THE CELTIC Way columnist Alan Morrison recently wrote an excellent analysis of Reo Hatate’s performance levels since his arrival at Parkhead, including thoughtful consideration of the human aspects relative to adjusting to life in Scotland and his relatively limited experience in the professional game.

It is important to build upon Alan’s work with regard to the crucial selection decisions facing Ange Postecoglou for the upcoming post-split fixtures.

While ensuring long-term patience with recently-signed players is sage advice, Celtic have a title to clinch and just five games in which to do so. Hatate’s performance in Sunday’s derby defeat appears to have raised increasing questions among elements within the support.

With that in mind, we have broken out performance metrics using three broad themes: overall value per on-ball value (OBV); defending and pressing; attacking. In addition, the analysis broke out the time series of OBV metrics, including the underlying components, in each league game over the period.

Celtic Way:

Matt O’Riley is included as the comparison within these radars for a couple of reasons. Firstly, he arrived at the club during the same transfer window as Hatate. Secondly, his minutes have been reasonably similar and they have played in similar 'attacking 8' positions on the field.

We can see from the radar above that, outside of dribble & carry OBV, O’Riley has offered significantly more value by OBV. In particular, his pass OBV advantage has been significant (it includes set-piece deliveries) and been at a level over double that of Tom Rogic.

This is a good time to revisit OBV and how it can be utilised as an analytical tool. Conceptually, we can consider it a way to net out the positive and negative events within player and team performances.

For example, with pass OBV, players’ value-creating passes are offset by those which do not come off. This is likely intuitive when looking at Hatate’s 0.01 value over his nearly 10 league game equivalents played to date, as his aggressive passing decision-making has often resulted in turning possession over to the opposition. At least by this model and metric, the good has largely been offset by the bad. In contrast, O’Riley has had an overwhelmingly impactful passing performance level so far.

Another way to think about OBV is to consider the importance of the relative components for each position and role. In the case of an attacking 8 in Postecoglou’s system and style of play, all four are pertinent. 

READ MORE: Celtic's 'irreplaceable unicorn' - Why Callum McGregor represents an under-the-radar recruitment problem

However, shots are a relatively low-volume metric and therefore shot OBV can swing quite significantly from game to game. Defensive actions offer significantly higher volume than shots, though are still well below that of carries and passes. Given the variability of shot OBV, it is useful to look at players’ and teams’ OBV while excluding it. Doing so in this exercise resulted in an adjusted OBV of 0.19 for Hatate and 0.36 for O’Riley, so the disparity remained large.

The radar represents an average for each player per 90 minutes played and, as with any performance data, simple averages may not offer a comprehensive display of underlying trends.

Some simple variations such as mean, median, and mode may be concepts you remember from stats or math classes of yesteryear. For this analysis, we looked at each game in each of the non-shot OBV components as a way to try and ascertain some context for performance consistency. In doing so, the pendulum swings even further in the direction of O’Riley.

Hatate and O’Riley made appearances in 13 and 11 league games over the period respectively. For Hatate, he had positive pass OBV in five of his appearances, defensive action OBV in eight and dribble & carry OBV in five too. In fact, his pass OBV and defensive action OBV components place him below average versus all SPFL Premiership midfielders over the duration of this season to date.

Given the overall performance levels of Celtic during this period, with the added context of his role within Postecoglou’s system, performing near a league-average level in those two facets of play is pretty poor.

O’Riley's consistency shone through using this perspective, as he had positives in nine, eight and seven of his 11 appearances within those three non-shot OBV components. In addition, his pass OBV placed him first among all SPFL Premiership midfielders. He was right about average for defensive action OBV.

Next, we drill down into defending and pressing metrics to try to gain additional perspective:

Celtic Way:

Here we can see that Hatate’s metrics have been uneven but with an issue that may not be surprising to many following the perception of Sunday’s performance: he’s been one of the worst midfielders in the league at winning tackles. This deficiency may compound issues in the Celtic midfield when he’s paired with Rogic, who has also been quite poor at winning tackles and even worse at defending versus opposing dribblers.

Now for the last radar:

Celtic Way:

This report offers an interesting context for the pass OBV metric shown above. We can see that Hatate has been creative, having averaged 0.19 in xG Assisted from open play. While well below O’Riley’s 0.26 rate, it compares favourably with David Turnbull’s 0.18, for example. However, the turnovers and resulting opportunities afforded to the opposition when his creative attempts have not come off have largely wiped out the value of his creativity overall.

What should Celtic do over the remaining five games? Player fitness and availability are important variables with limited outside insight, so preferences are contingent upon the obvious with that.

However, whether it is Nir Bitton pushing Callum McGregor forward into the 'attacking 8' role, or Turnbull re-entering the starting 11, the data suggests O’Riley has to start and that Hatate should be rested.