BladesPod: the last few weeks of SUFC

[Beans]

How do S24SU,

New podcast is up - I had a good chat with fellow Blade and stat enthusiast Jay aka Blades Analytics on Twitter, which I just put live for your listening pleasure.

We cover quite a bit, including:

• The defeat to Fulham (01:30), the general state of the Championship, and where United sit within it
• A strange few weeks for United (12:15) including the Hull meltdown, Wilder's comments after, takeover talk
• Bouncing back at Reading (21:20)
• Breaking down Moore and Blackman, and how lucky we are to have two good goalkeepers (23:45)
• Some stats about our still-excellent defence (28:00)
• Making the case for Leon Clarke's form (37:40)

Enjoyed making this one, Jay's raises some excellent points and has got stats coming out of his ears. Hope you enjoy listening to it - and thanks as always for any feedback, likes, shares, etc. UTB!

SoundCloud:

iTunes: https://itunes.apple.com/gb/podcast/bladespod/id1309198119?mt=2

 

[Sothall_Blade]

I've been looking forward to this. I will try to listen on the way home tonight. Thanks.

 

[Beans]

I've been looking forward to this. I will try to listen on the way home tonight. Thanks.

Cheers for listening!

 

[LoughboroBlade]

How do S24SU,

New podcast is up - I had a good chat with fellow Blade and stat enthusiast Jay aka Blades Analytics on Twitter, which I just put live for your listening pleasure.

We cover quite a bit, including:

• The defeat to Fulham (01:30), the general state of the Championship, and where United sit within it
• A strange few weeks for United (12:15) including the Hull meltdown, Wilder's comments after, takeover talk
• Bouncing back at Reading (21:20)
• Breaking down Moore and Blackman, and how lucky we are to have two good goalkeepers (23:45)
• Some stats about our still-excellent defence (28:00)
• Making the case for Leon Clarke's form (37:40)

Enjoyed making this one, Jay's raises some excellent points and has got stats coming out of his ears. Hope you enjoy listening to it - and thanks as always for any feedback, likes, shares, etc. UTB!

SoundCloud:

iTunes: https://itunes.apple.com/gb/podcast/bladespod/id1309198119?mt=2

Really enjoying these – I think Jay also makes a really good addition to it, interesting analysis from both of you.

 

[broomhillblade]

That's a relief. When I read "the last few weeks of SUFC" I thought it was warning of a doomsday scenario.

 

[Beans]

Really enjoying these – I think Jay also makes a really good addition to it, interesting analysis from both of you.

Yeah he's brilliant, a great follow on Twitter for United-based insight. Think SUFC need to get him hired as an analyst

That's a relief. When I read "the last few weeks of SUFC" I thought it was warning of a doomsday scenario.

Bloody hell Didn't think of that!

 

[I should coco]

Really enjoying these – I think Jay also makes a really good addition to it, interesting analysis from both of you

Hear hear!

Another sensible and considered analyis with time to reflect and ignore immediate post match hysteria.

Please keep it up and it deserves a much wider audience.

 

[Revolution]

How do S24SU,

New podcast is up - I had a good chat with fellow Blade and stat enthusiast Jay aka Blades Analytics on Twitter, which I just put live for your listening pleasure.

We cover quite a bit, including:

• The defeat to Fulham (01:30), the general state of the Championship, and where United sit within it
• A strange few weeks for United (12:15) including the Hull meltdown, Wilder's comments after, takeover talk
• Bouncing back at Reading (21:20)
• Breaking down Moore and Blackman, and how lucky we are to have two good goalkeepers (23:45)
• Some stats about our still-excellent defence (28:00)
• Making the case for Leon Clarke's form (37:40)

Enjoyed making this one, Jay's raises some excellent points and has got stats coming out of his ears. Hope you enjoy listening to it - and thanks as always for any feedback, likes, shares, etc. UTB!

SoundCloud:

iTunes: https://itunes.apple.com/gb/podcast/bladespod/id1309198119?mt=2

Really enjoying the podcasts - thanks for taking the time to do them. Jay's contribution is enjoyable.

