Dafuge & Data: How to Solve a Problem Like my Right Back
We’ve got a problem. And I’ve kind of telegraphed it in the title. It’s like a whodunnit where they reveal it at the start and then work backwards. Except instead of a butler who has committed a crime we have a situation where decent football has died at the hands of my terrible right back.
First signs of a problem
The first few months with the Seadogs has been going well. We brought in some fresh faces to fill out the squad and they’ve hit the ground running. A few wobbly moments but new strikers Varion and Odejayi have formed a promising pairing. Longball seems to be working, and the sacrificial bung performed in pre-season must have pleased the elder hoofball Gods.
We did hit a worrying patch though. Suddenly the wins dried up and the yellow draws started to creep into the form book. As we were overperforming we could just have been returning to mean. Basically what goes up, must come down, and vice versa. Statistically, there is a mean, and whilst you might deviate from it for a while you should eventually come back to it. So if we were just over-performing earlier on then it would stand to reason that we would see something of a drop as we found our level again. You can see the reverse happening when relegation fighting teams get a bounce from a new manager.
For a manager to be sacked they usually have to be underperforming in comparison to expectations, when the new manager comes in they do so at an opportune time. Right about when an underperforming team should be due a return to the expected levels of performance (according to Soccernomics anyway).
But maybe we are sort of good? Maybe we are meant to be in the top half of the table? In that case, it’s worth looking closely at the slip in results to stop the slump. And even if we aren’t actually good and the mean is on the horizon, we could still take a look to see if there’s anything we can do to postpone the inevitable and steal a few points.
A common theme in the worrying run of results was late goals. Or at least I felt I had noticed a problem in the amount of late goals. A few had stuck in my mind. But rather than rely on my memory I checked against the times of conceded goals below.
And it actually was a case of late goals having more of an impact in the memory. There’s a spike in the 80-90 range to 3 goals. But that’s not huge. We have a slightly bigger spike just before half time with 4. It maybe suggests a slight concentration issue rather than a tiredness issue.
But more importantly, it’s not conclusive of there being a big issue with last-minute goals. Last-minute goals just stick in the mind because of two reasons. The recency effect, and pain. The recency effect is something covered in the psychology of memory. We are better at remembering things that happen at the start of something, and the end (more recent events). Even though last-minute goals aren’t representative of what’s going on (only 3 out of 25! 12%) they are more available in the memory. The mind is a fickle thing and because it’s more available we assume it’s more common (here endeth Psychology of Memory 101). Oh, and I mentioned pain. Last-minute goals are important and they can painfully change a result, robbing you of points. They’re more emotionally charged than a 30th-minute goal with an hour left to salvage things.
What else could be the root of the problem then? One thing to always check is just how rubbish your keeper is. Don’t assume they are good. Assume they are the love child of Taibi and Hart and work from there to minimise disappointment.
I’ve always struggled with keepers, so last year I used xG against to work out whether my keepers were efficient and saving more than they should be on average. Or were terrible and were leaking more than they should. I did it for Belfast, by hand, and managed to slowly plug the leaks.
The game has it in now so it’s easy. Check what your xG against is for the games your keeper has played and total it. Check how many they have conceded. Take one away from the other.
We had conceded 25 goals at this point. So we’ve shipped almost 7 goals more than you would expect on the balance of probabilities. Bad, Keeper. Bad. It get worse if we just look at the last 4 games. An xG against of 4.09, but 7 goals against.
Is it all his fault though? Well, a lot of it is. But is our terrible keeper being put under undue pressure because of fellow bunglers in the team? Our keeper has faced 190 shots, 75 of them on target. There are busier keepers out there but that’s still a sizeable amount of action.
Where is the problem coming from?
We’re not throwing the keeper to the lions yet. He’s a problem but one we can deal with in the offseason when no one is looking. There’s no one to sign either. Maybe it’s a case of a weak link in the defence causing problems and putting pressure that could be avoided.
Diving a little deeper a few things jump out. One is that a lot of goals are coming from central areas. The opposition are getting into good central spaces for their shots and punishing us with the higher xG chances. Crunching the numbers again on average each shot against us has an xG of about 0.1 or 10%. We need fewer shots faced, and from poorer positions to plug the leak. The keeper clearly isn’t going to do it himself.
From the assist types, the biggest issues appear to be freekicks and crosses. Or from positions that are or might be out wide. If we add square balls to this and corners then we have a case for saying either we need to reduce the chances they are creating from wide or improve the centre of defense in the air (or positionally).
We have our suspects. The central defenders, or the full backs.
