Machiavellian Moves: Setting Snares with Defensive Corners in FM23
I am a big fan of set pieces in Football Manager, but have often looked at the attacking opportunities rather than spending much time on the defensive options. There’s a lot of exceptional guidance on how to get the most out of your attacking set pieces. Whether they are corners, free kicks or the impressively powerful throw in. Like a lot of people, I’ve been drawn towards the attacking possibilities like a moth to a flame. Or Rory Delap to a towel on the touch line.
They are a great way to squeeze more goals out of your game. They are great for those incremental improvements, the marginal gains, that can change your season. Once I’ve got my main tactic set up the very next I do is tweak the attacking set pieces, drawing heavy inspiration from Guido Merry’s work at Strikerless, to get the most effective set-ups I can.
But during FM23 I’ve found myself under a lot of pressure at times. A lot of the time, my tiny Mumbles side has been absolutely pelted and pinned backed by set pieces. I’ve been on the receiving end and I’ve not enjoyed it. So I’ve looked to the patron saint of philosophical shit housing, Machiavelli, for defensive set piece inspiration.
Well what does Niccolo have to say about this? His work is bursting with sinister and cynical quotes. But I think the below fits well.
“The lion cannot defend himself against snares and the fox cannot defend himself against wolves. Therefore, it is necessary to be a fox to discover the snares and a lion to terrify the wolves.”
Now, we are not the Lions in this scenario. We are the poor foxes getting tossed around like sad rag doll by the wolves. The opposition are the bigger boys who have come to steal our ball and kick over the pile of jumpers we’ve been using as goal posts.
But we can use the cunning of the fox to discover the snares and set them against the Lions, or Wolves, or Bears. Whatever. We are going to use our defensive set pieces as a secret attacking outlet. We are going to turn pressure into goals. By drawing teams further up the pitch because of the positions they take during their attacking set pieces, we can use any regained position and quick transitions to get up the pitch and into positions of maximum opportunity. Something we have done in past FM’s. We can even up the numbers between attack and defence in a way that we might struggle to do in open play as the under dogs.
New ‘Snare’ Set Up
It’s not especially complicated. We need to have options that stay by the halfway line ready to receive a quick pass or clearance. We need a player on the edge of the area that can either carry the ball into space and then lay it off to the more advanced player. Or that can move into space themselves without the ball if the pass has gone long to the advanced player.
There’s a variation of this for those feeling brave, with two upfront and one outside the area. I mix it up a bit, but I quite like that variant as it ‘fits’ with the attacking trio I have of two strikers and one attacking midfielder behind them.
It should in theory work with most formations though, as it’s not so much about player roles but the quick transition to attacking players that leaves the bulk of the opposition chasing our shadows. It might only buy us a few seconds of overload or better odds (2 v2 etc.) but that’s long enough to turn a defensive corner into a goal for us.
The more eagle eyed will note that I have the AMC and ST up front. Where is STCL? Who knows. I can’t find them on the set piece creator. I’m assuming a bug here, but basically they appear to be AWOL on defensive set ups.
Extra Considerations for the Corner
There are a few extra things to do as well to improve the odds. As a team instruction we have counter on, and the directness of our passing is pretty quick. I also have a version of the above with the fullbacks marking the post on their ‘side’ so they can also run and support in the right place rather than having to dash across the field to get in place.
Our keeper is encouraged to get the ball out quickly through the instructions of…well, distribute quickly and long kicks. Sometimes we will be clearing the ball, keeping it alive and passing using outfield players. And sometimes the keeper with be able to save/catch and distribute themselves.
Something that can make the keeper even more deadly in this situation as well is the player trait of ‘Uses Long Throw to Start Counter Attacks’. It is amazing. A keeper with a half decent throw can get it into the space beyond the penalty area at least to the player already waiting for it and start the move. If your keeper has been training for the worlds strongest man then they can sometimes get it to one of the other players further up. Combined with the long kick instruction it should mean that one way or the other if your keeper gets it they are going to try and bag themselves an assist.
But what’s the catch? It all sounds quite nice but if it’s all as peachy as it sounds why isn’t this the default? And I think that is simply because every tactic has a cost. In this case we are moving players away from our box and losing defensive cover. They might not be the best defenders but I’m losing 2-3 players that could mark or disrupt or get a second ball in the area. I’m sacrificing potential defensive stability and cover for the chance of more goals.
And the Stats?
Whilst I love to hoof and hope that hope is actually just firm belief in teachings of Big Sam and stats. Data is key. If I’m going to do this I need to know that it’s supported by the data. I need to know more than just do I score some goals from it. I need to know whether I score more goals than I normally would, and that I don’t concede more as a result as well. It would be nice to get some insight on the change in xG as well.
The Data Set
I’m about 8 seasons into my save. I made the change after two seasons. So I have a data set of 2 seasons data of standard set ups and about 5 and half with the snare. That equates to about 60 games on standard, and 176 games with the snare. Not a balanced data set but I’ll take what I can get.
I don’t want to waste my time or yours by looking at everything. I just need to look at the variables that potentially indicate whether this snare has had a positive effect. My hypothesis is essentially that this use of the snare should lead to a significant improvement in our performance. How I measure or operationalise performance is the important thing here. There’s no point collecting everything and looking at everything willy nilly.
