Experimental 3-6-1

[shoddycollins]

http://experimental361.com

I've taken a bit of an interest lately in the views of the people running this football stats site, and posted information from there on the last two match-threads for Carlisle.

Taking a statistical approach to football is always going to generate controversy, but in my opinion it's OK as long as have a realistic expectation for what the stats can tell you. At no point would I ever say stats can be used to prove one team did or didn't deserve to win a game. This sort of thing isn't about saying what should have happened, it's about looking at why what happened, did happen. In a way it's more about exposing the unknown factors that make football unpredictable, than predicting the outcome of games.

Their entire approach revolves around the concept of expected goals, and while I don't know the exact technique they use, it is basically the idea that not every shot on goal is equal. The problem will looking at shots off/on target is that is doesn't differentiate between beating the defense and getting 1-on-1 with the keeper, and taking a pot-shot from range.

Expected goals aims to make this distinction by giving each chance on goal a value for how good it is. 1-on-1 with the keeper, you expect to score so that chance is almost worth an entire expected goal in itself, whereas firing one from 30 yards is very speculative so might only be worth 0.1 of an expected goals.

Expected goals and actual goals rarely tally exactly, because there are still unknowns. If a team has expected goals tally of 2 in a game, but only score 1 then that means either they were creating changes they should expect to score from, but not scoring. That could mean either their strikers aren't clinical enough, or the opposition's goalie or defending players are doing well at stopping those shots, or simply luck, even a bad goalie guesses correctly some of the time, and even a good striker sometimes misses. You'd think luck would even itself out over the course of a season then.

The general idea then is that for the most part, expected goals scored, combined with expected goals conceded is a good measure of a team's effectiveness, as most of the game is spent trying to create good scoring chances, and for most of the players doing this is their entire reason for being on the pitch. What happens once a chance has been created, at the extreme ends of the pitch with goalies and strikers is still up for debate, but it's interesting to see which teams are doing well because they create chances they should score from and prevent the opposition getting chances they should score from, and which teams are doing well in spite of not creating chances or allowing too many chances.

You could say the former is an indication of a good overall team who should expect to continue to do well while the latter is an indication of a team who are more average overall, but have clinical strikers or a good rearguard, or both, allowing them to outperform themselves. They could also outperform themselves by scoring lots of goals from half-chances if they have players who can shoot from distance, or score free-kicks. The latter may be more vulnerable to a loss of form, or injuries to key players.

So here is League Two ranked in order of how much they dominate games. You'll tend to find at the top teams that both create good scoring chances and prevent opponents having them, while teams that can neither create nor prevent chances are at the bottom. In the middle you will find teams that are either average at both creating and preventing, or good at one but poor at the other.

Based by reading the following graph from bottom right to top left
http://experimental361.com/2015/12/22/season-update-league-2-22-dec-2015/

1 - Portsmouth
2 - Wimbledon (explanation required !?)
3 - Oxford United
4 - Accrington Stanley
5 - Bristol Rovers
---------------------------------------------------------------------- then a bit of a gap
6 - Leyton Orient
---------------------------------------------------------------------- then an even bigger gap
7 - Notts County
8 - Wycombe Wanderers
9 - Mansfield Town
10 - Barnet
11- Plymouth Argyle
12 - Carlisle United
--------------------------- below this line, teams are conceding better chances than they create
13 - Luton Town
14 - Cambridge United
15 - Northampton Town (very interesting indeed)
16 - Exeter City
17 - Crawley Town
18 - Morecambe
--------------------------------------------------------------------------then a bit of a gap
19 - Hartlepool United
20 - Newport County
21 - Stevenage
--------------------------------------------------------------------------then a massive gap
22 - York City
23 - Yeovil Town
24 - Dagenham and Redbridge

The fact that we're mid-table in this reminded me of another factor, it's not just the chances you create, but when you create them.

