ROB GUTMANN is joined by the Liverpool Echo’s Neil Docking, John Gibbons and Andy Heaton to discuss the ongoing ‘war’ between the stats guys and the purists .
Its nerds vs jocks. Football is the winner .
UNWRAPPED: STATS, LIES AND VIDEOTAPE
// TAW SHOP
// TAW APP
// PARTNERS
 |
|
|
|
|
|
|
|
People into American sports often go on about ‘intangibles’ in response to the reams of stats that come out of basketball, baseball etc. These are the hidden factors that make up a footballer that some of the more regressive members of the media point out when talking about footballers. While they are impossible to quantify they are still relevant to the modern football landscape. Some one who can communicate effectively will not be picked up by stats but they can be vital to the way a defense coordinates itself. Someone who has a particularly buoyant personality may inspire his teammates to pick themselves up after going down by a goal. Some of these intangibles are difficult to see from the outside – to the point where you have to be a teammate or part of the coaching staff to appreciate. As supporters we may argue that his pass completion is rubbish, but the manager may argue that he occasionally throws in a couple of snide challenges that have the opposition midfield worried about holding onto the ball for too long.
I think in a Liverpool FC specific context, the club has gone in for a couple of general characteristics as opposed to choosing individual players by highlighting their hidden strengths. Brendan Rodgers talked about character (understatement alert) and personality. When you examine the players he wanted, they do (to some degree) exhibit these characteristics. You could also argue that when he was saying character and personality what he really meant was obedience. A player that listens in training, doesn’t cause any problems off of the pitch and is generally agreeable. Players like this make the managers job easier for obvious reasons. However, by all accounts Michelangelo was a bit of a pain in the arse, so were Beethoven, Dickens and Cruyff. If you start a selection process, where inherent ability is vital, by deselecting anyone who isn’t a choir boy then its going to be a very difficult exercise. Throw in the pressures of the market, fees and where players girlfriends want to live and the chances of finding a footballing genius who calls his mum twice a day became near impossible.
Statistics on the other hand provide you with the exact opposite scenario. I’m not going to go into because I only need one cautionary word: Mario.
Basically, this is just an extremely convoluted way of saying each player needs to be selected by looking at both the intangibles and stats. If you select players looking for distinct characteristics across the board you will end up with a lot of Lovren’s and fuck all Leonardo’s.
Two points.
First, it would be useful if the folks commenting on the value of statistical analysis and statistical forecasting in football actually had a working knowledge of statistics and statistical inference.
Second, if one is going to make the case for the value of watching games relative to using summary statistics, it would be ideal if he/she actually made a habit of really watching a lot of football; at the very least, a lot of matches in which the players the summary statistics of which one is doubting.
If one combines watching (and rewatching) a lot of football matches with sensible use of proper statistics, one is likely to be better than doing either one of those things in isolation.
Long time listener first time caller lol
Been a subscriber to TAW player from the very beginner and listen to almost all shows. This is the first time I have felt the need to post a comment.
I thought Rob’s show on the statistical analysis within football was terrific. I don’t necessarily agree with anything any of the guys said but I loved the topic and the different ways the guys looked it and chatted about it. It’s what I love about TAW player, I don’t have to agree with what is said but I know every opinion comes from a good place and is thought through. You just don’t get this level of rationality about football anywhere else. Everyone else is either so blinded by their own bias they can’t see the bigger picture or they are promoting their own agenda.
Seriously keep up the good work guys. Honestly I would pay a lot more than £5 for content like this.
A decent attempt to discuss the issues from non-analysts guys.
The early discussion on the vitriol of the journalists writing these articles and the reasons for that was excellent. The journalist’s defences for their pieces were that they were just highlighting the influence of stats in football these days. That would be fine except that the language they used boiled down to mocking the use of stats and the geeks who use them. Each article also concluded by basically saying ‘Proper football men are right and I won’t change my mind’.
Rory Smith’s article was good but if there are battle lines being drawn then they are only on the side of the luddites raging against change. They set up straw man arguments that posit that stats can’t tell you everything/anything about a player so they can knock them down and by extension destroy these evil stat wizards.
Left on the other side are the analysts who believe that as with many other walks of life there is merit to collecting, analysing and utilising stats to increase knowledge and gain more understanding. But I don’t think I’ve ever read an article from a football analyst which claims that there is no need to actually scout a player and certainly not one that openly mocks old men in flat caps standing in the rain on a Tuesday night scouting passion and leadership.
