Experimental 361

[Gladders]

Because Stats!!

Am sure some of you have read this great blog and seen his graphs, so I thought I'd start a thread on them and fellow statto's can discuss them.

https://experimental361.com/

https://experimental361.com/2016/09/18/squad-rotation-in-the-efl-18-sep-2016/
Interesting one this week, with the average number of changes per game. Cambridge bottom with the most number of changes per week is no surprise I guess.

Another one I found interesting is the Squad churn from last season, especially the difference between the two coming up from the conference. Cheltenham hardly changing their squad and us massively changing it. Appears to be working for us better so far, but will it carry on?

https://experimental361.com/2016/09/01/efl-squad-churn-2016/

 

[BeesKnees]

We're still playing with our conference winning team. Even players deemed not good enough last season are back in.

The stuff Ben does on the 361 site are excellent. The expected goals give a really good insight into a sides strengths and weaknesses

 

[shoddycollins]

I was gonna post the squad churn one earlier in response to something one of our fans said about us having a lot of new players that need to bed-in. Ours could actually increase even more if we end up re-signing Asamoah.

 

[DearneValleyRover]

Our change in squad is unsurprising given the shite Dickov signed and our rotation is also unsurprising as we have 10 first teamed injured.

 

[shoddycollins]

Our change in squad is unsurprising given the shite Dickov signed and our rotation is also unsurprising as we have 10 first teamed injured.

Doesn't that render the whole 'best squad in League One' argument moot?

 

[DearneValleyRover]

Doesn't that render the whole 'best squad in League One' argument moot?

There won't be many who thought we had although it certainly should have been good enough to stay up.

 

[shoddycollins]

Another update from Experimental 3-6-1

E-ratings are basically an alternative type of form guide. You can read the original post here but I've summarised the important things to know below.

• This is based on Experimental 3-6-1's expected goals calculation which is based on quality of chances created rather than shots on/off target and goals scored. The second and third columns are the expected number of goals this team would score and concede against an average opponent with the first column, the 'e-rating' being the average size of win/defeat against the average opponent.
• The effect of any one game a team's attack and defence scored is weighted to take account of the opponent's attack and defence scores, so for example, creating one expected goal against Portsmouth will increase your attack rating a lot more than creating one expected goal against Morecambe. Morecambe's expected goals conceded against an average opponent is 1.88 so if you only create one then your chance creation has been worse than an average side. This is particularly relevant early in the season where differences in league position and ranking on the above table could be down to whether a team has been playing stronger or weaker opponents in their opening fixtures.
  
• As with everywhere that expected goals are used it rates a teams ability to create and prevent chances, not score and save them, the assumption is that this is a better barometer of overall team performance than simply goals scored/conceded but can a team with particularly clinical strikers or a good goalie could outperform their expected goals (and vice-versa). In a single game, luck can also be a factor, but these are calculated over 30 games which rules that out.
  
• Since this is over the last 30 league games, it goes back well into last season and you can see a number of teams have recent changes in direction which probably correspond with the beginning of the new season.

So what have I noticed?

• Portsmouth are a class ahead of everyone else.
• Morecambe are pretty poor in terms of creating chances and preventing them so to be where they are in the league must be the result of either short-term luck or good goalkeeping/finishing.
• Wycombe are getting progressively worse at an alarming rate.
• Plymouth have an extremely consistent defence, but a somewhat inconsistent attack
• Carlisle have the biggest number of combined expected goals (for and against), though seeing an upturn in defense since the summer
• Grimsby's defence has been getting worse but the decline began before they had to step up a division.
• Cheltenham haven't had a great start back to life in the League, but look a decent team, maybe they had a difficult start, or maybe they need to look at goalkeeping or finishing.
• Yeovil's defence was improving up until the summer, but is now going back in the other direction and may not be able to compensate for a pretty dire attack for much longer.
• Mansfield are tending towards mid-table boredom, and with the biggest difference in performance between attack and defence (strong defence, weak attack)
• The difference a summer makes is particular clear in Cambridge's case, with defence being where it's all gone wrong.
• Accrington are still a good side, and are actually improving slightly with a declining attack not quite fully cancelling out a distinctly improving defence.
• Hartlepool are consistently crap, just not quite crap enough to go down.

Based on that, I would lump on Portsmouth to finish champions and Wycombe and Newport to go down... not sure who else might get promoted though.

Morecambe might arrive in the relegation battle but probably with too much of a headstart to be seriously threatened, proving me right when I said at the start of the season that Morecambe always find a way to stay up despite predictions. Hartlepool will finish in the bottom four... again... giving their fans yet another season to remind us of their remarkable record of not getting relegated.

 

[BeesKnees]

Preview of tonight's games
https://experimental361.com/2016/09/26/league-2-previews-27-sep-2016/

 

[jacobncfc]

I didn't realise that it goes back thirty games, explains our very low rating despite a half decent start to the season.

I'd be interested to know how they work out the quality of chances, though. Is it just the position from which shots are taken?

 

[AdamStag]

Thought 3-6-1 was one of Murrays crap defensive formations!

 

[shoddycollins]

I didn't realise that it goes back thirty games, explains our very low rating despite a half decent start to the season.

I'd be interested to know how they work out the quality of chances, though. Is it just the position from which shots are taken?

I believe it's a combination of the position and type. It's only one guy calculating it all and he tends to get the stats updated pretty quickly so I'm guessing he just goes off the descriptions in the BBC match report timeline, things like 'a headed shot from inside the area'.