Just a word about expected goals. I think there's certainly a place for stats in sports analysis - all clubs at our level or above are employing it in some way, and it's radically changed other sports, most notably baseball - but some stats are better than others, and to me there's too much subjective "noise" in expected goals to make it very useful. The fact that United's expected goals against is so out of kilter with actual goals against is to me an indicator that the stat is flawed, and not that the defence is better than goals against would suggest.

I am surprised it's something that's had some play on Match of the Day, because there's more meaningful things that you cite in the podcast (eg save percentage, percentage of chances taken) that would be more worthy of discussion. Even then, one chance or one shot is not like another.

One thing I would be interested in learning is how many shots are being taken against United from outside the box, and how many of those are going in, and comparing that to other teams' records. My impression is that conceding goals from distance is not something that is usually "expected" but its something that United are bad at, not unlucky at.

 

[Blade56]

How do S24SU,

New podcast is up - I had a good chat with fellow Blade and stat enthusiast Jay aka Blades Analytics on Twitter, which I just put live for your listening pleasure.

We cover quite a bit, including:

• The defeat to Fulham (01:30), the general state of the Championship, and where United sit within it
• A strange few weeks for United (12:15) including the Hull meltdown, Wilder's comments after, takeover talk
• Bouncing back at Reading (21:20)
• Breaking down Moore and Blackman, and how lucky we are to have two good goalkeepers (23:45)
• Some stats about our still-excellent defence (28:00)
• Making the case for Leon Clarke's form (37:40)

Enjoyed making this one, Jay's raises some excellent points and has got stats coming out of his ears. Hope you enjoy listening to it - and thanks as always for any feedback, likes, shares, etc. UTB!

SoundCloud:

iTunes: https://itunes.apple.com/gb/podcast/bladespod/id1309198119?mt=2

Will make a point of listening, not aware of any previous podcasts from you.

Are you contemplating expanding into Vlogs? All it takes is a good array of nerdy hoodies, crap hairstyles, wooden script reading and a 99 year rights contract with Crab Industries. Sorry The Crab but this is a big opportunity.

If you would like me to appear in your stats section my smiley face system below (copyrighted) suggests from the second half performances we will end up with an additional 17 points.

 

[Beans]

Cheers Revolution, really appreciate the feedback. I am by no means a statistician or data scientist, mainly just a fan with an interest in sports analytics, so I probably don't have any particularly constructive counterpoints to your point there

However, I am aware that expected goals - like all stats, I suppose - need to be taken with a pinch of salt as you say. That said, I did try to position that stat about our defence being "the best" in the context of the other teams in the league. So sure, knowing that United "should" have conceded 28 but have actually conceded 40 isn't that useful, but the fact that there's such a big difference there compared to Wolves and Cardiff seems like a good discussion point.

Going on the below, we were -12 goals down on what we "should" be, whereas Wolves were +1 (conceded 28, expected 29), Cardiff +2 (conceded 27, expected 29), Fulham +2 (conceded 38, expected 40), Villa +4 (conceded 32, expected 36).

That was a bit back-of-a-beermat as I need to get back to work , apologies if I'm off with a couple, but the overall point is that compared to teams above us we have been conceding WAY more than we should have, based on the chances we limit our opponents to. That could easily have equated to another 5-6 points in our favour if we'd been conceding at a similar expected rate as our rivals.

As you say, there could easily be factors here that mean we're conceding more of these low-percentage chances than simply "we've been really unlucky". Thought it was an interesting stat though and one that merited a bit of discussion! I'm going to pass on your suggestion that United are "bad" at conceding long range shots rather than unlucky to Jay and see if he can figure it out

Will make a point of listening, not aware of any previous podcasts from you.

Are you contemplating expanding into Vlogs? All it takes is a good array of nerdy hoodies, crap hairstyles, wooden script reading and a 99 year rights contract with Crab Industries. Sorry The Crab but this is a big opportunity.

Are you offering to be my agent in negotiation with Crab Industries? Don't worry, I've already got the crap hairstyle down.

Edit - made a dog's dinner of those numbers, they're in the right order now.

 

[The Crab]

Will make a point of listening, not aware of any previous podcasts from you.