Similarly there was fair debate (mainly from Gibbo) on the criticism of the use of stats being boiled down to show me the evidence it works and if you can’t it must mean that the use of stats in general isn’t valid. That there is no leeway at all for other variables.
It’s basically the same as the argument over zonal vs man marking. Both systems have their advantages and disadvantages. Both can work if implemented properly with the right personnel. But if zonal marking (stats driven signing) fails because a player wanders out of his zone it is the system’s fault. If man-marking (scouting) fails because a player loses his man it is the individuals fault. Neither is infallible but rarely if ever are they used in mutually exclusive ways – much more often it is a combination of the two working together as they complement one another.
Unfortunately Andy Heaton puts himself firmly into the luddite category in this podcast as he did when he initially tweeted the Ashton piece under the heading of ‘Knives are out for the nerds – good’ – pretty embarrassing given the laughable article itself.
And this despite highlighting exactly how good stats analysis works in practice with his example of Mathieu Flamini – stats identified something he was good at, that stat was valued by the club and so they went to scout him and see for themselves before signing him.
I understand that he was there to act as the sceptic as he said himself but he spent most of the podcast setting up his own straw man arguments – ‘most stats analysts believe this’, ‘show me the evidence’, ‘Phil Babb had the highest pass completion rate’, ‘look what has happened at Brentford’.
1. No analysts I’ve read believe that stats are God at the expense of everything else – stats are just data just as scouting is data.
2. There is no evidence that it works as an absolute because no-one uses stats as an absolute.
3. Phil Babb had the highest pass completion rate in the EPL? Great – lets scout him and see why that is and what if anything he is doing different or better.
4. Brentford this season? Why does the initial struggle this season (with the admitted mitigating factors in injuries to key new signings) make what Benham is doing totally invalid? They are still punching above their weight after all. Does the success, this and last season, of Matthew Benham’s other club FC Midtjylland where stats are much more prevalent make this approach entirely valid or does it just work the other way when knocking stats down?
I think he is however correct that that kind of statistical analysis is probably going to give those in the middle to bottom of the pack more of a boost than those at the top but that is because there is more room to improve and more options available. It’s the reason that Southampton spending $50m well will see them rise 10 places whereas Liverpool spending $50m well will likely see them stay where they are.
Yes everyone can see that Aguero is a world class player now after 80 goals in 4 years in the EPL but Liverpool can’t afford him now can they? However, some statistical analysis combined with old-fashioned scouting could’ve indicated that Liverpool should have signed him from Atletico before he was a world star or even from Independiente as a 16 year old? That is what Liverpool and FSG are supposedly trying to do. Sign those players before every man and his dog can see how good they are just from their own eyes by which time they can’t afford them.
It’s an aid. It’s a tool. It’s a guide. It’s a starting point. It’s not the be all and end all. As Gibbo said if it is implemented as part of an overall structure at a club (Southampton or Swansea) that knows what type of football they want to play and the type of player they want to fit in and with everyone pulling in the same direction then statistical analysis is hugely beneficial. If a player has stats that you value that are way ahead of others playing in the same position then maybe you should go take a look and see why.
There is no magic bullet so critics should stop demanding to see evidence of one as proof before accepting statistical analysis as a useful tool. And no-one is advocating that it replaces scouting entirely except those who are claiming that it is replacing scouting!
Finally quite simply Liverpool should not be held up as an example of the use of stats to control transfer dealings because the failings of our transfer policy is not down to the use of stats but down to implementation, individuals and a lack of overarching strategy that everyone buys into.
Super stuff, well written and reasoned argument.
Fully agree, no analysts ever say that Stats are the be all and end all, they are an integrated tool along with scouts.
Drill down is the word of the day.
Didn’t seem to be much sympathy for the stats or the guys working in that field. Its made out that all these analysts are banging down the doors of clubs demanding a job. And surely clubs determine what the analysts are looking for.
We as a club haven’t done brilliantly for about 20 years with transfers, many years before ‘stats’ were involved and ‘real football intuition’ was the norm.
Just at the end there the Moreno dig, ‘Never heard of him’ – he was in the Spanish squad, touted as Alba’s replacement and very unlucky to miss the World Cup squad, with a lot of chat saying he should have. But yeah, never heard of him, deffo not a Prem Player that!.
I meant to write regards Moreno “….. unlucky to miss the World Cup squad, with a lot of chat saying he should have GONE”
Damn my impatient hands.
An interesting pod and you should invite Neil Docking back, he was good.
