One of the facets of analytics in football which remains in its relative infancy is that of goalkeeper play. Statsbomb is probably at the forefront of related data capture and metrics amongst the more widely used industry vendors, and John Harrison, head data scientist at Goalkeeper.com, has been a pioneer in capturing and modeling play for the English Premier League.
I have spent a fair amount of time over the past year learning from John’s work, as well as evaluating and working with Statsbomb’s keeper-related data. This process led me to ask the following question:
Does Scotland have a goalkeeping problem?
An answer to this question is ultimately a subjective one, as until the movement of the ball itself is tracked and quantified, measures of shooting and shot-stopping will remain less-than-precise. However, this does not suggest that the current generation of metrics offers no information or insight.
The two primary measures upon which I will focus today are goals saved above average (GSAA) and goalkeeper on-ball value (OBV). Statsbomb’s GSAA metric is the difference between the number of goals a keeper conceded compared to the total of post-shot xG the keeper faced - or to determine how many goals the keeper’s ‘shot-stopping’ either prevented or conceded relative to the average keeper.
The goalkeeper OBV metric is more ambitious, as it also incorporates other elements of keeper play such as claiming crosses and sweeping. For example, if an opposing team were to cross a ball into a range in which the average keeper would be expected to claim it, and the resulting unclaimed cross resulted in a saved shot, the keeper would get penalised for not claiming the cross and credited with preventing whatever the post-shot xG was. Goalkeeper OBV is designed to provide a single metric to offer a net value for the various facets of keeper play.
Statsbomb’s data began for the 2018-2019 season, and here we can see a scatter plot of the two selected metrics for all keepers who played at least 1,200 minutes that season:
Given the importance of shot stopping to keeper value, it makes sense that the two metrics plotted together would have a strong correlation. Rather than focus too much on any particular keeper, note the GSAA average for the league was 0.06 and goalkeeper OBV -0.06. In fact, for the keepers plotted on the graph, the total GSAA was positive 9.24 but the total goalkeeper OBV was -36.68!
The pandemic-shortened 2019-2020 season had the following:
That was a truly awful season for keeper play in the Premiership, as the aggregate for shot stopping for the keepers in the graph was -24.83 goals, total goalkeeper OBV was-46.62 with league average negative for both metrics as well.
The following season in empty stadiums, 2020-2021, brought with it improved keeper play overall:
While the total for GSAA and goalkeeper OBV for the keepers in the graph were -4.40 and -28.09, at least the league average levels were near zero.
Last season included a return to decent shot-stopping, but still underwhelming keeper play overall per goalkeeper OBV:
The keepers in the graph totaled 5.16 in GSAA and -27.63 in goalkeeper OBV.
Last but not least, the following graph includes keeper play so far this season for those who have played at least 300 minutes:
While the season is obviously less than a third completed, so far the GSAA and goalkeeper OBV are tracking at alarmingly poor levels, at -12.64 and -24.01, respectively.
From an analytics perspective, the persistently poor performance levels of keepers in the league introduce potentially significant questions as to how keepers are currently being developed, recruited, and coached.
Even big-name keepers in the league have offered relatively low goalkeeper OBV since the beginning of last season versus all keepers across all men's competitions covered by Statsbomb. Here was Joe Hart compared with Craig Gordon and Alan McGregor:
Gordon has faced a far higher volume of shots than either McGregor or Hart, and from a value perspective, it is easier for a good shot-stopper to accrue value when facing a higher volume of shots, but also suffer from negative value from increased exposure to cross claiming and sweeping, which appear to be significant negatives for Gordon.
In fact, with both Hart and Gordon at -0.07 in goalkeeper OBV over the period benchmarking around the 50th percentile across all men's competitions, the issues may not be specific to Scotland! Whether by design or luck, Livingston’s keeper play since the 2018-2019 season appears to have assisted in them persistently overperforming their relative wage bill.
Total OBV for keepers also includes their distribution/passing, and Livingston’s primary keepers have ranked as follows since 2018-2019: 2nd, 5th, 7th, 2nd, and 6th. It appears that John Harrison has fertile ground to tend to across the industry, and those clubs that improve their standard of goalkeeper analysis could benefit greatly
Comments: Our rules
We want our comments to be a lively and valuable part of our community - a place where readers can debate and engage with the most important local issues. The ability to comment on our stories is a privilege, not a right, however, and that privilege may be withdrawn if it is abused or misused.
Please report any comments that break our rules.
Read the rules here