Are you contemplating expanding into Vlogs? All it takes is a good array of nerdy hoodies, crap hairstyles, wooden script reading and a 99 year rights contract with Crab Industries. Sorry The Crab but this is a big opportunity.

If you would like me to appear in your stats section my smiley face system below (copyrighted) suggests from the second half performances we will end up with an additional 17 points.

Sorry.....................

 

[cooperblade]

Really enjoying the podcasts - thanks for taking the time to do them. Jay's contribution is enjoyable.

Just a word about expected goals. I think there's certainly a place for stats in sports analysis - all clubs at our level or above are employing it in some way, and it's radically changed other sports, most notably baseball - but some stats are better than others, and to me there's too much subjective "noise" in expected goals to make it very useful. The fact that United's expected goals against is so out of kilter with actual goals against is to me an indicator that the stat is flawed, and not that the defence is better than goals against would suggest.

I am surprised it's something that's had some play on Match of the Day, because there's more meaningful things that you cite in the podcast (eg save percentage, percentage of chances taken) that would be more worthy of discussion. Even then, one chance or one shot is not like another.

One thing I would be interested in learning is how many shots are being taken against United from outside the box, and how many of those are going in, and comparing that to other teams' records. My impression is that conceding goals from distance is not something that is usually "expected" but its something that United are bad at, not unlucky at.

Expected goals work well enough for one person (probably two) to have bought a football club and become seriously, seriously wealthy on the back of them.

Over a relatively small period of time, there can be a big difference between actual and expected goals. Over a longer period of time, they will prove much more accurate. For chances 8 yards out from the middle of the goal, there is a number of goals you'd expect to score per (say) 1,000 chances and this will be pretty accurate (over a large sample). You can, of course, miss 10 of these on the trot but the logic is, keep creating good chances and you'll score goals.

On the flip side (and these numbers are totally made up), 1 shot in 10 from 20 yards might go in, so that's 10 in 100 or 100 in 1000 - that means you might do a United and have 5 on the bounce go in and at that point, the difference between expected and actual is really big. Over time though, it'll regress to the mean and you'll end up conceding 100 out of 1000.

It's quite possible to have an entire season where you score more and / or concede less than you should (or the reverse) - look at Reading last season. They went against every stat going and ended up making the play-offs. Essentially they had a lucky season. The league table can and does lie in terms of how teams are playing. This season, their results are reflecting their performances and they're back where they should be.

There's a book called 'The numbers game' and they claim that football is roughly 50% ability and 50% luck - that's why a good team doesn't always beat a poor team.

In terms of United and goals conceded, it's going on for 2 seasons now where we're easier to score against than we should be (opponents need fewer chances to score against us than against other teams). We don't give up many chances though which limits how big a problem it's been. It is a long, sustained run though. Is it just a long run of bad luck? Is it a problem at goalkeeper? Is it something else?

 

[cooperblade]

Fantastic podcast by the way Beans - right up my street, really well done and very enjoyable. Thanks

 

[Revolution]

Expected goals work well enough for one person (probably two) to have bought a football club and become seriously, seriously wealthy on the back of them.

Over a relatively small period of time, there can be a big difference between actual and expected goals. Over a longer period of time, they will prove much more accurate. For chances 8 yards out from the middle of the goal, there is a number of goals you'd expect to score per (say) 1,000 chances and this will be pretty accurate (over a large sample). You can, of course, miss 10 of these on the trot but the logic is, keep creating good chances and you'll score goals.

On the flip side (and these numbers are totally made up), 1 shot in 10 from 20 yards might go in, so that's 10 in 100 or 100 in 1000 - that means you might do a United and have 5 on the bounce go in and at that point, the difference between expected and actual is really big. Over time though, it'll regress to the mean and you'll end up conceding 100 out of 1000.

It's quite possible to have an entire season where you score more and / or concede less than you should (or the reverse) - look at Reading last season. They went against every stat going and ended up making the play-offs. Essentially they had a lucky season. The league table can and does lie in terms of how teams are playing. This season, their results are reflecting their performances and they're back where they should be.

There's a book called 'The numbers game' and they claim that football is roughly 50% ability and 50% luck - that's why a good team doesn't always beat a poor team.