So I’m going to focus on:
Goals from Defensive Corners This is a pain to measure. I have to do this manually because there isn’t really a stat in the game for goals you score from corners you have defended. I also need to work out when my counter-attack ‘ends’. For this I’ve decided to use a Charles Reep inspire window of seven passes. Hugely flawed work in general, but more than seven passes I think can comfortably indicate a period of passing possession rather than a quick counter-attack.
Goals conceded from Defensive corners – The easier of the two, as this exists as a stat in the game.
Average xG per game generated by Defensive Corner counter-attacks – In for a penny in for a pound. I’m already making my eyes bleed manually counting I might as well throw the xG in.
Penalties Resulting from a Defensive Corner counter-attack – Sort of self-explanatory. How many extra penalties do we get from these counter-attacks.
Extracting the Data
There are more I’d be interested in looking at. Like xA involved in the counter-attacks, progressive passes involved. But I’m but one person. Admittedly, a nerd of a person with stats knowledge to burn, but I have to sleep at some point. To collect the data I was interested in I had to go to the analysis view after a match, filter to show opposition corners, and the go through each corner event to select linked events so I can see what happened. From there, I’ve been able to count and calculate. If I was just interested in goals I could have done the slightly quicker route of looking at my goals and the linked events, tracing back.
For speed I ended up basically cross-referencing the time of the opposition corners, with the times of the shots taken. So I could work out which shots followed a corner closely enough to be worth checking the linked events for. I ended up with shots visible and corners visible. Which meant a fair bit of sifting through shots and data that was relevant (our attacking corners and the oppositions shots).
I don’t recommend this manual crunching. And in fact, if I were to do this in the future, I would use a smaller sample to get a rough estimate and then update the estimate every few games rather than every game.
But, all going well, you should end up with nice linked events like this. The corner came in, the midfielder on the edge of the area collected it and punted it up field to the AMC, who then put through the striker for a shot on goal (with an xG of .07).
Before making the changes we had scored 3 (non-penalty) goals within 7 passes from a defensive corner, in 60 games. Bit of a mouthful, but the number was low. If we calculate per season that was 1.5 goals per season or 0.05 goals a game. We were generating on average a tiny 0.01 non penalty xG a match from defensive corners. Which I wasn’t surprised by when looking at the corner highlights, or low lights. Many corners led to a save or shot. The few that didn’t most were cleared into touch or back to the opposition as we just didn’t have any real outlets. Technically we were outperforming our pitiful xG a match but both numbers were so small as to be pointless.
We had conceded 3 goals per season from defensive corners (or 0.1 a game). To be fair here this was a really decent number, and probably down to the meat mountains we play in defence (an average backline height of 6′ 3 will do that).
At the other end we had won 1 penalty in total so, .5 of a penalty a season.
We managed a cool 14 goals across 167 games from defensive corners. About 0.08 per game, or 2.4 a season. A definite improvement in comparison before. We’re not talking huge numbers still but when the numbers were small to begin with you have to take what you can.
I think more tellingly our non-penalty xG per game from defensive corners had shot up to 0.09. You can be forgiven for not being wowed by that but it’s massive increase over the previous value, and at the equivalent of 2.7xG per season meant we were actually under performing our xG.
Not so positively, we were now conceding 3.8 goals per season from defensive corners (or 0.12 goals a game). It looks like having a few bodies back was having some impact. But by the same token having more people up front also meant we were winning 1.8 penalties a season
It perhaps gets a bit more telling if we look at things as a percentage change. Now personally I like to crunch the numbers more than this and look at statistical significance etc. but the nature of the sample and the variables means it’s not that useful in this case. We are also just applying this to our own team rather than FM as a whole so it’s less useful. Percentages will do.
A 60% increase in goals, and an 800% increase in xG per game from defensive corners does feel like it was worth it. And whilst the goals conceded did increase it was not by the same magnitude at all (20%). I think as a cost/benefit calculation it has been worth taking some of those players off defensive duties and pushing them further up field ready for the counter attack.
Aren’t you missing some stats?
Yeah I am. I’m going to pass the blame onto FM, but it’s hard to extract stats like how many progressive passes resulted from the new routine without manually counting each. There’s no easy filter for it. So I’ve just concentrated on the main takeaways.
Another reason for stripping it down is also the confounding variable of improvement over time. By this I mean each season we get better. We’ve been promoted, signed better players, and generally upped the quality of the squad. This is something that could skew the stats. And the more stats I report, the more opportunity there is for that bias to exert itself. By reducing the number of comparisons you make, you are reducing the likelihood of false positive (type I errors for the nerds in the know).
A third and final reason is because I overwrote some mid-season files as well, leaving me we only half seasons of data to look at. I’m normally very good at having a couple of saves per season so I don’t lose any data, but I fucked it. Ooops.
Would Machiavelli Be Proud?
Did we hit the Machiavellian brief for FM23 with these corners? Well, Mumbles Rangers were weak foxes in a league full of Welsh Wolves, and we did set a snare. Not everyone fell for it but I think an 800% increase in xG was worth it and a marker of some success.
Whether the minute increase in actual goals is worth it overall is a different matter. We are talking small numbers. But small numbers add up and become bigger ones. And given you can set up the routine and just leave it the few minutes of investment it takes are probably a cost I’m happy to accept.
It has made me really wish for a nice way to analyse the impact of all set pieces. Maybe for FM24?