KNOCKOUT football in general – but cup semi-finals in particular – are more prone to upset and anomaly than a 38-game league season.

Just ask Chelsea and Bayern Munich. Both played home and away Champions League ties and racked up more shots, more possession and higher expected goals yet find themselves dumped out of the tournament. That is heartening for those who don’t wish to see a direct correlation between money and success and thrilling for the neutral watching two dramatic ties.

Moving closer to home, and looking ahead to the upcoming Scottish Cup semi-final, can the data help predict such one-off ties, or do we throw our hands up and say ‘it’ll be random!’?

Form

Celtic supporters will hope that form prevails against Glasgow rivals Rangers - but only so far as this includes domestic Scottish action.

Since Van Bronckhorst replaced the Steven Gerrard, the comparable Scottish results have been:

Celtic Way:

Van Bronckhorst’s first day in the office was watching his side succumb 1-3 to Hibernian in the League Cup semi-final. This is a trophy Ange Postecoglou then picked up to match previous incumbent, Gerrard, in his medal haul after just a few months.

Since then, Celtic have turned around a six-point league deficit and forged a six-point advantage. Both sides have progressed in the Scottish Cup, obviously, while in Europe it is the Light Blues who navigated their group and then two rounds on knockout football.

In particular, their 6-4 aggregate victory of Borussia Dortmund was eye-catching. Yes, they have been on the right side of the variance fairy in terms of xG in that tie, and against Red Star Belgrade, but it is fair to say they have outperformed Celtic in Europe this season.

So the form guide seems to show the Hoops well ahead under their manager, but it is clearly a murky picture. What is more important: Celtic’s increasing Scottish dominance, or accounting for the fact Rangers can beat the likes of Borussia Dortmund?

Previous encounters

The two managers have clashed twice.  Once in front of a Celtic-only crowd, under the lights at Parkhead. That night saw Postecoglou’s men sweep their rivals away with a devasting first half of attacking football. They then showed a more pragmatic side, dropping 10 yards and containing a possession dominant away side that created only one very good chance on the 94th minute.

The main stats for that match were, from StatsBomb:

Celtic Way:

Celtic racked up most of their offensive output in the first half and Rangers in the second. Therefore, game-state is an important context to consider when assessing the overall match data.

The second meeting under the current custodians was recently at Ibrox. A potentially fatal title blow was struck by the men in green and white with a come-from-behind away win that was sealed at the end of the first half. With curious symmetry, Celtic then fell back into a medium block and headed away cross after cross (32 of them) as Rangers ran out of ideas on how to break them down.

Once again, game data would show Celtic dominant up to half-time, then containing after that. Using StatsBomb data once more:

Celtic Way:

Both these matches saw one team away from home with minimal or no support.

READ MORE: The treble tie that binds Ange Postecoglou and Martin O'Neill - Tony Haggerty

Also, the game contexts were specific. In February Celtic had to win to ensure they did not fall way behind at the top. In April, Rangers had to win to claw back the deficit at the top given the last fixture between the two will be at Parkhead.

That suggests Celtic have an edge in terms of dealing with the ‘must-win’ pressure… which is exactly what a semi-final is.

Hampden hoodoos

Celtic have recently completed a quadruple of trebles. The most remarkable aspect of this is navigating 36 consecutive rounds of cup football without mishap. A club with the wage bill advantage Celtic enjoy should win a league over a 38-game campaign. But cup football is different. Just ask Ronny Deila about Steven McLean. Or Neil Lennon about Ross County.

Since then, St Johnstone have become the kings of cup football in Scotland, capturing both trophies last season. Indeed, it is the Perth team’s mantle as second most successful side in Scotland over the last decade the Ibrox club now aims to emulate.

The Rangers squad have largely been together a while whereas Postecoglou fielded 12 new signings out of 16 deployed in the last league match. His charges have, however, lifted a trophy at Hampden. Both the League Cup semi-final and final saw tight one-goal victories over St Johnstone and Hibernian respectively.

Let’s look at the modern Hampden* records since the 2012-13 season:

Celtic Way:

* Celtic played one game at Murrayfield; ** Celtic’s two draws were a penalty shoot-out win vs Hearts and a penalty shoot-out loss to Rangers; *** Rangers’ draw was a penalty shoot-out win against Celtic

Specifically for semi-finals, the record is:

Celtic Way:

This Rangers side has a desperately bad record at Hampden. They have only won two out of 10 semi-finals and none of the two finals they have got to since 2012. The current squad is full of players with recent Hampden heartache close to hand.

On the other hand, Celtic have been largely imperious at Hampden (and Murrayfield) in recent seasons. Since the 2015-16 League Cup semi-final defeat to Ross County, only the same season’s Scottish Cup semi-final loss on penalties to Rangers has seen disappointment.  For six years, every Hampden appearance has seen success.

There is still a small core of Celtic players who see Hampden as a second home – Nir Bitton, James Forrest, Callum McGregor and Tom Rogic. However, as mentioned, the new players in the squad has already lifted a trophy there.

Will the experience and know-how of competing and winning knock out football in Scotland be telling for Celtic?

xG

Finally, we need to consider the staple of all data analysis: expected goals. What are the trends in terms of xG for and against for both sides?

First, Celtic:

Celtic Way:

What to focus on here is the following two points:

  1. Celtic’s overall xG difference is consistently above 1.5 per game. The higher the better as it reduces the impact of the odd piece of negative variance like a sending off, bizarre own goal, 30-yard screamer and so on.
  2. The variability in the data is reducing. That is, the gap between the top and bottom of the green lines is smaller meaning greater consistency of performances.

For Rangers:

Celtic Way:

Focusing on the same points:

  1. The xG difference for Rangers is much smaller – around 0.8, or half that of Celtic. That means they are more vulnerable to adverse fortunes impacting the final score.
  2. The volatility in xG difference has increased markedly under Van Bronckhorst. Under Steven Gerrard, their performances were very stable. This is not a surprise – he had them set up in a very compact, well organised and disciplined side. Especially in Europe, his team was hard to play against and gave little away. Van Bronckhorst has a more open and expansive style but there is greater inconsistency in their performances.

Summary

Celtic are the form team, are achieving greater consistency and xG differential in their performances and have a stellar recent Hampden record. Additionally, they have won the last two meetings between the respective managers with relative comfort.

However, Rangers’ volatility currently may be their secret weapon. Will Celtic get the version that defeated Borussia Dortmund, or the side that struggled against them last time and drew with Motherwell, Aberdeen and registered zero shots on target in the Braga first leg?

You can only control the variables to some extent. What Postecoglou has provided is a solid platform of team shape, organisation and purpose upon which to build success.