In Part Three of his statistical analysis of Yeovil Town’s centre midfield, Jonathan Hooper takes us through the varying combinations we’ve seen under Mark Cooper so far this season. You can read Part One and Part Two here.
Midfield Combinations
Before I dive into this, I love data and what it can tell you about what is happening on the pitch, but I also know it doesn’t give the full picture. Data such as xG is done on averages, as an example some teams may play above their xG for a period of time, such as German side Union Berlin did last season to qualify for the champions league but this isn’t sustainable for a long period of time, now this season they’re near the bottom of the Bundesliga. However, xG and other metrics are good indicators of the underlying performances of teams and players.
xG can only show so much, for example Yeovil had a higher xG than Welling despite losing 4-1. If we look at the xG story you can see Welling’s xG was superior to Yeovil’s until the very good Fisher chance late on.
Please bear this in mind as we look at performances from different combinations and analyse the performances.
THE DATA
Yeovil have lined up with 7 (seven) different midfield combinations so far this season in the league. Here they are:
| Combo | Games | Won | Drawn | Lost | Points | Goals | Conceded | PPM | Goals PM | GA PM | Oppo Strength | Expected PPM | Possesion |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Worthington & Cooper | 6 | 4 | 1 | 1 | 13 | 8 | 5 | 2.17 | 1.33 | 0.83 | 10.8 | 1.1 | 56.3% |
| Worthington, Lo-Everton & Cooper | 4 | 1 | 2 | 1 | 5 | 7 | 6 | 1.25 | 1.75 | 1.50 | 12.0 | 1.6 | 55.3% |
| Lo-Everton, Cooper & Williams | 1 | 0 | 1 | 0 | 1 | 1 | 1 | 1.00 | 1.00 | 1.00 | 10.0 | 2.0 | 58.0% |
| Worthington, Owers & Williams | 1 | 1 | 0 | 0 | 3 | 3 | 2 | 3.00 | 3.00 | 2.00 | 12.0 | 0.5 | 40.0% |
| Owers, Lo-Everton & Williams | 1 | 1 | 0 | 0 | 3 | 3 | 2 | 3.00 | 3.00 | 2.00 | 21.0 | 0.7 | 52.0% |
| Worthington & Owers | 3 | 3 | 0 | 0 | 9 | 7 | 2 | 3.00 | 2.33 | 0.67 | 10.0 | 1.6 | 56.0% |
| Worthington & Lo-Everton | 6 | 4 | 1 | 1 | 13 | 13 | 8 | 2.17 | 2.17 | 1.33 | 13.3 | 1.7 | 56.0% |
Worthington & Cooper and Worthington & Lo-Everton have both player 6 games and have the same record, won 4, drawn 1 and lost 1. This gives them a PPM of 2.17 which are both above the teams average PPM of 2.14 so far this season. Worth noting that Worthington & Cooper have played against harder opposition on average. Oppo strength = league position of opposition / number of games, lower the number harder the fixtures.
Next with 4 games Worthington, Lo-Everton & Cooper, lowest PPM of a combination to play more than 1 match together. Then it is Worthington & Owers with 3 games and 3 wins, against the hardest opposition as well.
Quick look at a 2 combo v 3 combo
2 in the middle: 15 games, 11 wins, 2 draws & 2 losses. PPM of 2.33
3 in the middle: 7 games, 3 wins, 3 draws & 1 loss. PPM of 1.71
Cohesion, in team sports, especially football, I believe this is one of the most underrated parts of the game. Fans want new signings nonstop, but at what point does that become detrimental to the team? Just look at the amount of signings Chelsea & Nottingham Forest have made recently and how their performances are in relation. One company I like is Gainline analytics, they are mainly Australian sport dominated but their piece on Leicester in 2015/16 is very good, You can view it here.
Keeping Mark Cooper over the summer gave Yeovil a better start to this season as it bought consistency, and his success in retaining many of the squad. During the winning run the back 5 stayed the same throughout, this can only help the team grow and understand as it moves forward. However, the rest of the team changed frequently, below I will go through the games, trying to read Mark Coopers mind, or should we call him Claudio Cooper after Claudio “the Tinkerman” Ranieri, lets find out…
START OF THE SEASON
First 4 games of the season, it was consistently Worthington & Cooper. Going by xG Yeovil slowly got better, the gap in xG between the teams grew each game. Yeovil finished ahead of their Expected points despite the loss to Hemel.
Worth pointing out Jordan Maguire-Drew, started just in front of the 2 in in midfield for each game.
| Opponent | Points | XG Diff | Expected Points |
|---|---|---|---|
| Hemel Hempstead Town | 0 | -0.12 | 0.70 |
| St Albans City | 3 | 0.23 | 1.50 |
| Truro City | 3 | 0.29 | 1.50 |
| Maidstone United | 1 | 0.33 | 1.50 |
Yeovil finished ahead of their Expected points despite the loss to Hemel. Actual 7 Expected 5.2.
