Data in Football

[bluebruce]

42 minutes ago, DE. said:

I built a relational database containing this season's stats which as far as I can tell seems to be accurate (BBC stats for possession). Assuming my statistics are correct and I didn't screw up the indexing somehow, it ends up like below:

45% or Less Possession

	Games	Points	PPG
Home	2	6	3
Away	4	8	2
Combined	6	14	2.3

46-55% Possession

	Games	Points	PPG
Home	5	8	1.6
Away	6	6	1
Combined	11	14	1.3

56-65% Possession

	Games	Points	PPG
Home	10	11	1.1
Away	7	3	0.4
Combined	17	14	0.8

66-75% Possession

	Games	Points	PPG
Home	6	9	1.5
Away	6	6	1
Combined	12	15	1.3

So yeah, assuming my calculations are accurate, the more possession we have the worse we get across the board until we reach the 66% mark, which is still marginally below 46-55% possession and some way below < 46%. 

Yep that's more or less exactly what I did, if you look. The only differences (apart from that mine needs updating) are the layout, the source for the possession stats, and the percentile segments selected. Interesting that both you and Joe talk about 55-65, whereas I went with 50-60 etc. I mostly picked that because I wanted to be able to delineate cleanly between when we do have more possession and when we don't, so I could also show a 51+ and a 50- section for direct contrast. Although I do get that there's not really any material difference between having 49% possession or 51% possession.

Despite picking slightly different sections of percentage, we both found very similar things. The main weakness in the data, however you collect it, is that the sample size of games where we didn't control possession is relatively very small, largely due to our hyper-focus on it this season, whilst opponents who figured us out soon began letting us have as much of it as we wanted. But the fact our season was very disappointing in general only emphasises that this isn't the right approach for this squad under this manager.

I do wonder what could have been if the manager had listened to this data and switched to a counter-attacking style, with through balls and balls over the top to run onto, or at least mixed it up a bit. We have the pace to frighten teams on the break, it's blindingly obvious.

[DE.]

3 minutes ago, bluebruce said:

Yep that's more or less exactly what I did, if you look. The only differences (apart from that mine needs updating) are the layout, the source for the possession stats, and the percentile segments selected. Interesting that both you and Joe talk about 55-65, whereas I went with 50-60 etc. I mostly picked that because I wanted to be able to delineate cleanly between when we do have more possession and when we don't, so I could also show a 51+ and a 50- section for direct contrast. Although I do get that there's not really any material difference between having 49% possession or 51% possession.

Despite picking slightly different sections of percentage, we both found very similar things. The main weakness in the data, however you collect it, is that the sample size of games where we didn't control possession is relatively very small, largely due to our hyper-focus on it this season, whilst opponents who figured us out soon began letting us have as much of it as we wanted. But the fact our season was very disappointing in general only emphasises that this isn't the right approach for this squad under this manager.

I do wonder what could have been if the manager had listened to this data and switched to a counter-attacking style, with through balls and balls over the top to run onto, or at least mixed it up a bit. We have the pace to frighten teams on the break, it's blindingly obvious.

I didn't really have any reasoning for my percentile segments, beyond 10% increments simply being easiest from 45-75. 

I'd agree that the sample size is relatively small in regards to our lower possession stats, and it should probably also be factored in that those two home matches where we had less than 46% possession were right at the end of the season, essentially in dead rubber matches. 

What the stats do show, beyond any reasonable doubt, is that possession-based football doesn't work for us. By pretty much any segment you'll rarely find much more than 1.3 PPG, combining home and away - but let's face it, you don't need data analysis to tell you that. You only have to look at where we are in the table. Anybody who's been watching us this season knows full well that our style of play is largely shit and the results are testament to that. The fact that our PPG really doesn't improve at all the more possession we have, though, suggests a focus on possession isn't the answer as far as our current players and manager are concerned. Whether less possession is the answer is up for debate. As you say, I don't think there's enough proportional data from this season to make a judgement on that. 

[JacknOry]

Ok, promotion sorted. We just need exactly 66% possession each match. 

[Guest]

2 hours ago, DE. said:

I didn't really have any reasoning for my percentile segments, beyond 10% increments simply being easiest from 45-75. 

I'd agree that the sample size is relatively small in regards to our lower possession stats, and it should probably also be factored in that those two home matches where we had less than 46% possession were right at the end of the season, essentially in dead rubber matches. 