I’ve got a few key attributes and stats on one of my defensive squad views. I’ve just looked here at our normal, 1st choice, back four. Below I’ve cut a few of the stats out that were fairly even between defenders.
A few things stand out here. First is the lack of key tackles. And that’s because FM21 is terribly bugged when it comes to stats. It’s lacking some polish. They are making key tackles. It’s just not being recorded anywhere. Another painful issue is that I can’t get blocks or clearances on here. They are on the detailed team lists for the division though. But just not here.
Looking at the squad view e can see that short arse Turner might be an issue. He’s not good in the air and the stats show it as well. Our DC’s look a little ropey because of the lack of tackling per game but the picture becomes a little clearer if we traipse over to the league stats.
Looking at the detailed team stats to fill the gaps reveals a few things. Richards and Weledji are near the top of the tables for interceptions, blocks, clearances and headers/key headers. I think it’s fair to say they are dealing well with the aerial game and by extension crosses. They don’t look like a weak link. Jackson the leftback isn’t far off for tackles and headers either. But Turner is nowhere to be seen until you get to this…
So he’s not putting a huge amount of tackles in. He’s not winning the ball in the air well. He’s not blocking, intercepting or clearing as well as the rest. And he keeps losing possession. About 17 times a game on average. Now some of this could be from long passes going awry. But some could also be from losing the ball in a difficult position.
I should point out here as well that context is important. When we look at his radar in some areas Turner compares well to the league average. But I’m not bothered too much about the league average, especially seeing as the league is gash. In a match situation, he’s not competing in a tackling competition with the other defenders after all.
And even with that in mind he’s not intercepting, heading, clearing, or even passing favourably. He’s better than the league average only in fouls, attempted (note attempted not successful) tackles, and blocks (which I do value to be honest).
Checking the Matches
So that stats say Turner might be a weak link. He could be a problem. The final check is to look at him in action again and see what’s happening when he is involved. Now it’s worth pointing out that the lack of timeline on the match view makes this difficult. The fact that the mistakes view doesn’t seem to actually fill with data makes it downright painful.
As much as I back my team I doubt they made no mistakes in this match against a team a league above us in the cup. I also doubt, looking at our other games, that we’ve never made a mistake. So gone is the opportunity to easily select a mistake and then view the attached highlight. Top work.
I did comb through some of the footage though. I saw plenty of poor marking, poor positioning and just a general lack of effort in trying to make an aerial challenge. Probably best summed up by this.
The cross comes in, and Turner is tracking back too slowly. He’s too far away, and he doesn’t even try to challenge in the air. The ball floats over him and the keeper is left with footballing pie all over his face. This combined with several instances every match of him playing a wayward crossfield pass putting us immediately under pressure again did not help.
Stay of Execution
Time to drop him right? He’s a weak link putting my already below average keeper under pressure. Well no. I mean he is, but no to completely dropping him. First I needed to get in some cover with what little wage budget I had in.
Liverpool youth reject comes in. There are no match stats to judge him on from his time at Liverpool but he seems like a better marker, passer, and header. He’s not great positionally but he does make better decisions. And just like his 5ft 9 twin he can’t jump. It’s not a vast improvement but it is some, and from a player with more potential. He should be a little bit better defensively now, and potentially a lot later on.
It took about a month to get Wilson in. So that gave Turner more time to spoil things for me.
The Halo Effect
If you’re not convinced by stats I can imagine you’re just asking why I didn’t watch the games and notice he was a complete liability? And the answer is I do watch, usually on comprehensive. But what was saving Turner was something called the Halo Effect. In the screenshot above he had 6 assists. By the end of the season, I think he had 10 in all competitions which isn’t shabby for a limited fullback. And that’s what was sticking in my mind. His attacking threat was much more exciting and memorable and than him not being able to jump.
I had a favourable opinion of him because the assists were sticking in my mind. Even though they weren’t what I needed from him they made me view him more favourably. Even though they weren’t representative of what he was doing every match (like the lost possession and lack of aerial challenges were) they covered for his many problems. And for me this is why you have to go back to the stats sometimes. Your impression of players and performances is skewed, it’s biased. Because the mind is a lazy thing and takes short cuts. But stats tend to lay bare the reality which you can then go back over and check.
Problem solved right? Well sort of. I have cover that might be better and shore up the defense a little whilst I wait for the summer to serve my keeper and Turner with their P45’s. Wilson may be better defensively for crunch games so they can go in. But Turner with his assists will get some game time if I’m desperate.
The real problem solving should come in the offseason when I’ve got a season’s worth of data and I’ve cleared deadwood from the wage list.