Michael Cox looks at the tactical battle between Roberto Mancini and Jose Mourinho.
Manchester City v Real Madrid, Wednesday 7:45, Sky Sports 2.
The meeting at the Bernabeu was one of the matches where Roberto Mancini switched to a back three midway through the match - but it wasn't City's formation that was the problem, more the lopsidedness with which they played it.
Both Aleksandar Kolarov and Gael Clichy were deployed on the left, but Pablo Zabaleta had to defend the right all by himself - Cristiano Ronaldo and Marcelo kept overloading him down that side, and predictably, both scored in the final 10 minutes to secure an important victory for the Spanish champions. Therefore, Mancini's first priority in this game should be to protect the flanks.
However, I think he'll go for two 'drifters' in the wide positions - David Silva and Samir Nasri either side of a midfield four. Those two will move inside into central playmaking positions, and I think this is the key area for City in an attacking sense - Real don't really have a natural ball-winner, with Xabi Alonso a deep playmaker and Sami Khedira more of a energetic box-to-box runner.
If Silva and Nasri can create overloads there, possibly along with Carlos Tevez dropping deep, City have a great chance of creating opportunities. The all-Argentine combination of Tevez and Sergio Aguero worked well against Ajax and also in the weekend win over Aston Villa - Tevez creeps between the lines, Aguero has blistering acceleration in behind defenders.
However, they must be patient. Real are arguably the best counter-attacking side in Europe, and there's much more to their threat than Ronaldo. Mesut Ozil travels with the ball on the break and makes excellent decisions in possession, while Angel di Maria starts in deeper positions on the right and can wreak havoc with his left foot, either shooting at goal or bending the ball through the defence for Ronaldo or Karim Benzema. If City commit too many resources forward, they're in danger of Real's swift transitions.
Because of that danger, they need a strong performance from a central midfielder. Mancini has options here - he could use Gareth Barry and Yaya Toure as his central two, but Javi Garcia is another option against his former club. The Spanish holding midfielder was poor against Ajax in City's previous Champions League game, though, and Mancini might ask Toure to play a cautious role alongside Barry - who won't want too many direct confrontations with Ozil, considering how he was embarrassed by him in England's 4-1 World Cup 2010 second round defeat to Germany.
There are unlikely to be many surprises from Jose Mourinho, although the use of Luka Modric instead of one of his usual midfielders would give Real more control in the centre of midfield. Mourinho doesn't mind a 'broken' team without the ball, and the front four will probably press at goal kicks but then drop deep into their own half. Mourinho will want to prevent City from settling into a good passing rhythm (and instead force Joe Hart to kick long towards two small players, Tevez and Aguero) but City's defenders aren't skilled enough in possession to warrant constant pressure. Soaking up pressure before counter-attacking is Real's likely strategy.
Expect goals here. This should be a high-tempo, action-packed game - and neither side have kept a clean sheet in their four Champions League matches so far this season. But how to play it? Both teams to score is an unappealing 1.51/2, and Over 2.5 goals is just 1.68/13. Over 3.5 goals looks much more tempting, at 2.56/4 or over - you can easily see four goals in this match. Watch out for a City comeback, too - they've specialised at that in recent months.
It's impossible to ignore City's problems at defending set-piece against Ajax - both Siem de Jong's goals came from corner situations. Ronaldo is an obvious threat in the box, but this might suit Sergio Ramos, who has a tremendous leap when given a free run. He scored a header against Athletic Bilbao at the weekend, and around 12.011/1 for Ramos to score again would be decent value.