What the stats do show, beyond any reasonable doubt, is that possession-based football doesn't work for us. By pretty much any segment you'll rarely find much more than 1.3 PPG, combining home and away - but let's face it, you don't need data analysis to tell you that. You only have to look at where we are in the table. Anybody who's been watching us this season knows full well that our style of play is largely shit and the results are testament to that. The fact that our PPG really doesn't improve at all the more possession we have, though, suggests a focus on possession isn't the answer as far as our current players and manager are concerned. Whether less possession is the answer is up for debate. As you say, I don't think there's enough proportional data from this season to make a judgement on that. 

True, we have to factor in that smaller sample size, so I'm going to try and figure out a data set across the Championship to see Possession % vs PPG and map out each side. We'll see which sides rank where.

[roversfan99]

With possession, often once a team scores first (and most likely goes onto win) they may happily concede possession to sit back or indeed because of less urgency to probe for a goal. In such regular situations, it would make correlating possession with results almost meaningless. Take the Derby game, we was 3 up after 20 minutes, it wasnt a win built on a lack of possession, the stat due to the way that the game went was essentially arbitrary.

[Guest]

Possession vs Points graph shows us some interesting things across the league. I've stuck in the points needed for Top 6 line and then a line for Rovers Possession % average so we can see the teams below and above each. 5 out of the 6 sides who occupied the Top 6 spots did so with less possession of the ball than Rovers. Only one team, Norwich, had more possession than us in 2020/21. Frightening. 

Are there any trends? Well I've slightly re-skinned the same graphic with a different view so we can see if there's any obvious trends here. In general there does seem to be a slight correlation, to a point, where more possession equals more points, but there are way too many examples of sides bucking that trend to make it a non-starter. Basically each side is different. Norwich can win games with so much ball because they have such a superior squad. Derby constantly chasing games makes their % a little higher, down in those danger spots at the bottom end of things.

Let's re-skin again and go to a cluster view. I think this shows us just how small the margins are, whilst still existing. The top 6 do average a little more as a cluster, with the midtable group slightly below and the bottom 3 just slight below that. Again. Huge outliers like Barnsley, Rovers ourselves and Derby mean that possession vs points per game isn't our best indicator.

Overview for me here is that each side is different. Talk of having less or more possession in order to win games is probably a poor idea, because possession alone doesn't affect enough to be the sole reason for goals for & against and points for & against.

Sheffield Wednesday have been plotted without their points deduction as this shouldn't affect a play based analysis.

Edited June 9, 2021 by JoeH

[bluebruce]

7 hours ago, JacknOry said:

Ok, promotion sorted. We just need exactly 66% possession each match. 

I think what the data more suggests is that when we absolutely and utterly control possession, yes it tends to work. If teams can barely get out of their half it's hard to score on the counter attack, and the pressure eventually tells. You can't expect to do this very often though. It may also more be a symptom of the teams that we are able to achieve that against being poorer teams. Even then, DE's figures showed that even when we absolutely hog the ball in the high 60's, it's still far less of a points return than when we actually see less of the ball than the opposition.

It just keeps backing up what we see with our own eyes, that we are a better counter-attack side and we don't get results when allowed to retain the ball. Last season's data showed it too, or it did up to the point I was tracking it. That's a long time to maintain a trend like this, especially backed up by what we see - I more than expected the numbers were going to say what they said.

I know you were being facetious but obviously nobody is suggesting we aim for a specific percentage, and it's not realistic to aim for the high 60s every week even if that was the better approach, which it isn't.

A different manager and players and maybe it could work, but this has always been more of a blood and thunder league anyway.

[bluebruce]

@JoeH Comparisons to other sides only confuse the matter, because as you say, each side is different. I've already stated my point here and in other threads like the one I linked you to, but I'll reiterate.

Our current style with a hyper focus on possession doesn't work for us. We do better with less of the ball, because we are a better counter-attacking side than we are a tiki-taka side. We have plenty of pace up top to catch out teams who overcommit to breaking us down. Our central defenders are also not the quickest, so are vulnerable themselves to counter attacks if we over commit our fullbacks. It's not so much that I'm saying possession intrinsically sucks so aim for 40%, more that our counter-attack is our best weapon and we don't get to use it enough because we have the ball most of the time. If we invite teams on a bit more, sitting with enough depth to give them the same problem they've been giving us, then spring the trap with a rapid front three, I think it would bring more dividends.

