Wolves 3 - Bristol 3 the more excitement in one game than the whole of last season ve

[Johnny75]

Football is so random that it is very difficult but there are clubs doing it very well and I think the next area of investment for clubs will come in analytics. You can get ahead of teams at the moment and then people will start to play catch up.

If teams are continously massively underperforming their expected goals then it is a good way to highlight that your forwards are a bit crap at finishing, or they just need to improve at something specifically. Whereas if all their shots were coming from somewhere you are far less likely to score you can focus on how you would create better quality of chances in better positions. So I definitely agree with you on that point but I would rather be creating a lot of opportunities and missing them than creating very little. So a team like Burnley quite often outperform their 'xG' but in the long term I don't think they can continue to do that.

Although it's a bit of a 'craze' as you say it is fast becoming the norm and is definitely here to stay.

I find the stats an interesting aside and I know that a lot of it is driven by the Moneyball phenomenon. The difference between the two is isolation, much like cricket. I know they're really struggling in American football and Basketball with stats based approaches and their roles are much less fluid than football. It'll be interesting to see how the stats pan out and how useful they are. As you and Mark have said, what's the point in having stats if they aren't defining the trend and proving effective at predicting the actual scoreline?

I'm sure there are mathematical minds far better than ours at work on this but I think the clubs may be looking for something where the patterns are so random and have so many variables it would be the equivalent of mathematically modelling the Universe. And I'm not sure Wolves v Bristol City merits that.

 

[machin05]

I like how we're effectively trying to create a wolf head with our pass maps. Maybe that's been the key to success all along??!

 

[Dublin Drummer Wolf]

As I expected to see, based on the quality of chances created most times we win this game. https://twitter.com/11tegen11/status/907864544386658304

Our pass map looks great you can really see the shape coming along.

Gold star for Leo and Silver for Doherty!

 

[Dublin Drummer Wolf]

Football is so random that it is very difficult but there are clubs doing it very well and I think the next area of investment for clubs will come in analytics. You can get ahead of teams at the moment and then people will start to play catch up.

If teams are continously massively underperforming their expected goals then it is a good way to highlight that your forwards are a bit crap at finishing, or they just need to improve at something specifically. Whereas if all their shots were coming from somewhere you are far less likely to score you can focus on how you would create better quality of chances in better positions. So I definitely agree with you on that point but I would rather be creating a lot of opportunities and missing them than creating very little. So a team like Burnley quite often outperform their 'xG' but in the long term I don't think they can continue to do that.

Although it's a bit of a 'craze' as you say it is fast becoming the norm and is definitely here to stay.

Very true, and not just for football. Detailed analysis generates insights and allows teams to measure trends and predict how players and teams will set up when using certain players, or which areas of the pitch to exploit.

We use Analytics at work to monitor trends in marketing and how users interact with digital ads and it's really interesting, and as you said, its only getting bigger.

 

[Deutsch Wolf]

Analytics and metrics currently work best in games that are essentially a series of set pieces. Far more difficult to quantify football in that respect, that's why Moneyball concepts generally don't work as a rule.

 

[GLASGOWOLF]

I'm not a Miranda fan so far, I've seen him weak in the tackle, get caught under the ball and switch off allowing runners in behind. Last night though he was (guessing Johnny [emoji6]) subbed because we were on the front foot trying to win the game and Saiss could push up into midfield and be better on the ball.

Made 2 of the best tackles from centre half I have seen against Brentford, 2 that would not have been out of place at prem or International level.

 

[GLASGOWOLF]

He's just jumping for the ball, it's a contact sport FFS.

Under modern day rules it's a penalty all day long

 

[MARKakaJIM]

Football is so random that it is very difficult but there are clubs doing it very well and I think the next area of investment for clubs will come in analytics. You can get ahead of teams at the moment and then people will start to play catch up.

If teams are continously massively underperforming their expected goals then it is a good way to highlight that your forwards are a bit crap at finishing, or they just need to improve at something specifically. Whereas if all their shots were coming from somewhere you are far less likely to score you can focus on how you would create better quality of chances in better positions. So I definitely agree with you on that point but I would rather be creating a lot of opportunities and missing them than creating very little. So a team like Burnley quite often outperform their 'xG' but in the long term I don't think they can continue to do that.

Although it's a bit of a 'craze' as you say it is fast becoming the norm and is definitely here to stay.

