HRT – Human Replacement Therapy

Liverpool’s league starting lineups 1992 to 2015

Managers cannot always send out their best sides. Sometimes they don’t even want to, despite Alex Ferguson’s sweeping assertion that a manager worth his salt always starts his best players. They might be injured, or otherwise indisposed. They might feel ‘off’, physically and/or mentally. They might have incurred the manager’s wrath, or broken their contract or even the law. Perhaps they’re being rested for a later game against a more difficult opponent, or exiled to international duty. Jurgen Klopp even had to have himself replaced on 2 January 2022, having a positive test for Covid-19. Who’d be a manager!

The 25-man squad rule was introduced for the 2010/11 season, but supplemented by the permitted use of u-21 players. An increase in the size of available squads must have an effect, making a change of lineup more likely than soldiering on with a team in difficulty, as well as promoting youth from the Academy, and encouraging rotation to keep players happier. So would the acquisition of new, potentially first-team, players who have to be bedded in and tested in real time with existing players. At the other end of their career with Liverpool, players wishing to move to other clubs can be showcased in order to justify and hopefully boost their transfer value. Managers who have been fortunate enough to win the Premiership earlier than the last game in the season can use the extra time for adjusting lineups and other experiments for which players would not normally have been chosen.

WBA boss Roy Hodgson and Kenny Dalglish shake hands in 2011.

We should therefore expect a myriad of considerations to lead to a certain amount of uncertainty in any analysis of how and why managers have chosen their best team on the day, but we can at least start with what actually happened before trying to explain any anomalies thrown up. Liverpool’s starting lineups in each Premier League season from 1992/93 have been examined, the differences averaged out season by season, manager by manager. We can then see the overall effect of a season or of a manager in the light of the overall, 22-year average, 1992/3 to October 2015, which is 2.10. In other words, LFC changed their starting PL lineups, between 0 (‘no change’) to 11 (‘clean sweep’), by an average of 2.10 players in every consecutive PL game after the first in each season. Note that the changes are calculated from the previous League game, not necessarily the previous game. (For LFC under Jurgen Klopp, see Appendix.)

The average figures for each manager should be seen in the context of this club average of 2.10.

Manager   Avg
Graeme Souness 1.77 
Roy Evans 1.24
Gérard Houllier 2.02
Rafa Benitez 3.09
Roy Hodgson 1.52
Kenny Dalglish (2) 2.42
Brendan Rodgers 1.74

* Dalglish's second reign

No Liverpool manager in the PL before Benitez made more than five changes from one match to the next, but the Spaniard increased this to eight, 10 days before Istanbul, and to nine, 10 days before the 2007 Champions League Cup Final. Dalglish (episode 2) subsequently used nine changes 4 days before facing the 2012 FA Cup final. Rodgers then reverted to using a maximum of four changes between games, having had no such exalted climaxes to cater for.

It is quite clear that, in normal times before Covid-19, by far the major factor in determining the number of changes is how far each manager has taken LFC in cup competitions. We’re always told of players’ hearts failing if they are selected to play in a match close to a final, interpreting that as being out of consideration for selection. Which LFC manager first used the technique of shielding the better players from the risk of injury before a final? If we go back, pre-Premier League, as far as Shankly, the first, and indeed only, suspicious case was the FA cup final at the end of 1985/86 when Dalglish used four players (Hansen, Molby, MOTM Rush, and himself) who had not played in the previous game, the SSSC semi-final, but three of them had played against Chelsea a week earlier. We need no reminder that LFC subsequently had a long PL period during which we did not enjoy the sort of final for which our best players were worth protecting, until Istanbul.

The heavy hand of the FA, which threatened any club with reprisals if they fielded a deliberately weakened team, had been the controlling factor. Removing top clubs from the clutches of the FA by starting the Premier League was the background allowing them more freedom to exploit their larger squads. Benitez changed eight in the match before Istanbul, with apparently no penalty. In 2010 Blackpool was fined £25,000 for fielding a weakened side against Aston Villa, but in 2011 Premier League clubs agreed that they could choose any player from their 25-man squad with only a very limited number from outside.
‘No change’, in which the starting lineup from the previous league game is repeated, has been relatively rare, totaling only 141 of the 871 games in which they were possible (i.e., all 22 seasons, but not counting the first game in each season). 77 were ‘one-offs’, 20 were doubles (i.e., the same lineup repeated twice), 5 were trebles, 1 quadruple, and in 1996/7, Roy Evans kept the same starting lineup five further times, plus a European game part way through the sequence! In all, he chose 52 ‘no changes’ in his 172 games. (As can be seen reflected in the tables, Evans was more likely to be loyal to former lineups than any other manager.)
Between the extremes of no change and replacing the whole deck, there are other factors worth investigating across the run-of-the-mill PL matches. For example, there is a relationship between the average number of changes made and the result of the previous league match.

