020-21 will probably take both more time and significant investment, and having been bitten by high-priced flops such as Haller and Anderson, West Ham’s ownership may be content to swim in shallower transfer waters for now too. Last season I felt that transfer backing could be the difference between lower mid-table security and something slightly more than that, and the equation is probably similar. Again West Ham have a choice to make, but this time they are making it from a higher vantage point, that of safety.


Want to read about another team? The rest of our Premier League season previews can be found here If you’re a club, media or gambling entity and want to know more about what StatsBomb can do for you, please contact us at Sales@StatsBomb.com We also provide education in this area, so if this taste of football analytics sparked interest, check out our Introduction to Football Analytics course Follow us on Twitter in English and Spanish and also on LinkedIn

The StatsBomb Premier League Season Previews 2020-21

We hope you’ve enjoyed the StatsBomb Premier League season previews! Here’s a handy place to keep them all. Just click the links to read about each team. Thanks to Nick Dorrington, Oli Walker and Ted Knutson for writing and thanks from me to you for reading. If you enjoyed these articles do share widely! Arsenal Aston Villa Brighton and Hove Albion Burnley Chelsea Crystal Palace Everton Fulham Leeds United Leicester City Liverpool Manchester City Manchester United Newcastle United Sheffield United Southampton Tottenham Hotspur West Bromwich Albion West Ham United Wolverhampton Wanderers  


  If you’re a club, media or gambling entity and want to know more about what StatsBomb can do for you, please contact us at Sales@StatsBomb.com We also provide education in this area, so if this taste of football analytics sparked interest, check out our Introduction to Football Analytics course Follow us on twitter in English and Spanish and also on LinkedIn

Pass Footedness in the Premier League

One of the many enhanced features of StatsBomb Data is pass footedness. Today we’re going to look at some results from looking at this feature of the data and show how pass footedness manifests within the Premier League. Footedness has historically been a feature that has been added to shot data and not passes, but shot volumes are small, and genuine clues around footedness can be concealed by player style or availability or opportunity; for example a centre back who finds the ball at their feet in their opponents six yard box will likely swing whatever boot is nearest to generate a shot. Back in their own half comfortable in possession, they are more likely to make a conscious choice as to which foot they pass with. Already we have a more detailed cut to look at player preferences.

Right: since this is new, let’s get some simple facts out of the way.

  • There are an average of 870 passes in a 2018-19 Premier League game.
  • Of those passes, around two thirds are made with a player’s right foot.
  • Of the players, around three quarters are dominantly right footed.
  • Right footed players average around 87% of their passes with their dominant foot, left footed players around 85%.
  • On average, right footed passes are one metre longer than left footed passes.
  • Right footed passes are completed around 81% of the time, left footed passes 78.5%. This trend holds regardless of the foot dominance of players.
  • The difference between left footed players completion rates for their left foot (78.3%) and right foot (81.2%) is slightly larger than that of right footed players: left foot (78.6%) and right foot (80.9%).
  • These trends all generally hold if you shrink the width of the pitch to the width of the penalty box, which was my idea of how to minimise the effect of full backs and wide players versus the touchline.

These are just counted stats, modelling work would no doubt start to highlight confounding factors but it’s interesting to see nonetheless.

We can see trends towards fewer left footed passes being completed across the pitch:

Here we see a fascinating signifier, every team in the Premier League bar one, Wolves, completes passes in the final third at a higher percentage with the right foot than left.

Team Breakdowns

Most teams will have a minimum of one left footed player–their left full back–but usually more. We can contrast opposites in the Premier League by looking at the two teams that attempt more passes than any other teams: Chelsea and Manchester City.

Even allowing for the effect of positional switches, a season long overview gives fairly strong indicators of player location and team structures and tendencies.

Of Chelsea players to have appeared in ten or more games, only their two left backs can be considered to be left footed. Pedro is the only attacker to approach two-footedness from a passing perspective and his Left/Right split is 27%:73%. Manchester City’s squad is chock full of left footed enterprise, with centre back Aymeric Laporte, converted midfielder-cum-full back Fabian Delph (who with a 35%:65% split would actually be a top three in the league two footed player, but lacks minutes, more on this later), Benjamin Mendy and Oleksandr Zinchenko all nominally part of the defensive corps, while David Silva and Phil Foden patrol midfield and Riyad Mahrez and Bernardo Silva usually occupy the right-to left orientated location so many left footed attacking players now do. Even keeper Ederson is left footed.