Looking at how other sides do with possession and points is just a side curiosity, because there are many variables at play from side to side. It does however emphasise how ill-suited we in particular are to it, because only one other side (Huddersfield) had possession in the same ballpark as us and yet finished worse. I thought that was where you were going with it at first, but (and sorry if I'm misunderstanding what you're driving at) at the end you seem to suggest possession makes no difference. I don't think that's true, the approach to it and personnel used in that approach make a difference. Not the possession stat itself. When we so consistently do worse with more possession, contrary to past conventional wisdom and the manager's 'wisdom', I don't think the correlation can be overlooked.

One big caveat in my suggesting more counter-attacking though - if we lose Arma, which is likely, our main threat in that approach is gone. I can't say how effective it would be without him.

Edited June 10, 2021 by bluebruce

[perthblue02]

Surely the most important data for any football club comes from the League table and the list of results. ToMo as already said he doesn't bother with those stats , so little hope for anything else. 

Maybe the recruitment stat lads / scouts could be cunning and present the data by way of different coloured iced French fancies next to players names to grab his attention in between naps.

Edited June 10, 2021 by perthblue02

[Guest]

1 hour ago, bluebruce said:

@JoeH Comparisons to other sides only confuse the matter, because as you say, each side is different. I've already stated my point here and in other threads like the one I linked you to, but I'll reiterate.

Our current style with a hyper focus on possession doesn't work for us. We do better with less of the ball, because we are a better counter-attacking side than we are a tiki-taka side. We have plenty of pace up top to catch out teams who overcommit to breaking us down. Our central defenders are also not the quickest, so are vulnerable themselves to counter attacks if we over commit our fullbacks. It's not so much that I'm saying possession intrinsically sucks so aim for 40%, more that our counter-attack is our best weapon and we don't get to use it enough because we have the ball most of the time. If we invite teams on a bit more, sitting with enough depth to give them the same problem they've been giving us, then spring the trap with a rapid front three, I think it would bring more dividends.

Looking at how other sides do with possession and points is just a side curiosity, because there are many variables at play from side to side. It does however emphasise how ill-suited we in particular are to it, because only one other side (Huddersfield) had possession in the same ballpark as us and yet finished worse. I thought that was where you were going with it at first, but (and sorry if I'm misunderstanding what you're driving at) at the end you seem to suggest possession makes no difference. I don't think that's true, the approach to it and personnel used in that approach make a difference. Not the possession stat itself. When we so consistently do worse with more possession, contrary to past conventional wisdom and the manager's 'wisdom', I don't think the correlation can be overlooked.

One big caveat in my suggesting more counter-attacking though - if we lose Arma, which is likely, our main threat in that approach is gone. I can't say how effective it would be without him.

I just think that this research shows that possession isn’t a useful measure really. Whilst we’ve discovered that Rovers tend to average more points with less P%, it doesn’t actually mean there’s a causation. I think this comes from the correlation of us having more p% as a result of chasing games. 

That’s just what I drew from the research really. Managers comments about dominating the ball seem pointless really because most teams in this league actively want their opposition to have the ball.

[DE.]

As far as Rovers' stats are concerned, it pretty much just shows the alternative hypothesis ('more possession = more points') can't be proven, so the null hypothesis ('more possession = irrelevant') is true. 

Obviously possession is just one of a number of intertwined statistics that make up a match. Our style of play, coaching and team selection from match to match all play huge factors as well (amongst other things) and aren't factored into simply possession stats. All we can say from the data here is that if TM claims dominating the ball leads to better results he is wrong. 

Edited June 10, 2021 by DE.

[Guest]

Can we please consider renaming this thread to something a little less condescending if we're actively going to make this the home of all data discussion and try and keep other threads away from straying to that topic? @K-Hod @J*B @Silas Data In Football would do wouldn't it?

[J*B]

27 minutes ago, JoeH said:

Can we please consider renaming this thread to something a little less condescending if we're actively going to make this the home of all data discussion and try and keep other threads away from straying to that topic? @K-Hod @J*B @Silas Data In Football would do wouldn't it?

Done. 