Average number of changes after LFC's results

Manager  Won Drawn Lost 
Graeme Souness  1.59  1.29  2.35
Roy Evans  0.95  1.46  1.65
Gérard Houllier  1.80  2.28  2.27
Rafa Benítez   2.83  3.27  3.51
Roy Hodgson  1.43  1.50  1.63
Kenny Dalglish (2)   1.48  2.86  2.92
Brendan Rodgers  1.43  1.55  2.48
 There is a consistent, and not unexpected, pattern in these figures, even in the short reign of Roy Hodgson, with more changes made following a loss than a win. Echo quoting Hodgson in depths of despair: ‘“There are moments giving me satisfaction.” For those moments to become much more consistent, then, surely it is time for Hodgson to play this lineup regularly and allow them to build up a rapport. Chopping and changing does no good and, if anything, only makes things more difficult.’ It looks as though Souness used the weapon of changes in the starting lineup following losses, but appears to have been happier after a draw than after a win! Evans was much more forgiving whatever the previous result had been. Houllier changed about twice as many starters as Evans had done following a win.

Rafa the Tinkerman

The outstanding anomaly in both tables is clearly Benitez, and some kind of explanation should be sought. In his six seasons at LFC, he fielded only eight unchanged league sides, four of which were in 2007/08. To the accusation that Benitez was over-fond of tinkering with his starting lineups, we should remember his proud record in cup competitions, which can easily be shown to have a significant impact in order to protect players approaching such a culmination to a season - Benitez took Liverpool to the Champions League final and the League Cup semi in 2004/05, the FA cup in 2005/06, the Champions League final in 2006/07, the Champions League semi in 2007/08, the Champions League quarter-final 2008/09, and the Europa League semi in 2009/10. However, after the thirteen wins between the preliminary Champions League round and the start of the FA cup on 7 January 2006, with no major trophy to consider, Benitez still made an average of 2.54 changes, keeping the starting lineup from the previous game only once. (He made a total of eight changes following his two defeats.)

Stephen Warnock, summed up his opinion of this aspect of Benitez’ reign as manager. "Even if you’d played really well one week, you never thought you would play the next. People were being rested in September and October and I just think that’s crazy. You don’t want to be rested at that stage of the season. I thought I needed to play week in, week out to improve myself as a player. But it was not just the team, it was the same with the squad as well. The squad used to go up on a Friday afternoon and you wouldn’t even know who was going to be in it. It was comical sometimes. No one would have a clue why certain people were not included but the manager would keep his reasons to himself." Warnock himself started five times in a row at the start of the 2005/06 league season, but in the rest of his two and a half years under Benitez he never started more than two in a row, and that only five times. Meanwhile, he was used on the bench 29 times, from which he was brought on in only eight.

It looks as though Benitez was addicted to change – even at Valencia he was known for a rotation policy which included resting his best players - and I wondered whether his recent (2021/22) run with Everton would confirm that suspicion. He started no single player in every one of the 20 league games, and only one game involved no change. His average changes of starting player was 2.53, less than the ‘tinkering’ with Liverpool, and almost exactly the same as Klopp but with no European matches to protect. In the last five games with Everton, as he approached his increasingly probable dismissal, the figure rose from 2.25 thus far to 3.80, giving at least the impression that he was trying desperately, through replacements, to find a novel solution to his eventually insoluble problem.

There are opposing views on changing lineups from one match to the next. On one hand, managers may be seen as identifying problems which need correction; alternatively, it may be feared as potentially disrupting the understating between players, which would then have the opposite effect. So, what do the results suggest? Is there any correlation between success in the match and the number of changes in the starting lineup?

The results of 871 matches in the whole period under review (from 1992/93 to October 2015) are unexpected. Almost exactly half were wins for LFC, no matter how many starting lineup changes had been made. A further quarter were draws, and the remainder were losses. The inescapable conclusion is that the number of such changes in any one match has little or no bearing on the result. If we look at the relatively few (statistically insignificant) games in which our managers started with five or more changes from the previous lineup, 24 resulted in wins, 14 in draws, and only 11 in losses. Changes to the defence are normally regarded as potential problems because of the perceived, necessary understanding between goalkeepers and their back line. Misunderstandings and lack of time playing together may lead to unnecessary goals being conceded so that you would expect managers to try to keep the same defence wherever possible. Altogether, of the 220 games lost between 1992/93 and October 2015, however, a goalkeeper or defender had not played in the most recent 114 league game. On at least five of the occasions in games we have lost, a goalie and a defender have been selected when neither has played in the previous game, and in 26 games we have lost with two new defenders in the starting lineup. Benitez stands out once more for the high number of changes involved. – of his 47 losses, 33 (70%) involved a change of defender and/or a goalkeeper in the lineup, many with two changes and occasionally three; yet, 90 (74%) of his 121 wins had the same defensive changes, 29 of them with two in the same game and twice with three.