The example of Man City is extreme, but it’s notable that five of ten high priced outfielders they have signed since the summer of 2016 are left footed, including the one player we’ve omitted so far: Leroy Sané. As a left footed, left sided forward, Sané is legitimately unique in the Premier League:

There’s only one Leroy Sané, literally.

That chart shows every left footed player (minimum 10 games) who has an average pass location within the final third, and the chart for 2017-18 shows the same trend. Not only are Sané’s pass locations on the other side of the pitch to all his inverted lefty compadres, but his passes are recorded higher up the pitch than any other player in the league. Raheem Sterling may have acquired the starting left forward role for Manchester City in recent months, but Sané remains a useful and different attacking weapon.

Two footed players

Part of the motivation to look at this was the search for that most elusive of player: a two footed genius, for which every pass is a decision made by the unconscious mind and for whom the balance of footedness is entirely natural. To my surprise, it turns out that idyllic vision is inhabited currently by just one player in the Premier League, Cardiff’s Harry Arter:


Credit @etmckinley for the viz concept

We’ll see some different scales on these as we go along so stay alert, but here we can see that although Arter doesn’t make a ton of passes (around 25 per 90), there’s a fair split between each foot. He also completes passes at the highest rate for his team at 78%. No player in the league comes close to his 48% : 52%, left to right split (sounds familiar…) and his range of mid to long passes forward with either foot have been a useful pressure release as Cardiff looked to remain in the Premier League this season.

We can see just how scarce this balance is here with this chart of open play passers:

(1000 minutes played to qualify)

The solitary red dot is Arter, and you can see the rest of the top twenty most two footed passers highlighted within and listed adjacent. There are simply very few players who balance their passing output between feet, especially at high volumes and even for those that do here, there’s perhaps a skew towards those who spend time in mostly central positions, which is logical. İlkay Gündoğan is the highest volume passer in the top list, and as we can see here, his general profile skews strongly towards his favoured right foot:

It’s interesting to think of players in these terms though. Even if a player is using their off-foot infrequently, it still feels like a useful skill, if only to avert trouble. The player with the least diverse footedness profile this season is Arsenal’s Shkodran Mustafi:

There’s no 48 to 52 split for Mustafi, he literally attempts one left footed pass for every 48 right footed attempts. The features of his passing; often long and upfield, or to his left, and far less to his right, gives a window into his positioning in possession, Arsenal’s use of a fairly high right full back, and perhaps a weakness, after all there are a limited set of directions Mustafi makes passes towards, and it will always be with his right foot.

Contrast quickly with Sokratis and Laurent Koscielny:

Each of Arsenal’s three main centre backs are almost entirely right footed, but have different passing profiles. Koscielny looks to switch between his fellow defenders, from side to side, while Sokratis will pass forwards with range, left or right. Should an opponent run a press against Arsenal, and consider blocking passing lanes and preventing easy outs, there is information here to work with.

We’ve just touched the surface of ideas around pass footedness here, but hopefully it’s given a flavour of how useful and informative it can be as a standard feature of a modern data specification. The reason why it helps as an aspect for pass modelling work or an input into recruitment analysis or opposition analysis: it scales. Every pass, in every game, in every league that StatsBomb collects holds body part information. Do remember too that StatsBomb has provided free data for women’s football and the 2018 World Cup, so if your interest is piqued by some of the ideas enclosed here, there are datasets available for you to work on too.

Now i’m off to petition the PFA to correct their Team of the Year oversight and replace Paul Pogba with Harry Arter.

An Overview of Pass Heights In the Premier League, English Football League and Scottish Premiership