 

[shoddycollins]

Thought 3-6-1 was one of Murrays crap defensive formations!

I'm sure we used 3-6-1 on at least one occasion last season

 

[AdamStag]

I'm sure we used 3-6-1 on at least one occasion last season

We mainly use 6-3-1 this season

 

[shoddycollins]

We mainly use 6-3-1 this season

We used 6-3-0 against Pompey (we'd had a man sent off and had a point to protect)

 

[GTFCfish]

We used 6-3-0 against Pompey (we'd had a man sent off and had a point to protect)

When Keegan was at Newcastle he took it to the other extreme and used a 1-3-6.

 

[shoddycollins]

Anyone down as far as Crawley could find themselves in 2nd spot come 5pm on Saturday

 

[shoddycollins]

https://experimental361.com/2016/10/20/scatter-graphics-league-2-20-oct-2016/

While the stats based on quality of chances created and lack of quality chances conceded that are used for the e-ratings still place Portsmouth as far and way the 'best' team in the division, the story of why they aren't romping home with the title is starting to show itself.

Not only are Portsmouth the second worst in the division at putting away the chances they create, but they're also the worst at preventing chances against them from ending up int he back of the net. Good footballing side let down by ability at either end of the pitch and if something such as injuries to key players were to upset their ability to dominate games then they could go on a losing run. I presume though this will be addressed via the transfer market in January.

 

[shoddycollins]

Also a new way of showing how each team could change position after tomorrow's games includes probabilities for each position they could potentially find themselves in. Could be a very interesting graphic come the final quarter of the season.

 

[Look Out There Are Llamas]

They're terrific.

Oh my God, Orient's attack is "energetically wasteful". That literally describes Ollie Palmer in a nutshell.

 

[George Reilly's Hairpiece]

We are more likely than not to be out of the bottom two after Saturday's games.

Presumably that means Orient are going to take a bigger hiding than us.

 

[shoddycollins]

We are more likely than not to be out of the bottom two after Saturday's games.

Presumably that means Orient are going to take a bigger hiding than us.

Not just Orient, as you're still more likely to be there than they are, but there's a few teams that could drop into that place so it would appear the chances of one of them doing shit enough for you to overtake are just over 50%

 

[George Reilly's Hairpiece]

Not just Orient, as you're still more likely to be there than they are, but there's a few teams that could drop into that place so it would appear the chances of one of them doing shit enough for you to overtake are just over 50%

I was kind of implying that I think we have no chance of getting anything from Wycombe (hence there is only one way we can move up the table), but clearly I was too subtle.

 

[shoddycollins]

I was kind of implying that I think we have no chance of getting anything from Wycombe (hence there is only one way we can move up the table), but clearly I was too subtle.

The opposite, but I felt that was too obvious.

 

[George Reilly's Hairpiece]

Get yourself on Betdaq. We've got business to discuss.

 

[shoddycollins]

Get yourself on Betdaq. We've got business to discuss.

I made my predictions (not on that, somewhere else) the other day and put this down as a 2-1 win to Wycombe. I took a look to see where Barnet would be afterwards and yep, still second bottom. I had Orient to pick up a point at Hartlepool.

 

[shoddycollins]

Not a good day for E361's predictions... or George Reilly's betdaq come to think of it.

Of the nine teams given a 40% or greater chance of winning.
Porstmouth LOST
Colchester DREW
Luton DREW
Crewe LOST
Grimsby LOST
Exeter LOST
Wycombe LOST
Accrington DREW
Plymouth WON!!!!!

 

[K-Win]

I was reading a website the other day about using a ratings system to calculate win/draw/loss percentages. Apparently, based on historical results, if two teams with exactly the same rating are playing each other, the possibility of the home team winning is around 46.5%. Thus you should be looking at 47% and higher for 'fancied' home teams.

That would leave you with Pompey, Colchester and Luton as the genuinely underperforming home teams last weekend.

 

[Vanni]

This percentage thing is just a load of tosh if you ask me.

Case in point no. 1 - Exeter (with a 0-1-5/GS 1 GC 8 home record prior to Saturday's game) vs Cambridge (1-2-3/GS 4 GC 7 away record), and yet Exeter had 43.4% chance of winning. That's right, Exeter had only scored one goal at home and without a home win so far, so 43.4% is a bit on the high side no?

C.I.P. No 2 - Crawley (3-2-1/7-5) vs Stanley (2-2-2/7-7) with Accrington having a 51.2% chance of winning!

C.I.P No 3 - Colchester (3-1-2/11-9) vs Morecambe (4-1-1/10-8) and yet Colchester had a 54.6% chance while Morecambe only had a 22.75% chance, even though they had won 4 of their 6 away matches.

 

[jacobncfc]

The calculations still take into account the last bit of last season I think, hence some of those odd readings. It's over the last 30 games or something.

It's why we're so low down all their ratings despite being decent this season - our horrible spell of losing 5-0 all the time at the end of last season is still affecting it.

 

[Vanni]

The calculations still take into account the last bit of last season I think, hence some of those odd readings. It's over the last 30 games or something.

It's why we're so low down all their ratings despite being decent this season - our horrible spell of losing 5-0 all the time at the end of last season is still affecting it.

Ah that explains it then, cheers! Still I can't help but think that basing the calculations over 30 games is a bit silly. I'd get it if was over the past 5 or 6 games, as that's a proper 'form table', but 30 games?