The power of customisation in the Twenty3 Football Lab
Everybody does things differently. Often, an off-the-shelf solution will work well enough, but it’s always better to let people go their own way if possible. It’s with that in mind that we’ve added the ability to make custom stat groups in the Twenty3 Toolbox’s Football Lab.
Since the Toolbox’s launch, we’ve grouped statistics in the Lab into easy-to-use and sensibly organised categories. You could, for example, use the ‘Passing’ group on our Rank pages and compare a range of pass-related metrics. Here we’ve sorted by crosses completed per 90 minutes (and added a minutes threshold to filter out the ‘only played one half’ outliers).
However, this might not capture what you want to do. Three of these top four are full-backs (a sign of the times that they’re not wingers), and the range of things that this position is judged by stretches beyond conventional, broad groupings.
With this in mind, we can create a brand new stat group specific to this purpose, with a mix of statistics from across the range available in the Toolbox. Given that most full-backs are attacking these days, let’s start off with defensive actions successful, a mix of passing and other on-ball actions, and then some of our new Twenty3 Advanced Metrics.
Each stat group needs to have a default to sort by and we’ve chosen successful defensive actions — these are defenders, after all, even if we want them to do much more than that. Immediately, we see that the full-backs and wing-backs making the most defensive actions aren’t the same as the ones completing two or more crosses per 90.
That doesn’t necessarily mean that these are all stay-at-home defenders though. Kyle Walker-Peters makes three progressive runs per 90, and at the bottom of this top 10, Joe Bryan is making 2.87 dribbles and 0.72 shot assists per 90.
We can order by any of the stats in the stat group though, so let’s go back to the more exciting full-backs. Our new advanced metrics include a statistic which is the percentage of sequences that a player’s involved in which end in the final third. It’s partly affected by the team, as we see when we sort by it (below), but part of it is about the individual as well.
We have the most-used full-backs of Manchester City, Liverpool, and Chelsea in the top 10, as well as additions from Leeds, Brighton, and Manchester United. The City players make a particularly interesting case for the ‘assigned role vs player performance’ balance: João Cancelo’s figure is several percentage points higher than both Benjamin Mendy and Kyle Walker, whose percentages are virtually inseparable.
Manchester United are similar. Although Luke Shaw isn’t quite in the top 10, 19.6% of the sequences he’s involved in end in the final third, very similar to Alex Telles’ 20.49%. It probably won’t surprise United fans to learn that Aaron Wan-Bissaka’s figure is much lower, at just 12.66%.
These stat groups can be used across other areas of the Football Lab. You can assign weightings to each individual metric in the stat group, for use in our Discover tool. For this example, let’s say that we really value defensive actions, and also open-play expected assists. We might also be less fussed about the percentage of sequences that the player’s involved in reaching the final third, because we’ve seen how that can be affected by the team they’re on.
Now we’re just into the fun of Discover. We’ll want to filter by the player’s position, of course, and let’s have a look at players who are 25 or under in the Championship, Eredivisie, Ligue 1, or Portugal’s Primeira Liga in 2019/20 (for a reasonable sample size). The stat group will be used to judge the players, and the weightings will help separate them according to our chosen values. Players who make more successful defensive actions will get more of a boost than those who make more completed crosses.
Although we put a higher weighting on defensive actions, the number of passing or creative stats in the group means that some players high up this list can get away with smaller figures for defensive actions successful. We could tweak the weightings further directly in Discover if we wanted to.
For now, let’s look at our top recommended player — Nuno Tavares — in more depth. By clicking on his name we get the option to go to see more of his stats in the Analyse page.
In Analyse, which has also experienced a recent update, we can choose which stat groups show on the page. You could perhaps create one stat group for use in Discover, and then another for further stats that are of interest, but not crucial to filtering players. For now, we’ll have the Full-backs group alongside the default Defending and Passing groups.
With the options to create and save stat groups, the entire suite of Football Lab tools become even more customisable. If the default groups that have always been in the Toolbox are generally good apart from a couple of stats here or there, then now you can create your own version. If there’s a position- or role-specific group, that can be created too.
Everybody does things differently. So that’s what we’re letting you do in the Football Lab too.
If you’d like to learn more about the Football Lab and its many different features, or our other tools within the Twenty3 Toolbox, please get in touch.