StatsBomb’s dataspec was designed with a view to capture more of what happens on the football pitch. To that regard, passing was an area in which it was felt that improvements could be made from the perspectives of simply recording how the game is played right through to the functional utility of applying analytical processes on top of that data. Pass footedness is one of the primary upgraded features of StatsBomb data and a useful indicator for player evaluation, but today we are going to look primarily at graded pass heights. Passes within StatsBomb Data can be either “High” (above shoulder height), “Low” (below shoulder height) or “Ground” (self explanatory). For some added clarity we will omit headers from the following charts. As you will see, team and league styles can be quite pronounced.   One of the benefits of collecting data across the entire 92 Premier League and Football League clubs, as well as the Scottish Premiership is to enable stylistic comparisons. We can see that at least in relation to the volume of ground passes. The big six Premier League clubs are in a league of their own–as are Celtic in Scotland. This is no surprise, but as we move elsewhere, there are further intriguing comparisons: Cardiff are complete stylistic outliers in the Premier League this season, and their completion rate here of 83.6% is third lowest of all 104 clubs in the sample. That’s a full six percent beneath the rates that Crystal Palace (89.7%) and Newcastle (89.9%) connect with their ground passes. Of course opportunity impacts volumes, but we see that the Championship possesses a clutch of teams that focus on a ground passing strategy: Graham Potter’s Swansea, Marcelo Bielsa’s Leeds United, Daniel Farke’s Norwich City to name just three. One stylistic feature not shown here that emerged strongly from Leeds’ overall profile was the high volume of low passes attempted; with over 48 per game, they try a full six more than any other team listed. With Bielsa ever interesting as a manager to study, this statistical quirk could well be worth further investigation. Far more compressed are Leagues One and Two. They appear similar in general scope, but surprisingly, League One sees a slightly lower average volume and completion rate for both low and ground passes. Individual teams that stand out are Barnsley and Forest Green Rovers, both organisations that have forward thinking inputs, despite their relative positions in the league ladder. Discussions around the quality of the Scottish Premiership often yield arguments, but at least here, stylistically, the pass volumes and completion rates look more like the lower two leagues than the Championship, with Celtic a considerable outlier.   If we flip this to high passes, again clear distinctions between the leagues are apparent. Have Swansea refound the Swansea Way? The last two charts imply that they might have. Celtic once more look different to everyone else in Scotland and Leagues One and Two are again very similar in profile. One standout trend from this whole chart is the way that pass completion rates of high passes trend differently to ground passes. In the last chart we saw a fairly simple and logical correlation between the volume of passes attempted and completion rate (r=0.84), here we see a weaker negative correlation (r=-0.50) but also what appears to be a skew towards completion rates for high passes at their highest in the Premier League. This offers ideas around player quality, both in passing ability and the ability to make space to attempt and receive such passes. Again logical, but nonetheless distinct.   To get ideas around attacking intent we can extend our look into passes into the box from open play. General trends persist and once more most of the big Premier League clubs and Celtic stand out. But what of Bournemouth? Or Sheffield United, Burton or Bury? All clearly use a ground based philosophy when attacking their opponents’ box and appear to have reasonable success. At the other end, teams are attempting just a handful of such passes and completing barely any. These are teams that likely use the set-piece as a weapon or prioritise higher balls over low and ground passes. Overall trends we saw in general play appear weaker when filtered down to this level.   Now we see genuine stylistic curiosity–recall these are open play box entries via high passes. Would anyone expect that Liverpool would lead the Premier League numbers here? It’s clear that they are well capable of attacking the box either on the deck or aerially and they show high volumes for both. Tottenham and Arsenal both show up less so for volume but for the rate in which they complete these type of passes. Swansea are disinterested in this method as are Norwich, but Leeds–purveyors of high “ground pass” and “low pass” volume– frequently put the ball in the air in the final third. Sheffield United, like Liverpool, show a varied and successful method of box entry. AFC Wimbledon and MK Dons top their respective divisions showing that regardless of which direction the club went, the old Wimbledon way looks to have persisted…


Hopefully this look at pass heights has offered some insight into how considering trends within the whole league pyramid can be informative. We’ve just scratched the surface, but when recruitment is so often local, it makes sense to understand the dynamics of teams across a whole market, and detailed data such as that collected by StatsBomb can help identify teams that play preferred styles for loan opportunities too. Look out for more of these charts on the twitter feed soon.    

Inside League One: Teams, Trends And Metrics

StatsBomb Data has been collected for League One 2017-18 and will be collected for the forthcoming season too. Last week we looked at some players who stood out via the numbers to examine how to get a starting point for player evaluation, so do check out that article if you missed it. This time round we’re going to look at some metrics that are part of or derivable from StatsBomb Data and discuss the results.

