Haha dugout! yes. Jr Performance Analyst doing first team cone setups
United appoint Dominic Jordan as director of data science
- Thread starter Tom Van Persie
- Start date
I agree with what you're saying, and it's extremely important for data analysts to understand what they're applying the data towards. And when it comes to recruitment, data analysts do need guidance from the football men to best utilise their approach towards data. I call it the guiding principle, which is a guidance that should come from the head coach.
I've in the past shared the thoughts of the head of data science at Ajax (Vosse de Boode) on this forum. And according to her, data is useless unless there's football people guiding her on how to apply data towards the type of football they want to implement.
If anyone is interested (need more than @Revan, the nerd) I can create a separate thread in the Football forum of analytic specific initiatives across football in recent years. A lot of work going on trying to better quantify football metrics beyond xG.
Im interested!
do it Please
I have no clue what you just said, but I think I would enjoy reading it.
I’d like to partake. The challenge is that a lot of the more advanced metrics like expected threat as an example aren’t freely available.
shame your insight doesn't spot humourWe are so fortunate to have such insightful posters around here
Will you embrace "you're just a bus stop in Stretford?" when the RaRa's sing it at you in the same way we have with the Hounslow version though?
This would be interesting. Is the data gained through ML and Python? I have a friend with a BSC in statistics so he uses R for whatever data he needs for his customers.
I reckon’ there will be very little embracing done.
"Nobody" thinks that? I wish that was the reality, but read some of the responses and you will see - plenty of people think that.
Case in point:
Anyway, United has had data team for years. Not much will change with this new appointment. No data can fix having a bunch of toxic, arrogant, and lazy players.
Our only hope is Ten Hag who will ship these a-holes out.
Last edited:
On the analytics side, what you can do with one, you can do with the other. Won't make a difference unless you're talking about end to end solutions & deployment where Python stands a clear winner. What will be more interesting than the programming language or the technique is the type of data & the quality of data collected for the analysis.
I don't want to be associated with nerds. So that's a no from me
The point is that this Liverpool team has been largely built with the data-model being a strong contributor to transfers.
Will it make much difference to this current group of players? Probably not. But it could help decide which positions to prioritise improving on, and more importantly which players we should be looking to buy for these positions. There are signs that things are changing in terms of organisation throughout the club, and (especially with Ralf and ETH hopefully pushing in the same direction) hopefully this is something that will have a larger contribution going forward.
Laughing so much at you claiming that’s what I meant. Absolutely clueless attempt at covering for your own idiotic statement
Ruder your response more it shows your insecurities. Confident, intelligent, and emotionally stable people don't need to bark at people, even the ones they disagree with
Yes I understand. My point is: the role of data in decisions is largely exaggerated. Every team at that level collects and processes data, to some extent but City and Liverpool are strong because of Pep and Klopp first and foremost. I don't believe that there is material difference in the quality of data processed by them and United. Sure they have made more football decisions and we are making decisions driven by Glazers, but that is not anything head of data team has any influence over. They just collect, process, and present data. If some moron keeps Lingard and Jones way past the time we should have, or doesn't fire Ole quickly enough, or keeps Maguire as captain, it isn't because we had a poor data team
I’m long past caring what people think about me, particularly the ones who’ve completely made up my position on a subject because of they realised they were wrong and need somebody to paint as more wrong than them. Fact is that kind of person isn’t worth dignifying with time because they’re clearly racked with their own intellectual insecurities and crave being “right”
That's a lot of response for somebody who doesn't care. Cheer up, man. Seriously. Why are you so bitter. Feck this shit
Okay so let's say I want to see the data of Dalot using his left foot to make a cross during the whole season:
1. How do I go by collecting this kind of data?
2. How do I present it to the end user?
Thought poor Dominic had gotten the chop when I saw this was bumped.
Can honestly say I've never heard of him until now anyway!
Can honestly say I've never heard of him until now anyway!
Python and R are just programming languages, so you can use either. Python is more used mostly cause it is a better language, and it has a lot of good libraries (outside of deep learning libraries, so had R).
Kudos to you and @bringbackbebe for the explanation. Any good tips/links for learning material? This is a very interesting topic.
For Python, there is an absolutely great course of MIT given on MIT OpenCourseware and also edx. The edx version is better cause it also has graded assignments. I think it might be called something like ‘Introduction to computer science and programming with Python’. In edx it has a sequel that deals with some basic data science libraries like numpy, pandas, sklearn etc (In MIT opencourseware they are joined).
For Machine Learning, there is nothing to beat Andrew Ng’s CS229, but you need decent math background. If you don’t have it, then he has a version of it in Coursera that is very watered down and easier, but still quite nice. Or you can just directly read some ML book (Murphy’s; Bishop’s; or Hastie/Tibshirani are probably the best).
1. There's a number of ways to do this.
-> A reflective approach where a follow through is done on the coach's hunch. For eg, coach observes that Dalot is a weak left footed crosser of the ball & asks the analytics team to study the the last few matches to see his pass percent by each position, neutralizing other factors. The team focuses only on this.
-> A prospective approach where you have an army of data entry operators from developing countries that break down every aspect of the game into quantifiable statistic as a series of line entries. For eg: Player, position where he recd the ball, number of touches, position he passed the ball to, if intercepted?, which foot was used?, type of pass etc etc etc. This is then used to build a mathematical model to suggest to the coach how Dalot should play. Problem with this is it's labor intensive and there's a bigger chance of errors.