In terms of United and goals conceded, it's going on for 2 seasons now where we're easier to score against than we should be (opponents need fewer chances to score against us than against other teams). We don't give up many chances though which limits how big a problem it's been. It is a long, sustained run though. Is it just a long run of bad luck? Is it a problem at goalkeeper? Is it something else?

I take your point re expected goals being fairly accurate over time, but I still do not see it as reliable as, say, data they gather in baseball using pitch tracking. That's because there are more variables when you have a shot than when you throw a pitch. And I agree that "fluke" seasons exist: Everton's fourth place year in 2005 when they had a lot of narrow victories springs to mind, as does the season Huddersfield had last year.

The difference between our expected goals and actual goals suggests that something is not being measured properly. The data suggests we've conceded nearly 50% more goals this season than a team who's had those chances created against them should have. That's a massive variation over a reasonable number of games.

If I had to guess I'd say its a combination of:

- random chance
- poor goalkeeping, in terms of positioning for long shots (see how far Moore is off his line for Bristol City's opener, for example, or the hash he made of Hull's goal at the Lane).
- our players not picking up enough loose balls in the 10 to 15 yards outside the penalty area and our defenders not closing players down quick enough

If someone has the data showing how we fare with shots from outside the box as compared to other teams in the division, I'd be interested to see it.

 

[cooperblade]

I take your point re expected goals being fairly accurate over time, but I still do not see it as reliable as, say, data they gather in baseball using pitch tracking. That's because there are more variables when you have a shot than when you throw a pitch. And I agree that "fluke" seasons exist: Everton's fourth place year in 2005 when they had a lot of narrow victories springs to mind, as does the season Huddersfield had last year.

The difference between our expected goals and actual goals suggests that something is not being measured properly. The data suggests we've conceded nearly 50% more goals this season than a team who's had those chances created against them should have. That's a massive variation over a reasonable number of games.

If I had to guess I'd say its a combination of:

- random chance
- poor goalkeeping, in terms of positioning for long shots (see how far Moore is off his line for Bristol City's opener, for example, or the hash he made of Hull's goal at the Lane).
- our players not picking up enough loose balls in the 10 to 15 yards outside the penalty area and our defenders not closing players down quick enough

If someone has the data showing how we fare with shots from outside the box as compared to other teams in the division, I'd be interested to see it.

As far as I'm aware (and I might be wrong), the models measure where the chance was on the pitch, I'm not sure how much they (certainly the basic ones) differentiate between a shot from 20 yards with 3 defenders closing you down and one where you've got time to pick your spot. It would seem safe to assume that whatever models Bloom / Benham use have more sophistication than some of the (excellent) stuff you can access on the internet for free like via the BladesPod or Experimental 3-6-1. You're right though, it's not an exact science like it would be in Baseball. It can still provide predictions which are more accurate than the market though so it's very, very good for a sport with so many variables.

I think you're right that the fact we concede more goals than we should is down to a number of factors. In terms of goalkeeping, without too much effort I can think of a couple more examples where Moore should've done better with long range shots. I find it harder to think of examples with Blackman (although that might be confirmation bias as I rate Blackman higher than I do Moore).

The data about shots from outside the box definitely exists, we just need someone to crunch the numbers - Jay?

 

[Revolution]

Also just a quick mention for something else re expected goals and our defence. The defence has conceded a goal in every single away league game bar one (Bolton) this season. That record does not suggest to me that we have an excellent defence playing in bad luck. It's more indicative of a defence that's overrated in some way in terms of expected goals.

 

[cooperblade]

Also just a quick mention for something else re expected goals and our defence. The defence has conceded a goal in every single away league game bar one (Bolton) this season. That record does not suggest to me that we have an excellent defence playing in bad luck. It's more indicative of a defence that's overrated in some way in terms of expected goals.

It's felt like we're easy to score against for a while without ever giving up loads of chances. In the pre-Wilder era, especially at Bramall Lane, it felt like we'd often be on top, creating chances and then the opposition would score the first time they went forward.

 

[Blade56]

Beans .......Are you offering to be my agent in negotiation with Crab Industries? Don't worry, I've already got the crap hairstyle down.


You are fekin joking, last person The Crab negotiated for were Skerpol and his daughter