The Dubai Myth?
By James Dailey
Today’s installment continues my “campaign” against 10 in a row complacency. Much has been made of this past season’s winter break, “what happened in Dubai” for Celtic and The Rangers, and the subsequent 13-point lead. Various narratives about winning mentality by Celtic have emerged, so I decided to analyze things from an advanced stats perspective. Following my review, I believe that narrative may be overblown. In honor of my friend Andrew on Twitter, I have distilled this down to an executive summary:
Celtic’s wage advantage of around £9 million empowered a deeper squad better able to endure injuries. The Rangers had enjoyed relative health pre-break and that good fortune ended, which left their relative lack of squad depth exposed (Ed – they had the largest squad in Scotland so it was a lack of quality not quantity?).
Morelos experienced a collapse in output despite a comparatively modest drop-off in advanced stats, all while having widely reported issues in his personal life.
Lennon’s shift to a 3-5-2 addressed various structural issues which dramatically increased the offensive production of both McGregor and Edouard (and the team overall), while re-introducing Griffiths and increasing the balance of our attack across the pitch. The result was a consistent offensive explosion and Celtic dominating domestic opponents to a level where the risks of dropping points plummeted.
I will be breaking this analysis down into two separate pieces. This first one will review overall performance stats pre- and post-break along with squad depth, while in the second I will review the performance disparity between Edouard/Morelos and the shift to the 3-5-2 formation.
Pre vs Post
The following data breaks out each teams’ performance and output (including Europe) per 90 minutes excluding penalties for the pre- and post-winter break periods:
Celtic’s underlying xG and xG differential went from good to very good, while goal output remained above what xG would suggest – i.e. Celtic consistently outperformed their xG.
On the other hand, Rangers went from outperforming their xG during the pre-break period at a level comparable to Celtic to dramatically underperforming it.
Celtic most definitely took performance levels post-break up a notch, while Rangers remained at a comparable level. It was their goal output which collapsed compared to the scoring chances created.
Rangers’ last two league results prior to the shutdown are excellent examples: the 1-0 loss to Hamilton at Ibrox and 1-0 victory in Dingwall vs Ross County. They accumulated xG of 5.66 and surrendered just 1.05 combined in those two games. Incredibly, they outperformed Hamilton by xG of 3.45 to 0.61 on 31 total shots. To place that game into perspective, the xG gaps and number of shots were greater than those in Celtic’s 5-0 victory over St. Mirren in the last league game of the season. Rangers lost two other fixtures which were basically draws from an xG perspective, which means that they could have easily accumulated an additional 5 to 9 points post-break.
Variance between underlying performance stats like xG in single or a small number of games can be significant. Celtic were truly dominate in the league this season, but a relatively normal variance in luck could have swung the league table to be much closer.
Pre-break there were 5 Celtic victories in which xG was close enough (less than 1 goal) where normal variance could have resulted in dropped points, including 1 each against the likes of Hamilton and Hearts. By comparison, The Rangers had only 3 victories pre-break which were that close. It would have taken just a few of those games, and/or the three post-break for Rangers, to “break” against Celtic for the league to have been uncomfortably close.
Squad Depth
The pre-break period for The Rangers was exceptionally fortunate when it came to injuries. Outside of the cheap shot-induced injury to Jordan Jones, no player of real standing missed meaningful time due to injury until Helander was injured in the December Cup Final.
In addition, Morelos’ disciplinary record pre-break was relatively clean. In contrast, the post-break period of 10 league games included missed time by the following players due to injury: Tavernier, Jack, Arfield, Defoe, Barisic, and Helander, while Morelos missed several games as his disciplinary issue re-emerged.
Would Rangers have dropped more points pre-break had the timing of injuries been spread out more within the season? That is certainly a plausible scenario.
However, the condensed timing of the injuries into the post-break period was likely even more impactful due to their relative lack of depth. Celtic certainly experienced a fair share of injuries throughout the season, but they were more dispersed and the wage disparity of about £9 million is the equivalent of 4 to 6 additional high-quality players and depth. For example, Celtic had the depth to have Christie, Bolingoli, and Bitton on the bench, while The Rangers had players like Barker, Halliday, and Edmundson.
Conclusion
I believe an objective review of this past season’s performance data suggest that the ultimate 13-point margin was not an accurate indication of how close the league was. The post-break explosion in the points gap was largely driven by a combination of a rash of injuries to a weaker Rangers squad combined with a collective swing in goal production of about 1.60 goals per game in Celtic’s favor.
Approximately 0.40 of that was due to improved Celtic performance levels, while about 1.20 was due to a collapse in Morelos’ ability to convert chances - both of which I will analyze in my next piece.