Yesterday I released a chart examining players in the League of Ireland. As far as I know, this kind of work is new to the league, though correct me if I am wrong. After all, I might have missed something along the way with my head buried so deep in Excel and Tableau. There is a possibility that some club analysts may already be using these charts for recruitment purposes. Although I do find that unlikely. You’d be surprised how many Premier League club’s don’t do this yet either. Using these kinds of dashboards to convince coaches and chairpersons that the player they are pursuing is a fabulous visual aid. So without further ado, I have attached the visual comparing two strikers below: Cork City’s Graham Cummins (30) and Dundalk’s Patrick Hoban (26). That picture of Cummins is deceiving…he looked mid 20’s to me.
So let’s get started. First and foremost, the data collected comes from Instat. They cover a wide range of leagues and players worldwide and data is collected on a massive scale. I do not have the slightest idea how they collect their data as they often require a video feed to analyse the games (or so I think at least). This requirement for a video feed can often lead to below par video footage (which I myself have) often seen on occasions. Either how they collect a bunch of metrics which can be seen in the visual above. The number of metrics is not particularly the issue, rather the quality of them. Firstly, I personally am more familiar with Opta’s data and therefore trust it’s reliability and it has been used more often in the public domain (that is not to say that Instat’s collection is wrong, more than it needs validity and reliability tests). Secondly, the quality of metrics is not as in-depth as I would like. For example, the visual above shows Shots & Goals. That is great as a start but we would like more granular information, such as shot locations and expected goals. These are now considered the norm in evaluating players, even though they do have their own problems.
As for the definitions of the metrics used, Instat does not provide these. I assume most are self-explanatory – ball recoveries is another way of saying Interceptions?. Where Tackles and Challenges differ is another world to me..
Back to the visual itself. The easiest way to “read” the visual is to examine each “box” and identify the dot that is furthest to the right. The visual is designed to compare two players against each other at the same time. In essence, the further to the right the dot is, the better that player is in that metric. Important to remember that not all games played are included in the dataset. The reason being that the data provider has not covered all games. The easiest and most effective way, therefore, is to examine players on a p90 basis. Please read through p90 before continuing below.
Let’s take a look at the snapshot to the left. While Cummins has played fewer games and took fewer shots than Hoban, he has a higher goal per game ratio than Hoban. Cummins on average also provides assists to his team more often than Hoban. Is he, therefore, a better striker? I do not know the answer to that question. What we can establish from the data is the following:
- Cummins may be more effective in converting shots to goals. Yes, I said it! That totally depends on several things: his shot selection, his age, his playing style and a few other factors – currently which I do not have access to. For example, if we had access to shot locations, we could establish an expected goals total for the two players and compare their actual vs expected figures. This may help identify a “better” shooter.
- Or the season is still too early and doing any of this work is just plain nonsense. Agree, with the first part. It is still very early on in the season and a small sample size is a big plague (for now). Let’s revisit this towards the end of the season.
Sadly, that is all the data I have access to at present. The metrics, for now, are most likely benefiting the attacking rather than the defensive players. It is a small side project that I run out of enjoyment and a personal goal of adding analytical value to football in Ireland. You are more than welcome to ping me and request player comparison graphs for the League of Ireland and I will post them on Twitter. That’s all for now folks.