I find the stats an interesting aside and I know that a lot of it is driven by the Moneyball phenomenon. The difference between the two is isolation, much like cricket. I know they're really struggling in American football and Basketball with stats based approaches and their roles are much less fluid than football. It'll be interesting to see how the stats pan out and how useful they are. As you and Mark have said, what's the point in having stats if they aren't defining the trend and proving effective at predicting the actual scoreline?

I'm sure there are mathematical minds far better than ours at work on this but I think the clubs may be looking for something where the patterns are so random and have so many variables it would be the equivalent of mathematically modelling the Universe. And I'm not sure Wolves v Bristol City merits that.

It's definitely something that's developing, and surely only going to get more detailed, a few years ago no-one would've cared about any stats at all and now we've already passed through basics of people being obsessed with possession share from the Guardiola/Barca years into far more specific players stats and now looking more at interactions like the pass maps.

At the moment though it's like trying to spot a needle in a haystack to pick something relevant from the data, like you look at Wolves falling short of their expected goals and an easy conclusion is to say Bonatini needs to work on his finishing but then you check that pass map and you see he is significantly deeper than Jota/Cavaleiro so then you've already got different options like working on Jota/Cavaleiro's finishing or working on Bonatini's movement to try and get him back in those advanced areas more often. It's very difficult to pick these things apart in a game like football where pretty much every action will have widespread effects elsewhere, you try and push Bonatini further forward to improve chance conversion but you may hamper the chance creation in doing so, it's all a bit of a minefield.

At the moment they're useful for identifying a trend but there's still a lot of interpretation to the causes and possible solutions that requires thought to piece together.

 

[Elephant Pyjamas]

I think the only thing Danny beats Rode on is physicality and headers. More to come from Miranda whereas Tuesday was about Batth's best.

Which are pretty bloody important if you are a CB!

Sent from my SM-G935F using Tapatalk

 

[MARKakaJIM]

Which are pretty bloody important if you are a CB!

Sent from my SM-G935F using Tapatalk

Depends what you're playing against doesn't it, wouldn't get him very far against a forward who's good running with the ball and keeps getting it played into feet, as we've seen before with Batth.

 

[Johnny75]

Depends what you're playing against doesn't it, wouldn't get him very far against a forward who's good running with the ball and keeps getting it played into feet, as we've seen before with Batth.

Leko rinsed both Batth and Doherty twice in the first half. No end product luckily enough for both of them

 

[Dublin Drummer Wolf]

Analytics and metrics currently work best in games that are essentially a series of set pieces. Far more difficult to quantify football in that respect, that's why Moneyball concepts generally don't work as a rule.

I get you, but it also depends on which metrics you are looking at, as certain metrics will be consistent across all games, formations, tactics, possession etc.

You then have the different variable such as number of men on the pitch, weather conditions etc. But with the massive amount of games taking place each season, all of these can be evaluated and placed into their own groups and then each attribute can effectively be measured against a 'control group' (other teams playing the same formation with 10 men in the wind for example).
All of these things can be easily computed through software data and thats what we are seeing now, through the likes of that 11tegen11 guy.

Everything is measurable, getting accurate results and being able to use them going forward is what makes the difference.

 

[Lord Knows]

My argument throughout this discussion has been that Batth played better against Bristol City than Miranda. Those stats help to back that up.

Sent from my SM-G935F using Tapatalk

They both bring different qualities to the line up and complement each other, so unless your Tony Pulis you wouldn't be selecting like for like players in your defensive line.

 

[Johnny75]

I get you, but it also depends on which metrics you are looking at, as certain metrics will be consistent across all games, formations, tactics, possession etc.

You then have the different variable such as number of men on the pitch, weather conditions etc. But with the massive amount of games taking place each season, all of these can be evaluated and placed into their own groups and then each attribute can effectively be measured against a 'control group' (other teams playing the same formation with 10 men in the wind for example).
All of these things can be easily computed through software data and thats what we are seeing now, through the likes of that 11tegen11 guy.

Everything is measurable, getting accurate results and being able to use them going forward is what makes the difference.

In principle you're right but it doesn't work in practice as the guy you have linked shows. Nobody has done it yet and they've been at it for a decade now because the variable are too great. In simple terms you are right but that data doesn't reflect match outcome. Books like Soccernomics prove that if you have data and don't know what it is good for then you come up with the wrong outcome. The best example I have of this is the guy in Denmark who did just what you are talking about and was a success in Denmark but when he and Benham at Brentford tried the same thing in a more competitive league (i.e. the Championship) it failed to work because of the variables.