[Guest]

On the possession conversation, it's often a conclusion that more possession in the oppositions half, or the final 3rd of the pitch, would equal more success. But some league comparison shows this not to be the case. Think this, especially when you see how close Rovers' & Swansea's data is here, shows us that possession means very little. We shouldn't draw too much conclusion from pure average ball possession OR positionally adapted possession data.

[Ianrally]

In the words of Jim Royale, “Data my arse, they stuck it in the onion bag twice and we didn’t at all therefore we lost 2-0”

[bluebruce]

4 hours ago, Ianrally said:

In the words of Jim Royale, “Data my arse, they stuck it in the onion bag twice and we didn’t at all therefore we lost 2-0”

It sounds like this might not be the thread for you.

[tomphil]

So in terms of doing anything other than staying in mid/lower mid table the possession game does nothing for us.  It doesn't suit the squad the manager put together and it doesn't play to their strengths.

He's chosen a one size fits all, try to to make it work approach and although some of the stats say it wasn't a disaster the league position and subsequent happy to go nowhere mentality says it was.

God only knows what he'll come up with next but if he changes it again it's probably the right thing to do. Yet that also makes a fool of him and the crap he's trotted out the last 2 seasons. Shows it all up for the complete wasteful experiment its been.

We so so need a change of direction here but we are stuck in a Mogga muddle.

[Guest]

On 13/06/2021 at 09:29, tomphil said:

the possession game does nothing for us

Against some teams I think it does. Wycombe's, Coventry's, Birmingham's, Huddersfield's this season for example. But against sides above us, with more knowhow and game plan, I think we have to be more counter based personally.

[Kie_BRFC]

2 hours ago, JoeH said:

Against some teams I think it does. Wycombe's, Coventry's, Birmingham's, Huddersfield's this season for example. But against sides above us, with more knowhow and game plan, I think we have to be more counter based personally.

Didn't we lose to Wycombe and Huddersfield

[tomphil]

3 hours ago, JoeH said:

Against some teams I think it does. Wycombe's, Coventry's, Birmingham's, Huddersfield's this season for example. But against sides above us, with more knowhow and game plan, I think we have to be more counter based personally.

I agree the counter attack game seems to suit this squad as does the higher tempo get in the faces stuff. To me that's how a young vibrant team should be used to get the best effect as it's making the most of the young energy levels.  Difficult to play like that every game twice a week obviously but if it becomes drilled into them they'll get used to it.

That's the way to get up this division imo as iv'e never known an average side turn into a total football side overnight and prosper. Just doesn't happen it tends to be teams with strong squads who've maybe come down a season or two before that get their act together and football their way back up. 

In truth though as you've pointed out being able to do different things is the best way but can lead to over complication in our case. I just think the core style should be higher tempo slightly more aggressive pressing game. Then if you have to go away to top teams or even at home against one and sit back a bit and keep the ball so be it. 

Odd teams aside i don't think the gap between 80% of the division is that big at all, that's why it's so competitive and unpredictable at times. Select the strongest side that gels and apply yourselves properly and anybody can beat anybody. That's the mentality we should have not this being happy playing keep ball and losing by the odd goal that's crept in as an excuse for possession football.

[StubbsUK]

Anyone here a member of Wyscout or similar?

[Guest]

9 hours ago, StubbsUK said:

Anyone here a member of Wyscout or similar?

Yes can I help?

[Guest]

Don't want to take other topics off point but I think it's worth noting that's ANOTHER win with less than 50% off the ball at our feet. We're so much better as a pressing side. Since Derby away last season I've said we always look better forcing teams to play through us.

Will post some data on this one tomorrow once I've managed to gather it and sort through it.

[roversfan99]

I do agree that endless laborious possession is not a wise tactic, and indeed we do look better when pressing, but I think it is a little simplistic to put it all down to that.

The last 2 games have been fine margins, we haven't offered a threat at all from open play, but we look far better from set pieces which have got us the goals we need.

[ben_the_beast]

8 hours ago, roversfan99 said:

I do agree that endless laborious possession is not a wise tactic, and indeed we do look better when pressing, but I think it is a little simplistic to put it all down to that.

The last 2 games have been fine margins, we haven't offered a threat at all from open play, but we look far better from set pieces which have got us the goals we need.

At the same time, although yes we have conceded in all our games this season it has only been by one goal. Endless possession leads us to over committing players forwards and we looked so vulnerable. Goals from set pieces have seen us take points in games because we're not having to score 3 goals to win.