New Blades Data Dashboard

[Steve Mackan]

I've made a new data dashboard on The Pinch.

Hope some fellow nerds get something out of it. But first, for all the people who like a good graph, I'd be really keen to get any thoughts on how it all works.

I'm biased, but I don't think any other club really has this. Yes, the dashboard comes entirely from publicly available data. But what we (Riley and me) noticed, was that this data is often so often under-explored at club level, which makes sense because it's bloody labour of love to do. So, we've tried to dig a little deeper whilst providing data that can actually be explored.

This isn't a series of screenshots or tables you can't sort. We've built something that (with fingers crossed it all works) is a proper statistical sandbox. You can sort tables, search tables and interact with graphs. You can isolate stats from one game and compare them with another. Altogether, there are 12 charts/tables, covering Team Stats, Match-by-Match stats and Player Stats.

Now, I'm totally prepped for the xJUYEC (expected 'just use your eyes' comments). And I have a lot of sympathy with anyone who dislikes the way data in football is covered. But our intention with this is to make Blades By Numbers a bit of a playground; the entire opposite to the StatManDavification in mainstream media, with nobody shoving the xG of a shot down your throat.

Those are a lot more words than I'd planned. Hope a few of you have a play around! And any feedback hugely appreciated.

Ta, Sam!

 

[DifferentClass]

Love a good bit of data porn

 

[Geordie Blade]

Phwoooaaarrrrr. Love it.

 

[Cassius Kray]

This is interesting, thanks. There are some stats I've not seen before and being able to easily compare them (e.g. across matches) provides additional insights. Appreciate the definitions under each visual as well.

Where do you get the data from (I see 'Opta' but I don't know where Opta publish this)? Is it a data feed or are you planning on manually updating it? Is there any particular reason you chose Datawrapper to create the visuals?

For some of the XG data, such as finishing performance, I wonder whether it would be useful to include an additional column for performance as a %? May not make a big difference with the current values but it could be interesting to see who has the worst performance relative to their opportunities, as well as the worst performance overall.

A minor point but you might wish to consider bars over the doughnut charts for the player share visuals - radial visualisations are generally considered bad practice for anything more than couple of values because it's more difficult to intuit sizes and therefore make comparisons (it's not immediately obvious, for instance, that Burrows' share of xg is almost exactly half of Peck's).

 

[Steve Mackan]

This is interesting, thanks. There are some stats I've not seen before and being able to easily compare them (e.g. across matches) provides additional insights. Appreciate the definitions under each visual as well.

Where do you get the data from (I see 'Opta' but I don't know where Opta publish this)? Is it a data feed or are you planning on manually updating it? Is there any particular reason you chose Datawrapper to create the visuals?

For some of the XG data, such as finishing performance, I wonder whether it would be useful to include an additional column for performance as a %? May not make a big difference with the current values but it could be interesting to see who has the worst performance relative to their opportunities, as well as the worst performance overall.

A minor point but you might wish to consider bars over the doughnut charts for the player share visuals - radial visualisations are generally considered bad practice for anything more than couple of values because it's more difficult to intuit sizes and therefore make comparisons (it's not immediately obvious, for instance, that Burrows' share of xg is almost exactly half of Peck's).

Thanks CK - really appreciate those thoughts.

1. The Data - it comes from publicly available Opta sources, mostly FBref but FotMob for shot-by-shot xG used for the 0-0 tables
2. Datawrapper - because it's the only such software to be integrated with Substack; and, to be fair, it's a pretty solid tool, and see #3
3. The extent of manual work is limited to building everything in the first place; now it's all setup we can paste raw data into googlesheets and datawrapper has the functionality to pull the info from google - which is handy for semi-automation
4. Extra % column - really like this idea, and definitely something we'll look to add as we go along and note improvements down
5. Noted re the donuts - I suppose what I liked about these charts that it would clearly show where most of our xg/xA comes from, but you're right that it makes it harder to intuit differences; another one for the list
6. Thanks very much for the thoughtful responses!

 

[DI36O]

This is great that you've done all this specifically for united so thank you.

I like looking at data. I find it difficult to make many inferences from it in football especially early in the season. That Hull game for instance, that baffles me how our stats are so good. And 4 for prevention against Bristol City seems insane. There's nothing wrong with data itself though, its just the conclusions we find ourselves drawing from it.

Anyway thanks again for making this and I'll keep checking it over the season. Hopefully some positive patterns emerge.