StatsBomb CEO Update - October 2018
“Yo, boss… international break is a really good time for a piece from you talking about StatsBomb in general, and what’s coming next. Do you have time to do one of those?” asked Goodman.
“Sure, should be no problem,” replied yours truly.
*international break ends*
My only real defense here is that we did get two podcasts out over the break, which is probably one above expectation.
It has been a spectacularly busy year.
Anyway, Mike was right in that now seems like a really good time to talk about what we have done in the last year, and to look forward at what’s coming in the next year.
For those who are new around here, or who have their own busy lives to keep up with, it's worth a recap: What have we done?
Launched StatsBomb Data
This is a massive one. We decided that existing event data on the market wasn’t good enough, so spent ten hard months developing our own. Upgrades include
- Positions of GK and all defenders and attackers in the frame on every shot
- Pressures (off the ball defensive actions)
- Actions Under Pressure
- Passing footedness
- Pass height
- Ball receptions
- Intended Pass Recipient
...and much more.
All of these upgrades result in StatsBomb collecting about 1000 more events per Premier League match than the competition. Customers tell us we have created the world's best football event data and having done a massive amount of our own research in this area, we are forced to agree.
We're also collecting this data on unique leagues like Poland Ekstraklasa, Scottish Premiership, England League One and Two, and the completion of our new collection centre will allow us to double the number of leagues we collect in the coming months.
What else have we done?
Turned StatsBomb.com into a daily content site.
Added Free Women’s Data to the Resource Centre of StatsBomb.com, including FA WSL updated weekly.
Added Free FIFA Men’s World Cup data to the Resource Centre as well.
One of the problems for people interested in working with football data is access to football data. Another of the problems is a lack of qualified, interested women in an obnoxiously male-dominated field.
By releasing women's data, we are trying to support growth in the women's game while providing great data to anyone who is interested.
Oh... and we gave the world our 2018 FIFA World Cup data set to play with because that just seemed like a cool thing to do.
Begun to Unlock the Power of StatsBomb Data in StatsBomb IQ
This is a big one too, because all this new data delivered new information that we don't understand, which triggered new research and analysis. We now have xG models that incorporate GK and defender position into evaluating chance quality. We just passed 100K shots in the model data set and we’re seeing standard errors as good as anything we saw on competitor data with ten times as many shots.
And our data collection isn’t biased around big chances, which may not matter for media use, but really really matters when it comes to predictive models.
We also visualise pressure at the player and team level. This has allowed our customers to better evaluate their own players and potential transfers in a new light based on defensive work rate that has never been available before.
Customers tell us StatsBomb IQ is already the best football analytics platform on the market*, but the only way it will stay there is if we keep relentlessly improving it, which is why we do upgrades at least monthly that include new data science research, new visualisations, and basic quality of life improvements.
The way I view IQ is that it's our Amazon Prime - we're going to keep packaging in so much good stuff into the platform that it becomes a value proposition customers simply can't live without.
*I know it sounds like marketing, but customers are almost weirdly positive about StatsBomb stuff even when my first question to them is usually, "Okay, tell me what you hate." Absorbing and incorporating negative feedback about our work is hugely important to our future improvement.
We're not perfect. We know we're not perfect. We tell people up front we're not perfect. But we are really honest about what our imperfections actually are, and how we're going to fight to constantly improve them.
So that’s what we’ve done since May… what’s next?
The StatsBomb IQ Goalkeeper Module
Those of you who listen to our podcast know that I have been very cautious about using stats to help evaluate GKs over the years. I never felt comfortable the data we had access to allowed us to add meaning to GK impact on stopping shots. Now that we have GK position on every shot, that has changed. This week, we release the first phase of our GK Module into StatsBomb IQ that will help ourselves and our customers evaluate GKs in a way no company has before. (And yes, slightly to my chagrin, data scientist Derrick Yam has forced me to add GK radars to the spec, but that will have to wait until we release phase 2 next week.)
The amount of stuff we don't know or understand about goalkeeping, at least from a statistical perspective, is astounding. xG, post-shot xG, optimal positioning on shots, range factors, cross collection activity and its value, distribution, and so much more probably matter, but analysing and quantifying it is all new, and the perspective is NOT the same as it is for outfield players.
I’m sure we’ll talk a lot more about this in the coming weeks, but this is a fundamental game changer in player analysis and recruitment.
Unlocking More on Actions Under Pressure
I think I said it at the Data launch, but I view Actions Under Pressure as the most dramatic improvement we have made on this data set. Understanding what happens at the team and player level when players are put under pressure versus how they behave when free is a monumental step forward when it comes to understanding the game. Our analysis department has finished a significant project in this area, and we're just waiting for free development cycles to add it to StatsBomb IQ, where it will trickle into our analysis on the website and off.
Engine Room 2.0
Engine Room is the tactical suite inside StatsBomb IQ, though at the moment, the size of it is more like a studio apartment. We did an initial release last year, but I put further development on hold while waiting for pressure information to come online from SB Data. Once the GK Module is out the door, Engine Room will probably be the last major set of development sprints we have for 2018.
Why does it matter? Because using data to profile team tactical information, from broad top down trends to granular player level insights, is one of the most valuable possible uses I can think of for football data. We are committed to understanding the game better ourselves, but also in making that understanding available to coaches and analysts in easy-to-digest formats.
At the end of it, ER 2.0 should be the best tactical profiling tool for team analysis and opposition scouting anywhere.
And as I noted above, we'll then let our customers tell us if we got there or not.
Secret Project Due End of 2018
We are doing a massive amount of analysis on right now that we're delighted to be involved in and quite proud of, but I can't tell you any more about it until everything is complete.
No one has done this type of project before, but the hope is that the report will be released at the end of the year or early 2019.
Last year, we held a private launch in May for SB Data, and it was a complicated little dance to get people to show up to something that was mostly secret until launch day. We mostly succeeded, but we also recorded all the talks for posterity. (Top tip: Always have a back up plan.)
This year we’ve pivoted a bit and decided to host our own conference, probably in May again.
The good news is that you - the general you - are invited. That's because we're making our conference focusing on analytics, innovation, and new research in football data open to the public.
There will also be some segment of tickets reserved for specific yous as well (helloooo customers and team invites), but the goal here is to deliver a day long conference with outstanding speakers in the sports data space. Additionally, with a nod to our data launch, this will remain the place where we announce new, cutting edge shit that we are working on, and that is ready for primetime consumption.
I’m going to stay light on further details here until we get everything completely nailed down, but know that we’re doing it, it’s going to be great, and you should be excited and want to attend because you will learn new things and probably laugh once or twice as well.
(Possibly with/at James... Because there will be a StatsBomb Pod Live.)
So there you have it - a monster year leading to even bigger plans for the future. I feel like I have been running a million miles an hour since this time last year when we were first developing the prototype for SB Data, but that's what we needed to do to get here. That choice unlocked the entire football analytics space for us, and allowed us to deliver fantastic new insights to our customers and the world that were simply not available before.
Our team is doing great things at StatsBomb, both in the public-facing work you guys see, and in the private stuff available only to customers, and I could not be more proud of the progress we've made in a year.
Keep watching this space though, because we are still just getting started.