-> An automated model that replicates the above using computer vision. There's a lot of gas floating around but doubt anyone actually uses this to record data yet, even in big tech companies outside of football. Computer vision models are still rudimentary.
2. The end user here is the coach and eventually the player. In a reflective case, it'll either be a yes or no (Dalot is a weak left footed crosser of the ball). In the prospective case, since we're looking at this holistically, it'll be a suggestion - Dalot is a weak left footed crosser of the ball in general, but it's ok for him to cross when he has enough time on the ball and the person marking Dalot is not a defender.
Given the current season, you might as well just appoint Jordan herself.
Guys… I know random bump. BUT, I was just curious about our Director of Data Sciences and checked his resume.
2020-22 - Director of Data Science at N Brown Group
2011-20 - Chief Data Scientist at INRIX (logistics)
2006-11 - Head of Innovations at ITIS holdings (traffic)
1999-06 - Data Systems Manager at NUS (Student services)
I understand there aren’t very many sports data scientists out there of high caliber, but this guy doesn’t have any sporting background and has worked at… let’s say not well known positions. But he’s a Manchester lad and fan and has a degree from Cambridge. So clearly smart.
Then I checked Liverpool’s Lead Data Scientist… William Spearman. Let me start at his beginning to give proper context:
2005-08 - 2 Bachelor’s Degrees - Arts and Physics! University of Dallas
2007- Summer Student at CERN
2008- 09 - Fulbright Scholar at University of Geneva!
2009-14 - Masters in Physics, PhD in Particle Physics! Thesis: TO MEASURE THE HIGGS feckING BOSON!
2011-13 - He was a graduate student AT CERN TRYING TO FIND THE HIGGS BOSON!
Higgs Boson was detected in 2012… HE WAS THERE!
He then completely switched tracks to football! I mean you achieved your peak in particle physics what else would you do!
2014-2018 - He worked on the concept of pitch control visualisation. He studied passing, open space and scoring to power his model. His paper ended up at MIT Sloan as a breakthrough in using geo-spatial data to model pass probabilities.
2018 - Liverpool and the rest is history.
Compare the two and tell me how you feel.
2020-22 - Director of Data Science at N Brown Group
2011-20 - Chief Data Scientist at INRIX (logistics)
2006-11 - Head of Innovations at ITIS holdings (traffic)
1999-06 - Data Systems Manager at NUS (Student services)
I understand there aren’t very many sports data scientists out there of high caliber, but this guy doesn’t have any sporting background and has worked at… let’s say not well known positions. But he’s a Manchester lad and fan and has a degree from Cambridge. So clearly smart.
Then I checked Liverpool’s Lead Data Scientist… William Spearman. Let me start at his beginning to give proper context:
2005-08 - 2 Bachelor’s Degrees - Arts and Physics! University of Dallas
2007- Summer Student at CERN
2008- 09 - Fulbright Scholar at University of Geneva!
2009-14 - Masters in Physics, PhD in Particle Physics! Thesis: TO MEASURE THE HIGGS feckING BOSON!
2011-13 - He was a graduate student AT CERN TRYING TO FIND THE HIGGS BOSON!
Higgs Boson was detected in 2012… HE WAS THERE!
He then completely switched tracks to football! I mean you achieved your peak in particle physics what else would you do!
2014-2018 - He worked on the concept of pitch control visualisation. He studied passing, open space and scoring to power his model. His paper ended up at MIT Sloan as a breakthrough in using geo-spatial data to model pass probabilities.
2018 - Liverpool and the rest is history.
Compare the two and tell me how you feel.
I'm not sure what you are trying to prove here. Because someone worked at CERN, he's better at Data Science?
Our Director of Data Science has +20 year experience in ... data science.
So we got one with 20+ years in management and leadership roles with experience at the director level in the data science field.
Then we got a 10+ year student who studied a lot of physics and doesn’t seem to have had an actual job before?
Compare the two and tell me how you feel.
The main thing is that Liverpool were doing this stuff over 4 years ago while it sounds like we're just starting out. And we wonder why their scouting is so much better than ours.
No concern's. Not sounding wishing to be too disrespectful here but I assume your argument is Liverpool is better because Will Spearman has on paper a more high profile CV ??
Ultimately a data scientist is exactly that, science. It follows a very rigid discipline of collecting data and analysing data. Analysing data on the Higgs Boson or or data on how many people drive into the Trafford Centre on a Weds morning between 10.30 and 12.00 will follow the same discipline. Also worth noting that Will Spearman has a speciality in Physics that would lean him into projects like Higgs Boson.
Ultimately the data is the data, what is more important, is how we analyse it and in turn how that impacts what is done in training and in turn, matches. The data scientist is just a cog in the machine.
Commercial data science is basically marketing with spreadsheets. A scientist from CERN on the other hand, if he switched to football would be doing it out of a genuinely different motivation and shelf of ideas.
I think Jordan is hired to build our data science department. Like Hunter at Liverpool. Spearman is just the guy doing the work. In other words, Jordan needs to hire a Spearman.
No I’m trying to say Spearman is a genius for one, and spent 4 years actually applying data science in football. Unlike our guy.
Uhh… what?
All I’m saying is that Spearman actually applied his genius IQ in football and innovated with visualising space and space creation on a football pitch.
The man who runs that team there is an innovator and path breaker. Not a management man, which our guy seems to be.
So unless he’s hired some incredibly smart football data scientists, our operation will take a few years to truly come to fruition.
Bingo.