In The Clear

One of the major upgrades that StatsBomb Data incorporates is player positioning for all shots. Not just the shooter, but everyone visible within the “frame” of the shot for both teams. This Freeze Frame means that we are able to evaluate the pressure a player might be under at the time of shooting, whether there are blockers prevent a good sight of goal–for the shooter or the goalkeeper–or if a teammate was better placed or open, whether they made the wrong decision to shoot. The data is fundamentally objective too. We are not making a binary decision whether or not a chance is a certain size or simply counting players. We are mapping the location of the players. The potential for exploration within this part of the data is extensive and we are only scratching the surface here. Much more can be done whether you are a StatsBomb Data customer or enjoying the free women’s football data that we have made available. In this instance we are going to look at shots “In The Clear”. Definition: Shots where only the goalkeeper was in between shot-taker and goal, i.e: within a cone between the shot-taker and the goalposts. Open play shots only. Here are a couple of shots from League One top scorer, Peterborough’s Jack Marriott, that qualify as in the clear: As we can see, the defence trails behind him and he’s significantly ahead of his teammates–none of whom are in the frame. These are particularly obvious examples, but a free header would likely also qualify as would plenty of shots in which the player has made space for a shot quite normally. So how did League One shape up in total? Let’s look at how many of each team’s shots were In The Clear: Peterborough (powered at least in part by Marriott) created the highest percentage of shots here, and in fact Marriott himself not only led the league for shot volume but also for in the clear volume too. Of the players with 100 shots or more in the division, the player with the highest percentage of in the clear shots was Ellis Harrison, the player with the lowest percentage was Nick Powell: Visually the differences aren’t hugely pronounced between the two players, but we know that Harrison took a ton more shots that were in the clear while Powell’s total was augmented by lower value less clear shots from range. So we’ve added an extra layer to our shooting analysis, thanks to the information contained in the Freeze Frames. As mentioned, this is just a starting point but we have a great basic example of scalable data analysis that once enacted can be applied across an entire dataset, be that a single league or many.

Pressure in the opposition half

Pressure events are another unique aspect of StatsBomb Data for which there are many avenues of exploration ahead. Here we’re going to take a straightforward look at events higher up the pitch in League One. The team that recorded the most pressure events in the opposition half of the pitch were the champions Wigan. This may be something of a surprise given they were also the team that recorded the most possession (54%) but it shows that their work rate was first class in the zones where it mattered for their attack: By way of contrast, Bradford City recorded the fewest pressure events in the opposition half in the division–over eight per game fewer than next lowest Oldham Athletic:   Their passivity extended over most of the pitch but perhaps crucially not within their own defensive zone. Overall, this less aggressive style did not appear to have had a strong negative impact on their performances as they landed squarely in mid-table in eleventh place. However, they offer a distinct contrast to every other team in League One in skewing so heavily away from pressure events up the pitch. In relation to players, a willingness to chase down the opposition can be a useful string to a forward’s bow as Devante Cole found. He had a stellar first half to 2017-18 at Fleetwood Town, part of which meant that he led the league for opposition half pressure events per game: Exactly the kind of profile that might attract a team oriented towards opposition half pressing like Wigan… who promptly signed him in the January transfer window. Sadly for Cole, he only managed six all too brief substitute appearances for the champions in the second half of the season, but we can at least understand part of the thinking behind his signing–beyond the 10 goals he scored for Fleetwood. Expected Goals Lastly, a quick note about expected goals in the division. StatsBomb Data comes with expected goals values for each shot as part of the package so full player and team analysis can be conducted straight away using the benefit of our extensive knowledge in this area. There is occasionally talk that expected goals is a less reliable indicator in non-top divisions. However, using the StatsBomb xG model, we found that there were few surprises at the top of the League One table in 2017-18. The three teams most impressive via expected goals finished first, second and fourth and all ended up being promoted:   At the bottom of the table, four of the seven lowest ranked teams found themselves relegated, and perhaps Gillingham can consider themselves the biggest beneficiaries of a positive skew in their numbers. By landing 56 points they ended up two wins clear of the drop, yet in doing so overshot expectation by the greatest margin in the whole league: Plymouth were a team who had a season of two halves, 24 points before Christmas and 44 points after and they too look as though they benefited from a positive skew ahead of expectation. For both of these teams expectation ahead of 2018-19 may well need to be tempered, as the structural basis for their final 2017-18 positions indicate possible problems in the future–all things being equal–but it’s best to know this ahead of time, rather than be surprised on the other side of any potential reversion.


StatsBomb will be collecting high quality data for the English Premier League, Championship, League One and League Two in 2018-19 as well as a host of other leagues. Get in touch with us via sales@statsbomb.com if interested or submit our online form. Thanks for reading part two of our look at League One. Part One is here, where we majored on players highlighted by the data. And ahead of the new season we have been tweeting out further highlights from last season’s League One data here. Enjoy the new season!