Unlucky or Inaccurate?

How did Celtic’s attacking effectiveness compare between the 17/18 and 18/19 Champions League Qualifying campaigns?

Celtic played six matches in both seasons which is a very small sample so bear that in mind.

Possession

Celtic appeared to have less overall control compared to last season whilst continuing to dominate possession.

They averaged 81 less passes per match and 7% less overall possession. Two anomalies to call out.

The Rosenborg away game this season saw Celtic happy to absorb attacks whilst comfortably seeing out a 3-1 aggregate lead – also know as “enduring a torrid night” if you write for one of the Scottish papers. Celtic had “only” 47% possession probably as much by design.

Secondly in 17/18 Celtic played Linfield, part-timers, akin to playing Brechin City. None of the 18/19 opponents were part-time.

Even if you remove Linfield from the analysis Celtic’s overall possession still drops from 66% to 63%.

So, less overall control but does that translate to less chances being created?

Creativity

Celtic’s total Expected Assist values for the two campaigns are:

17/18 11.643

18/19 9.156

Again, 5.995 of the 17/18 total is accountable to the Linfield game. Remove that and Celtic created slightly better chances this season, averaging 1.526 per game versus 1.412 in 17/18.

The number of chances created per match fell this season from 18.2 to 14.8 but 51 chances were created against Linfield.

Where Celtic have massively underperformed compared to last season is in breaking the defensive lines with forward passes. Forward passes that remove opponents from the game are called Pack Passes. The total value of opponents taken out the game (defenders getting 3 points per defender “packed”) is called Impect.

Celtic completed on average 14 less forward passes per game for an average 95 lower Impect score.

As a result, they took out an average of 18 defenders per match on 17/18 but only 12 this season.

Even removing Linfield from the analysis results in a decline in passing through the lines.

Goal Attempts

Celtic’s chance quality is measured using Expected Goals. The comparative values were:

17/18 13.036

18/19 9.490

Celtic could have been expected to score 4 less goals this season. And they did! Over performing the xG value in both seasons scoring 11 this season and 15 last.

I mentioned at the time that Celtic’s On Target Shot Conversion of 37% in the qualifiers last season would be very difficult to maintain. Over the season Celtic averaged 31% conversion of shots on target. In 18/19 this was only 26% – 5% below the average. An 11% drop off in On Target Conversion will be difficult to overcome without a very tight defence. Oh.

Celtic’s shot selection also deteriorated with 12 shots per game from inside the box in 17/18 falling to only 9 in 18/19. Shots on target fell from 6.7 to 5.8 per match.

In general play Celtic got possession of the ball 22.5 times per 90m within the opposition box. This was a decrease from 24.5 last season.

Conclusion

Such analysis is tricky given small samples and the distorting effect of playing part-time Linfield. If I remove both Alashkert and Linfield from the analysis then there is a decline in virtually all team attacking metrics however.

There are likely many contributory factors as always, and in both seasons the forward line personnel has been hugely disrupted through injury.

But the Shot On Target Conversion against Astana and Rosenborg in 17/18 was 40% whilst against AEK and Rosenborg in 18/19 it was 21%.

It is too simplistic to say “Celtic didn’t take their chances”. Celtic outperformed their xG last season and this by around 2 goals. This season they were expected to score around 5.8 goals against Rosenborg and AEK and scored 5.

There has also been a real decline in passing ability through the lines (we can call this the Armstrong Effect) which is an area for concern.

There certainly hasn’t been improvement and that is sobering.

To complete the picture I must next look at defensive performance for further clues. Wish me luck!