Form- Data Visualisation

[GazLaz]

What the hell has happened? The numbers I have used are pretty sophisticated xG figures that take into account every attack for each team during a game, quantify them and then generate a total figure.

We have fallen off a cliff after improving.

The improvement came from improving defensively, the attacking output remained stagnant.

We are in big trouble.

[Chris Black come back]

It’s shite to watch and delivers terrible results. We visualise that every match.

[sedwardsdrfc]

Don’t need to count every forward pass to figure out we’ve lost any semblance of a been a team that can win a game without extreme luck.

I wonder what’s been done during the week in training to lead to this. There has to be a reason or something we are doing that is just plain wrong.

Maybe we just practice 5 yard goal kicks. But we’re shite at those too. Certainly don’t practice throw ins.

[mpc123]

This does sum it up. But really didn't need software for it. Thanks Gazlaz for the info.

[Mike_F]

They showed that XG for graph at the MTO event to demonstrate how we were making real progress under DS. It was on the back of three wins and was looking very healthy.

They clearly use and analyse these data to track and evaluate success/progress so with that in mind if I was Danny Schofield I'd be very concerned about my job security right now.

[BillyStubbsTears]

What the hell has happened? The numbers I have used are pretty sophisticated xG figures that take into account every attack for each team during a game, quantify them and then generate a total figure.

We have fallen off a cliff after improving.

The improvement came from improving defensively, the attacking output remained stagnant.

We are in big trouble.

Interesting stuff Gaz.

I do a lot of time series analysis professionally and that often includes taking rolling averages to find trends in noisy data. I have a question about this. How does the rolling 8 game average work? Is it the average of the previous 8 games? In which case, how does that work at the start of the season? Does it roll over from last season?

[Filo]

I would suggest, despite what some players say about Schofield, they have stopped playing for him, that drop off is of biblical proportions

[Mike_F]

That's an interesting query which also leads me to thinking how it would work with a change of manager, the assumption being on the charts above that there would be an exponential demonstration of the impact that Schofield had as the first eight games under his leadership accrued with every passing game effectively deleting one of McSheffrey's from the front end of the data set.

It would also suggest that B isn't as impressive a stat for DS as it might first appear because it's partially inclusive of A whilst the delta at C moves out of the range.

[BillyStubbsTears]

The data for match 32 is far worse than the data for match 30.

If the average is a lagging one, the average for match 30 is based on the data from matches 23-30 and the average for match 32 is based on the data from matches 25-32.

That implies that matches 23-24 were very good for us and 31-32 very bad.

23-24 were the wins against Rochdale and Carlisle. 31-32 were the abject defeats against Sutton and Bradford.

On one level then this is just a useful way of visualising what we all know from watching the performances. It kind of says that, up until the Sutton game, Schofield had got us performing at about 8-10th place standard. Sometimes a tad above, sometimes a tad below.

That's about the standard I think for the distinctly limited squad we've put together.

Again, as we know, the wheels have come off since then. Questions are: why and can he fix it?

Having calmed down a bit from Tuesday night, I'd be prepared to give him a bit of leeway on the "why". We've been badly disrupted by injuries (Maxwell, Lakin, Biggins, Anderson, Close all missing games, Rowe clearly not at 100%, now Miller out). For a squad as rank bad as the one we've put together over the past 2 years, it's always going to be tough to deal with that sort of disruption.

It also didn't help that at 0-0 at Sutton, Miller missed the easiest chance of the season when, once again, he refused to use his left foot to roll a ball into an empty net from 3 yards out. On such moments can form change...

The "can he improve it" question is harder. On one hand, I think until recently be had developed a style of play that at least got the side performing somewhere near the sum of its very limited parts. Could he do better with better recruitment? No reason to think not. But he's certainly not got the side playing to a standard better than the sum of its parts, which is the mark of an excellent manager


On balance, I think I'd give him the summer and first half of next season and hope for a bit of luck on injuries and a bit of professional diligence on recruitment.

[GazLaz]

What the hell has happened? The numbers I have used are pretty sophisticated xG figures that take into account every attack for each team during a game, quantify them and then generate a total figure.

We have fallen off a cliff after improving.

The improvement came from improving defensively, the attacking output remained stagnant.

We are in big trouble.

Interesting stuff Gaz.

I do a lot of time series analysis professionally and that often includes taking rolling averages to find trends in noisy data. I have a question about this. How does the rolling 8 game average work? Is it the average of the previous 8 games? In which case, how does that work at the start of the season? Does it roll over from last season?

Yes it is and it isn’t carried over, I just built it up over the first 8 games, 1 game, 2 games rolling, 3 games rolling, up to 8.

[BillyStubbsTears]

Cheers Gaz. Much appreciated.

You might want to try a centred 7 game rolling average.

So the value for match 30 for example is the average of matches 27-33  and you taper the average off at either end.


(So game 1 data is just game 1, game 2 data is average of games 1-3, game 3 is average of games 1-5, and the same at the latest end - game 37 is game 37, 36 is average of 35-37, 35 is average of 33-37).

Advantage of this is threefold.

1) It gives you a better feel at each stage for what was the form averaged out over the period for which your date is the centre, rather than using data from a long time earlier. This makes it easier to spot when major reversals of form started.

2) You get a better averaging at the start of the season.

3) You might just start to see when a change in form is starting to come in at the right end of you time series and be able to predict better where we are going (because you averaging at the end of the series isn't influenced by what happened 5,6,7,8 games ago).