The flow of league games through a season is interrupted by the intrusion and excitement of the FA Cup (in which LFC reached three finals since 1992) and League Cup games (six finals). LFC’s approach to both competitions has evolved during the Premier League era in which the all-too occasional added pressures of international matches will also have played a part. The following brief analysis is on the basis of changes from the immediately preceding games, not necessarily those in the cup competitions.
From 1992/93 to October 2015, we played 75 FA Cup games, in which managers made 196 changes from the previous (usually PL) games in the starting lineup, an average of 2.6 per game (compared with the 2.10 for one PL game to the next). The same enquiry for the League Cup gives 379 changes in 91 games, average 4.16 per game. Even Souness and Evans were more likely to make changes for the League Cup than for the FA Cup.

It comes as no surprise that the average number of changes made by Benitez was very high - 4.6 in the FA Cup, and 8.1 in the League Cup – not far off twice as many as by the other managers. No manager before Benitez had dared to change all eleven players in either competition. He, Dalglish and Rodgers did so in the League Cup. ‘No change’ (i.e., repeating the whole starting lineup of the previous match) is much rarer than in PL games, a total of 6 in the FA Cup (all before Houllier), and 10 in the League Cup. 

Klopp uses his squad to the fullest


Jurgen Klopp. October 2015 to present day May 2022

Klopp’s figures for his choice of lineups in the PL have been significantly determined by the influence of international matches and Covid-19, as well as the large number of wins! They have therefore been included as an appendix in order to prevent them distorting the overall findings for his predecessors. It is believed, however, that he prefers consistency to rotation for its own sake.

When it comes to HRT, Jurgen Klopp is not the normal one. His starting lineups for the 251 league games have averaged 2.61 changes per game, higher than any other manager except Benitez. His most erratic season in this respect was his first, with an average 3.76. Thereafter his figures have swung from the most stable since Evans in 2016/17 to the least stable of any during the Premier League era in 2017/18 and 2021/22.

Until 10 April 2016, Jurgen Klopp changed his starting personnel an average of 2.47 times, but the figure increased to 7.37 in the last 8 league games of 2015/16 as he protected his chances of progressing through the quarter- and semi-finals of the Europa League Cup. Three days before the final, he culminated this policy with a clean sweep, changing all eleven players for a match at West Brom. He is not averse to keeping the same team twice in the league, doing so 18 times so far including a ‘triple’ towards the end of the 2016/17 season.

It is often said that a manager’s approach to cup games can be judged by the number of starting lineup changes compared with league matches. Until the present season, Klopp averaged 8.3 changes in the FA cup, but this fell to 6.8 for the Cup-winning year, still far higher than for league games, of course. With a successful cup run in sight, however, ignoring the final warmup injuries, he made only seven changes in the two semi-finals and final together. Klopp has worked towards a clean sweep policy so far that while he made no change for the League Cup final in his first season, he gradually separated his squad into what looks suspiciously like a ‘B’ team for the League Cup, making eleven changes in each of the only three matches in 2019/20. The same increasing consistency which has accompanied success is also clear for the League Cup runs. We have been knocked out before Round 4 in three of his six seasons. Before 2021/22, he made 7.53 changes per game, falling to 7.14 in the year of our success. For the first semi-final at Anfield he made 8 changes, but only 2 for the second leg at the Emirates, and 3 in the final. It will be interesting to see if the content of the 2022 trophy cabinet will have any effect on his lineups for the domestic cup games next season.

As for Klopp’s European teams (again comparing the lineups in the games immediately preceding them), 2015/16 in the Europa League produced the most changes, though the figure (4.27 per game) is distorted somewhat by his clean sweep for the final with Sevilla, following seven changes in the return leg against Villerreal. The averages fell to 3.27, 2.92 and 3.22 after the fallow season (2016/17), but climbing to 3.80 and finally 4.23 in the most recent. Only once has he faced a European opponent having made no changes from the previous game.

The Sevilla final apart, his willingness to experiment is largely confined to the group stages, especially when little rests on the outcome of a specific match, or after an easy ride in the first leg. For example, he made eight changes for the return legs against Midtjylland in 2020/21, and against AC Milan in 2021/22.

Copyright - Colin Rogers for


We've got all the results from official games, appearance stats, goal stats and basically every conceivable statistic from 1892 to the present, every single line-up and substitutions!