This is just the old debate about the old and the new, jumpers for goal posts. Look at Rafa and rotation. Stats is just another tool and like any new tool it is not the answer to everything but it has its place. Stats should flag up a player but then you have to look at the player as an individual. A classic example of this is Glenn Hoddle, a good technical player, but a million miles away from a Liverpool player at the time.
A more important question is will this player fit into your team and what should happen is all the strengths the player has should be reflected in the stats that you want.
What is difficult to prove by stats is personal courage, Clive Woodward took the 2003 England Rugby team took them to an army training camp to find out which players would crack under pressure. Look at the make up of that team and the individual courage each player have. This last quality is very important for teams that want to win things. Not so important for mid table teams that are never in the situation to win anything. A good 5 million pound player for West Ham may be useless for Liverpool and a 17 Million pound Liverpool player may be wasted at West Ham. West Ham preferred to play Brendan Mullian than Masherano
I think like all things the truth lies in the middle. When Twitter came out I was into stats and quoted them a fair bit. I’ve been through the honeymoon period and I’m now left thinking they’re a load of shite. I’m sure they serve a use somewhere, injury records perhaps or well, I can’t even think of another example. What put me off was Liverpool buying Mignolet because he was top of the stats table. I don’t care what stats tell me. I’ve got a pair of eyes and I watch football. I know when a player is good and when he’s not and it’s got nothing to do with what’s on a spreadsheet. I’d go as far as to say, I despise people who post stats on Twitter. It really winds me up. I’m sure they have a use somewhere but it’s true that I can produce stats for any argument and make them look good. They’re misleading and in the wrong hands can be dangerous. Much as I loved the 80’s I absolutely love most aspects of modern football but stats do me head in. Just watch players play a few times. If they’re good, sign them. If they’re not, don’t. Simple.
Excellent show. Despite how it may appear I don’t have time to listen to all the shows but this is one I’ve got in my diary to listen to every Wednesday evening. Love it.
I think you guys tried to be fair to both sides, and mostly it came out fair, but I think the discussion could have used at least one person who knew something about statistics and analysis like Paul Tomkins, Michael Caley, or Dan Kennett. Statistics is a really broad topic, even in the context of a football club, and labeling it as one thing does it and the clubs that use it a disservice.
The thing most people seem to have a problem with is statistics models applied to player recruitment. Statistics is useless without analysis, and any analysis is incomplete without experimentation and observation. Any club that used statistics alone to pick a player would be really dumb. Stats may reduce the size of the pool of players you need to look at, but you still need to send scouts to observe the players and confirm that they look as good as they sound. Or if your scouts have two players that appear almost the same, you could look at their stats to see if one had a slight edge.
Andy had a good point about “how do we know which statistics matter?”. Professional and amateur analysts will be the first to tell you that they’re still trying to figure that out. Some mean more and some mean less and it is ever evolving. What is repeatable year after year? Goals scored by a striker and goals saved by a keeper are nice measurements but they have no bearing on the next season out of context. Number of shots on target and shots from good locations is more meaningful. If you’re doing enough shots in the right situations eventually they’ll go in. The only way to prove any of this is to propose an idea and then observe it over time to see if results match up. Theories that don’t work are thrown out. It’s the scientific method.
What John said about the recruitment team and the manager having different ideas about what is important in a player is also absolutely true. None of this matters if the recruitment team is recommending the wrong players or if the manager doesn’t buy into it and won’t play the players selected. That doesn’t mean that analytics is failing or invalid, but it could mean that you need to replace your analyst(s) or your manager (see Brentford). You’re only as good as the people you have hired and everyone needs to be working together.
Enjoyed the pod.
Think the shear turnover of players at clubs now has put stats at the forefront. Then there’s the development of stats and the macros that you can measure and learning how to translate them & then the turnover of managers and how they want to use them.
So say for example we use the Viper system and have been using it for a while under Rodgers. If Klopp decides not to use it or use it significantly less than the quality of the learning and data generation would drop off until maybe the next guy comes in.
I think John was bang on in terms of how the club use stats rather than the incumbent manager uses them. If you think the average turnover of manager is 18 months so the club has a direction in playing style that effects your choice of manager and thus how you set the club up long term.
Then there’s the well West Ham are doing this with technology etc etc shouldn’t we be doing it. I guess there’s always pressure to seek out the micro gains both on and off the pitch and the example of Sky Cycling was a good one to pull out, they’ve taken it to a whole new level.
Did anyone see the article by this fella the other month re sleep.
http://www.theguardian.com/football/blog/2015/jul/23/nick-littlehales-the-man-who-taught-cristiano-ronaldo-how-to-sleep