Disadvantage is if you have one outlier performance in your most recent game, the final point on your graph can go up or down by a lot without it really being a trend.

[DonnyOsmond]

Interesting that the tumble started when Maxwell got injured.

[BillyStubbsTears]

Should have said, the centred average will, for most of the data, more or less just shift the trend line you already have a few games to the left on the graph. But as I say, it makes it easier to see where the change in trend actually started.

I'd guess that, what looks like a big change in net xG around game 32 is actually telling us the real change was from games 28-29 onwards.

That's interesting, because we won those two matches (Tranmere and Swindon) and the next (Barrow) but the consensus was that we hadn't been particularly good in any of the games.

[BillyStubbsTears]

Interesting that the tumble started when Maxwell got injured.

And (whisper it) Ro-Shaun Williams.

Interesting...

[BillyStubbsTears]

Whisper it even more quietly. The dip also started about the time that Rowe came back from injury.

[GazLaz]

Whisper it even more quietly. The dip also started about the time that Rowe came back from injury.

He’s gone hasn’t he?? Hope not admittedly. His technical ability had always been ordinary so when his physical assets go he’s done.

[pib]

Whisper it even more quietly. The dip also started about the time that Rowe came back from injury.

Think it's been widely acknowledged among the fanbase that he's been miles off it recently.

I'd still be interested to see how effective he'd be further forward, but it could just be that he's gone well past his best.

[CJK]

Little bit unfair on Rowe I think. Played out of position since he came back into the team and it also looks like, we assume, he's not 100% fit. If he doesn't have a fitness issue then yes, he's not looking good. Would just like to see what he looks like playing in a more suitable role.

[i_ateallthepies]

Tommy seems to be slower in his thinking more than his physical speed.  There was one situation early-ish in the game Tuesday where a Crawley long punt upfield was going out for a goal kick and Tommy was jogging along behind it to make sure it did.  He carelessly stood off it by a couple of feet and the chasing striker just ghosted past him and back heeled it against Tommy for a corner.  Gob-smackingly slack play.

[sedwardsdrfc]

Rowe does seem to be gone. Hopefully he’s just playing with niggles out of necessity and with a bit of time he’s closer to the player we know.

Think DS should play Faulkner ahead of Rowe and use Rowe as experience from the bench till he’s actually fit

[Lesonthewest]

He was our top scorer not long ago, God forbid we started him further forward & looked a bit more threatening.

[Chris Black come back]

Close, Taylor, Anderson and Rowe. Probably our top earners? Certainly our four senior players in the squad. Lot of injuries there and loss of form.

[BillyStubbsTears]

If they aren't our top earners, the recruitment over the past two years has been even worse than I thought.

[sedwardsdrfc]

Copps said in his interview about building the team around Rowe and Anderson. I thought that was worrying. Yes good pros who it’s nice to have around the squad. They I would rather they aren’t expected to be 1st choice. Certainly can’t see Rowe been that player again. Hope I’m wrong

[GazLaz]

Just revisiting this.

I said in March that the data said we were in big trouble. We picked up 4 points in the last 9 games after this.

Now the data is saying we’ve improved massively. We have to caveat all this in the near future due to the horrific injuries. In the short term we could really struggle until we get bodies back. In the medium term we will be ok. February to May is massive for us. A strong run in for us is needed so we can hit the ground running next season. I think it’s likely to happen.

[Filo]

Although I have no understanding of xg and how it is calculated, I find these stats interesting, and more or less match our performances

[BillyStubbsTears]

Interesting stuff Gaz. Thanks for posting that.

I'm going to take issue with some of what you said though, and in doing so, I'll be the lightning conductor again for some bile I suspect. But I can't help myself.

You're (rightly) pointing out that we have a horrific injury crisis. And it's absolutely right to view performances and results in that light. But  we had a far worse crisis in terms of players available at the back end of last season, when things really collapsed.

Also, your figure screams very clearly that our data this season has only just briefly got up to the levels we were at for most of the period between matches 23-31 last year. I suspect that's because we are allowing so many more chances than the middle of last year, and not creating enough chnaces of our own to balance that. Is that right?

Whatever the reason, it puts an interesting perspective on the relative performances of Schofield and McCann. It looks like Schofield was, until the squad started falling apart and the wheels came off, actually quite effective at making us quite effective, even if it was grim to watch. My gut at the time said that we were being unfairly harsh on him, perhaps because some of us expected us to romp to promotion. Maybe part of the reason McCann is getting an easier time (more attractive football too, certainly) is because expectations are lower after last year? Having said that, it's now time for me to be retiring to the bunker.

[GazLaz]

We were a low chance creation, more solid unit at our best under DS. Thats never going to be as easy on the eye as a higher chance creation, more open defensively team.

With DS, I couldn’t see where the improvement in front of goal was going to come from and I don’t think the players did either. That added to the demise when the system essentially started boring them too.

The solutions to McCann’s issues are potentially easier to find. We have to bin the 3 centre half system for me. More bodies defensively doesn’t necessarily mean more solidity. The amount of goals we concede by one of the centre halves getting dragged out of position and a deep runner exposing the space is incredible really.

[Chris Black come back]

It’s shite, that’s what it is. Piss weak in defence and shipping goals pretty much every game.

[Colin C No.3]

That’s really, really, really interesting stuff GazLaz.