Next up was Tonbridge, we started with Worthington, Cooper, & Lo-Everton for the first time. We dominate the game, xG difference is 1.10 and xPts is 2.3. Here its looks like a good change from the gaffer, 2 goals and a clean sheet. JMD didn’t start this game, in his place was Jordan Young for his first start of the season.
Next up was Havant after a 2 day turn around. Cooper selected the same trio as Saturday we started with Worthington, Cooper, & Lo-Everton, 2 changes in the back 4 and all 3 up front. However the system remained the same. We lose 4-3, we also lose the xG battle by 0.66 and our expected points is 0.5.
THE UNBEATEN RUN
Chelmsford at home sees the introduction of Williams as a DM, playing just behind a duo of Lo-Everton and Cooper. Yeovil win the xG battle by 0.75 and their xPts is 2.00 but the Glovers fail to make their chances count and draw 1-1. A positive performance.
Next up was Weston away, first of plenty Somerset derbies this season. Yeovil start Owers for the first time, Worthington and Williams joining him. Cooper misses via a suspension. Yeovil win 3-2, thanks to 3 set piece goals. However, the data suggests Yeovil got lucky. Losing the xG battle by 0.66 that means the Glovers expected points from the game was 0.5. This is where the xG story I mentioned above comes into it a bit better, Yeovil scored 2 goals before the 20th minute. WSM got one back from a long range effort late in the first half. Yeovil get a 3rd in the 55th minute. Yeovil shut up shop and WSM throw everything at it and get a goal late on. Interestingly this game was Yeovil’s lowest possession game so far this season in the league with 40% of the ball.
Last two games, going by data, dominate one and draw, get dominated and win.
Eastbourne were next on the list, Yeovil lined up with Owers, Lo-Everton and Williams in the middle. Worthington started as well but in more of a left wing role. Again according to the data Yeovil were lucky to come away with the win. losing the xG battle by 0.29. their xPts was 0.7.
Winning whilst not playing well, champions stuff. Despite the wins, was Cooper happy? I don’t know but for the next match he reverted back to…
Worthington & Cooper to face Hampton and Richmond. Yeovil got the win but once again lost on data and xG difference of -0.15 and xPts 0.7. However, H&R got their goal from a late penalty, in xG a penalty is worth roughly 0.75, which means on average a penalty is scored 3 out 4 times. If you were to take the penalty away, Yeovil would have won on the data.
Aveley were the next visitors, Mark Cooper changes the midfield again, due to an injury to Charlie Cooper. Incomes Josh Owers to partner Matt Worthington. this is the midfield duo for the next 3 league games (4 if you include FA Cup against Southend) These games are also the first time Yeovil reverted to a more traditional 4-4-2. Here is how the data lines up
| Opponent | Points | XG Diff | Expected Points |
|---|---|---|---|
| Aveley | 3 | 0.64 | 2.00 |
| Worthing | 3 | -0.19 | 0.70 |
| Weymouth | 3 | 0.85 | 2.00 |
A pretty successful run of games, winning on data comfortably on two of them. The other Worthing were awarded a penalty late on, same as the H&R game. Without that penalty Yeovil would have won each game by over 0.5 on xG.
Despite the success of this partnership, it was broken up for the Braintree game. I am unsure if this was because of injury to Owers or another reason but it was the start of a six game run for a new partnership Lo-Everton and Worthington.
| Opponent | Points | XG Diff | Expected Points |
|---|---|---|---|
| Braintree Town | 3 | -0.07 | 0.70 |
| Torquay United | 3 | 0.35 | 1.50 |
| Dover Athletic | 3 | 3.00 | 2.70 |
| Farnborough | 3 | 0.34 | 1.50 |
| Chippenham Town | 1 | 1.01 | 2.30 |
| Welling United | 0 | 0.08 | 1.50 |
You can see against Braintree it was a pretty even affair, not helped by the weather. Yeovil got a bit lucky in this game as the data shows. Then a strong run of form winning each game on data as well as on the pitch.
Chippenham, looking at the data it was one of our best away performances of the season, we did struggle to score on the day but the xG suggests we should have been comfortable winners.
Welling, as I mentioned at the start, we won the xG but they were leading until late on. It wasn’t Yeovil’s best performance of the season but not one where you’d expect them to concede 4.
RECENT GAMES
After the loss to Welling Mark Cooper decided to change up the midfield duo and formation.
| Opponent | Points | XG Diff | Expected Points |
|---|---|---|---|
| Bath City | 3 | -0.16 | 0.70 |
| Dartford | 1 | 1.65 | 2.70 |
| Hampton & Richmond | 1 | -0.13 | 0.70 |
Bath game, Cooper replaces Lo-Everton and partners Worthington. we got lucky, had one chance and took it. That said Bath never really created a golden opportunity and only just beat us on xG. Dartford, we revert back to 3 in midfield with Worthington, Cooper and Lo-Everton, the data suggests we played well, created numerous chances but didn’t take all of them, whilst Dartford were the clinical ones. Similar to the Chippenham game.