We're outperforming ourselves to be 5th in the table but not because we're especially clinical, we genuinely are creating lots of changes per game so the goals we've scored are because we 'should' be scoring loads given the chances we make. Neither is it simply a case of our rearguard being good as we've conceded about as many as you'd expect a team who give the opposition so many chances to concede. Our goal-difference is pretty modest; we have won a lot of games by the odd goal, in fact of our ten league wins this season, only our most recent two, at home to Crawley and Notts have been by more than one goal, as have two of our five defeats. This could be down to luck, we just happen to create our chances at the right time, or it could be by design we get into the lead and then we ease off... or it could be to do with motivation, our players don't intend to, but subconsciously work harder when we need to score.

 

[Vanni]

Or, in other words, Wimbledon need to change their back four and keeper, whilst Northampton are just plain lucky.

ps - If I have understood the graph correctly that is.
pss - Actually, it's the other way round. Wimbledon need better forwards? I'm confused now!

 

[EricSabin]

But then I saw a stat that places us in the most (if not the most) clinical finishers in the division.

EDIT: We are the most clinical finishers, we need 5 shots for every goal

 

[shoddycollins]

But then I saw a stat that places us in the most (if not the most) clinical finishers in the division.

EDIT: We are the most clinical finishers, we need 5 shots for every goal

Well yeah, this is alluded to on experimental 3-6-1's site, I did try and add every caveat possible to my post as I don't want to make it sound like a total mystery as to how Northampton find themselves in the top 3... it does suggest that Northampton and Oxford are doing well for different reasons though, and the interesting thing would be to know whether Northampton have clinical finishers throughout their team, or whether it is specifically Marc Richards who is the clinical one, as if it is the latter you could be in trouble if he gets injured.

 

[shoddycollins]

Or, in other words, Wimbledon need to change their back four and keeper, whilst Northampton are just plain lucky.

ps - If I have understood the graph correctly that is.
pss - Actually, it's the other way round. Wimbledon need better forwards? I'm confused now!

There is a full set of graphs on the site with further information
http://experimental361.com/2015/12/22/season-update-league-2-22-dec-2015/

They classify Wimbledon as being 7th in terms of shots created (and also 7th in terms of expected goals), which suggests that they are creating plenty of good chances to score, however they have the 4th highest number of shots taken to score... it classes them as 'energetically wasteful'... or 'create a lot but can't take advantage of what they create'.

Defensively, Wimbledon are also interesting, only Portsmouth face fewer shots per match, but only Barnet concede from a greater proportion of shots faced. Wimbledon are second lowest in terms of expected goals conceded too, which suggests that their opponents aren't creating particularly good chances.

So what this means is you're right on both counts... Wimbledon create a lot and prevent their opponents from creating much... however these advantages are cancelled out by the fact they need to create a lot of opportunities to score an average amount of goals and their opponents don't need to create all that much against them to score. Wimbledon could benefit from a better goalie (and to a lesser extent better centre backs, as their current centre backs could be helping them prevent opponents creating chances even if they aren't doing much to stop opponents scoring from the chances they do create) and better strikers.

Northampton are better than just lucky, they are pretty good and effective defensively, they're just scoring goals from very few chances, there's nothing lucky about having clinical strikers, but personally I'd rather be creating enough chances that strikers will score goals whether they are clinical or not.

 

[Trapdoor]

Great thread. Great site.

 

[Glasshalffullpools]

Taking a statistical approach to football is fine ......if you have no wife , no girlfriend , no meaningful outside interest and a half arsed attempt at a beard ..........sandals completely optional

 

[RavenBish]

 

[Trapdoor]

 

[thewwfc]

Hartlepool attack "ineffectual"? King Rakish? How dare they.

 

[PoolieTom]

So we have a shit defence, shit attack, we are shit at making chances, we are shit at taking the few chances we do make, we concede lots of attacks, and of the attacks we do concede, they very often lead to a goal.