The difficultly is really in what to measure and how to link it to the outcome. If anybody works that out then they might be worth chatting to for your accumulator that week.

 

[Deutsch Wolf]

If I were scouting a player then the first three things I'd be looking for are:

- First touch
- Positional responsibility
- Awareness, on and off the ball

There aren't numbers that can really measure any of that. Maybe a heat map for positional stuff but even then it's going to depend on the match situation. There'll have been games in 08/09 where Kevin Foley was miles up the pitch for most of the game but that doesn't mean he was indisciplined, we just had all the play and he supported the attack. Whereas the same map for Zubar clowning around against better players would indicate that he was a brainless fucking idiot.

 

[MARKakaJIM]

Even if you're nailing down things like formation, weather, etc. to make your comparisons more accurate you've still got every individual on the pitch as another variable to consider, so what happened for say Chelsea playing 343 against 10 men on a sunny afternoon could be completely different for what Wolves would have to do against 10 men playing the same formation with the same weather conditions. Even if you were comparing it purely with Wolves in the same circumstances then you'd potentially have different players from one game to another, fitness levels may be different from one game to the next, it's so difficult to get a real like for like comparison in football.

 

[MARKakaJIM]

If I were scouting a player then the first three things I'd be looking for are:

- First touch
- Positional responsibility
- Awareness, on and off the ball

There aren't numbers that can really measure any of that. Maybe a heat map for positional stuff but even then it's going to depend on the match situation. There'll have been games in 08/09 where Kevin Foley was miles up the pitch for most of the game but that doesn't mean he was indisciplined, we just had all the play and he supported the attack. Whereas the same map for Zubar clowning around against better players would indicate that he was a brainless fucking idiot.

I imagine for scouting purposes you'd use the stats for more headline figures, pass completion, goals, assists, whatever else and then you'd start actively scouting them to pick up on the points like you mention.

 

[Deutsch Wolf]

I imagine for scouting purposes you'd use the stats for more headline figures, pass completion, goals, assists, whatever else and then you'd start actively scouting them to pick up on the points like you mention.

Yeah, sure. But I think the point of Moneyball is that you purely trust the numbers. So you're picking up players that don't outwardly look anything special, but the metrics say they'll fit what you want them to do. If anyone ever comes up with a system that works like that in football I'll be deeply impressed, it 100% does not exist at the moment.

Apparently Arsenal's board are a bit annoyed that they installed a system at Wenger's insistence, it came up with Lucas Perez and Shkodran Mustafi and they've both turned out to be sub-par.

 

[Dublin Drummer Wolf]

I didn't realise we were talking about using analytics to buy players, I was talking about using it for insights and trying to create predictive models about how teams play and where and who they look to create from. As Mark said, there are many variables to consider, but thats the thing about Analytics, it measures details about what happens in games, who runs where, who passes to who etc..... it's just facts, it's not a time machine that will guarantee anything that might happen in the future.

At work, we use Analytics to look for trends: if 100,000 people saw an ad on YouTube, and 10,000 visited a site after clicking that ad, does that mean 10% of YouTube users will visit the site when they see the same ad?
No, theres other variables to look at, time of day, device, location etc.

We look at the top numbers and the bottom numbers to work out the average and then look at historical data in order to create trend paths from there, to be used as insights. They don't guarantee anything ever, but can be used to help advertisers make big changes quickly and stay ahead of the game, and it has to be something similar in football.

There's literally so much data out there, for every player in every game in every league, it makes sense to collect this data and create models designed to give managers an extra edge over their competitors. Of course, it won't get it right all the time, trends are there to be bucked and data gets skewed, but in the majority of the time, once you have enough data, trends will adhere to a glide path over time and using that info can help teams compete and not get caught out when a manager changes shape or they go down to 10 men.

I read somewhere recently that Man City use analytical KPI's to determine bonuses, so if a player does not complete a certain number of passes per season etc, they lose out on a performance bonus. I would imagine that this will become the norm in football soon enough as it is becoming the norm across other industries.

 

[Deutsch Wolf]

Don't get me wrong, I'm no Luddite. The whole thing interests me.

However you don't need incredible analytical software to tell me that between January 2016 and May 2017, Wolves were incredibly vulnerable down their left hand side. I've no doubt the numbers will tell you that. I'll tell you why it is, it's because we had Matt Doherty there and he is a shit defender and not a left back.