Hampton & Richmond, continuing with the trio in midfield of Worthington, Cooper and Lo-Everton. Pretty even game one that we did lose the xG battle on just but a draw was probably a fair result.
RANKING
I have scored each of the combinations, based on fixture difficulty and xPts, here is my rankings
| Ranking | Combo | Score |
|---|---|---|
| 1 | Worthington Owers | 0.47 |
| 2 | Worthington Lo-Everton | 0.28 |
| 3 | Worthington Cooper | 0.22 |
| 4 | Lo-Everton Cooper Williams | 0.20 |
| 5 | Worthington Lo-Everton Cooper | 0.16 |
| 6 | Worthington Owers Williams | 0.13 |
| 7 | Owers Lo-Everton Williams | 0.10 |
Worthington & Owers come out on top, do you think they should start more often?
HOME AND AWAY
I am often interested in how teams fair differently playing away from home. Yeovil’s PPM is 0.80 per game less away from home. What causes this I do not know. Teams do seem braver at home. The fact Yeovil have lost to 2 of the bottom 4 also baffles me. Lets have a look to see what combinations have done what home and away
| Combination | Home Games | Home Points | Home PPM | Home xPts | Away Games | Away Points | Away PPM | Away xPts |
|---|---|---|---|---|---|---|---|---|
| Worthington Cooper | 3 | 7 | 2.33 | 4.50 | 3 | 6 | 2.00 | 2.10 |
| Worthington Lo-Everton Cooper | 2 | 4 | 2.00 | 3.00 | 2 | 1 | 0.50 | 3.20 |
| Lo-Everton Cooper Williams | 1 | 1 | 1.00 | 2.00 | 0 | 0 | 0.00 | 0.00 |
| Worthington Owers Williams | 0 | 0 | 0.00 | 0.00 | 1 | 3 | 3.00 | 0.50 |
| OwersLo-Everton Williams | 1 | 3 | 3.00 | 0.70 | 0 | 0 | 0.00 | 0.00 |
| Worthington Owers | 2 | 6 | 3.00 | 4.00 | 1 | 3 | 3.00 | 0.70 |
| Worthington Lo-Everton | 3 | 9 | 3.00 | 4.90 | 3 | 4 | 1.33 | 5.30 |
Conclusion
In the first part what struck me was the slight imbalance of having 2 or 3 right footed players in central midfield. I didn’t have the data to compare the midfield to a midfield with one right and one left footed player.
The second part we could start to see different aspects of players games from different metrics and stats. each player had their stand out aspect, from Owers being more feisty and getting stuck in to Lo-Everton creating chances.
In this part we can see how the different combinations fair. Owers & Worthington finished top in my metric, is that because of the balance of a right & left footed player, luck or something else? Do the tweaks that Mark Cooper makes help or hinder the side over a long term?
As Mark Cooper said in his recent press conference, we may not be winning every game currently but the data performances are generally good, if not better now than they were during the winning run. As I said at the start data can only tell you so much, if you take an early lead then look to control the game may be a success getting results but the data looks poor. We all know which one we would prefer to win between actual points and xG/xPts.
What we can see from the data is that every player and every combination has their strengths & weaknesses, lets all hope Mark Cooper can get this best out of them. Go on the Glovers!
What duo or trio would you pick to start?
SPEED OF PLAY
One thing I’ve heard recently from other supporters is that some players slows down play, I do not have the data to prove or disprove this. However, as a bonus, below is a chart showing passes per 90 and passing rate for every team in the league. Higher the passing rate the faster a team moves the ball.
Do you have any thoughts about Jonathan’s epic three-parter? Leave a comment below.
I think Mark Cooper stated after the Weymouth game Owers looked tired & needed a rest. Given the results when he played I’m surprised he’s been out of the team for such a long time (often not even making the subs bench).
I think on the whole, Mark Cooper has managed the rotation of the squad very well this season, considering the huge number of games, quality (or lack of) in some of the pitches, weather, injuries, etc. Cooper and Worthington are solid – CC will bring the ball forward and be brave on the ball, MW works very hard. If playing both then I think we look more effective with a creative/more dynamic-type (Young or Lo-Everton) playing centrally in the #10 role – finding pockets of space and opening up space on the wings for Michael Smith to bomb forward (the Wannell to Smith cross-field ball is something they have clearly been working on lots). Having so much flexibility is definitely helping us though – vs Bath on Saturday it was interesting to watch how MC changed the system and personnel depending on the state of the game (e.g. Smith noticeably less adventurous in the second half/Whittle and Foulston both on the left to shut down Bath’s dangerous wingers). Call me greedy but I would like to see us be a bit more dynamic, particularly when winning 1-0, to kill teams off. With our quality we should be winning 2-0/3-0 more often (e.g. Taunton on Boxing Day). Easier said than done when teams are playing deep and narrow though.