Machiavellian Moves: The Reducer in FM23
The Reducer. Letting them know you are there. Putting your mark on the game and the other player’s shins. We’re not strangers to that here at FMTahiti. But it’s something we’re going to revisit and check out with a few stats. That’s right, violence and numbers. What a combination. It does get stats heavy towards the end but there’s also a GIF of Pepe breaking the Geneva convention if you read on.
What’s the point of a bone crunching tackle, legal or not, though? Tactically what’s the point? Well, you can listen to Machiavelli or Vinnie Jones.
It’s partly psychological. Instilling fear and hesitation. As Vinnie Jones points out letting them know you are there and that they might be next. There’s a great BT Sport documentary about the Crazy Gang (that seems to have vanished) where someone points out when the ball is up in the air you gain an advantage if the opposition are scared into looking for where the next elbow is coming from rather than where the ball is going to land. Machiavelli perhaps makes the point a little differently (but often).
What it is
When brutality is the only option left, it is holy.
If you don’t have many options then brutality is next to godliness. Preach Machiavelli.
For, in truth, there is no sure way of holding other than by destroying
Getting a bit philosophical and meta-physical with FM23, but the heart of creation is destruction.
Without an opportunity, their abilities would have been wasted, and without their abilities, the opportunity would have arisen in vain
This is perhaps a little more tactical from Machiavelli. If you are constantly crunching the bones of a player, stopping them from playing, taking part, and potentially even staying on the pitch, then you have taken them and their skills out of the equation. Forget Total Football, we are talking Total War. Lets take a leaf out of Sgt. Zim’s book from Starship Troopers.
If you are a small club and can’t go toe to toe with the bigger boys in terms of skill then this is how you level the playing field. It’s how smaller sides have done it for years. It’s much maligned, but in real world football it works. What about FM23?
Route One Rovers and the Reducer
In previous blogs in the series you’ll have seen me managing the mighty Mumbles Rangers amongst other teams. I’ve actually jumped ship to an entirely different save. Using DanFMDatabases 13 Level database for England I’ve gone all the way to the bottom to find a team. I do love building a non-league side up, just like in my Seadogs save.
I searched some regional divisions near where I’m from, and one team jumped out. Route One Rovers. How could I of all people resist the call of a team called Route One Rovers, just in case the Sunday league vibes were not strong enough.
In reality they are a fairly new but quite successful team from Bradford. They have won some silverware and seem to be on the up. I don’t know much more about them but I’ve emailed them asking about the inspiration for the name. I’m hoping it’s because they are Big Sam devotee’s.
Why does the team matter though? Well once again I’m in a position where I don’t have money or pull in the transfer market. I’m not likely to produce wonder kids or train the players I do have to a decent standard. I’m in a position where I need to squeeze all the advantages out of match I can. I’ve got two promotions in a row, my team are already psychotically violent on occasion, and we play as route one as we can (hoof has been turned up all the way to 11). But we need more.
Bring on the Reducer.
What do we consider a Reducer in FM23?
We have a few different opposition instructions to choose from. So what are we going to use to emulate a reducer? All of them? Probably not. For me a reducer is a fairly robust tackle that needs a bit of a run up. It’s a tackle with momentum and follow through. Not a standing tackle. It’s not jostling and shirt pulling. It has a crunch. Take a look at some examples! The Pepe one is just for fun.
Very much a case of playing the man, not the ball. For me I’m going to just use hard tackling and trigger press to emulate the reducer. A common factor is quickly closing the distance to put a tackle in. That is key. They are not being tightly marked or shadowed. They are on the defender’s radar, and the moment the ball gets to them the trap is sprung.
We could use show onto weak foot and mark tightly but for me that is less a reducer and more just trying to completely swamp the player out of the game. And I don’t want to do it that way, and probably can’t with how poor some of my players are. Many can’t mark. Most can kick hard.
Our Reducer Target Criteria
For our purposes putting a reducer on someone is going to be the equivalent of triggering a press and tackling them hard. So far so simple. But how do we measure if it was effective? What is a positive effect of the reducer for us?
One measure could be whether we completely stop a player. If the key stat we’ve used to target them doesn’t improve at all then we could consider that to be success. For example, if they had 3 key passes before we targetted them and they end with 3 key passes. That’s a bit all or nothing though and doesn’t give any wiggle room for FM counting stats poorly, or a freak effort.
In line with the reducer theme maybe we should measure success as reducing the impact the targeted player has. With that in mind, I’ve settled on a simple approach of multiplying the impact the player has had in that key stat as being the benchmark. How much it is multiplied depends on whether we made changes at 30 minutes or at 45. If it’s made at 30 minutes then we multiply by 3, and if it’s at 45 we multiply by 2.
Let’s say John Victim has got 2 key passes at the 30-minute mark, and we mark him our target at that point. Then if he finishes the game with 5 or less key passes, we’ve stopped him. We’ve stopped him accruing key passes at the rate he has set in the first 30 mins.
If we had waited to half time, and he had 2 key passes, then we would need to limit them to 3 key passes at most to count as stopped.
Extra Reducer Stats
I’ll also record whether they were subbed or moved, injured, received a card, and what their average rating change was after the reducer was put on them. I’m not too bothered about the match rating as it can be so wildly skewed by basic events and the weighting that prefers attacking actions. A freak assist for example c could give a big boost even if they have been quiet for the whole game. I’ll record it still, but out of interest rather than expectation of it being a good measure.
I think subbings will be an interesting secondary measure. If we are truly stopping them, or hurting them, then I’d expect to see them subbed off more than chance would suggest. All things being equal, there being 3 subs, and assuming the keeper isn’t going to get subbed, then the normal chance of getting subbed is 3/10 or 30%. We get significantly higher than that then we are cooking with gas.
There are some issues with this approach. One is that it focusses only on one or two key stats. I’ll be using key passes as my main focus, but may use CCC, dribbles and pass completion for tie-breaks. But that is a limited range of stats. But we have to work with what we have, and to go more in depth would take me more time than I have.
Another issue is that these could be skewed by non-open play stats. So I’ll have to do my best to remove them. I don’t want a player getting an easy pass because they also happen to take free kicks, for example.
For the stats savvy as well you might have noticed that the multiplication approach means a quiet first half where only one or two key stats are hit give you very little wiggle room in the second half for stopping them. A player with 2 key passes on needs 2 more to no longer be ‘stopped’ whereas one with 4 could make 3 passes and still be considered ‘stopped’ in the second half. It’s something of a floor effect when low numbers are involved so the best way round it is to think of percentage performance rather than outright numbers. The second player getting 3 passes has worked at 75% of their previous half’s output. The first player with 2 passes is working at 100%. It’s all relative.
Thinking of it as percentages also better covers how effect a reducer might be against different strength opposition.
Why not pre-match or man marking?
Good questions. Why am I doing this on the fly rather than ahead of the game. Like Vinnie Jones says in the video above he’d have started his run up two weeks before the match. The simple answer is that in the lower leagues with poor scouting it is hard to identify who the best players to put the reducer one would be. It’s not impossible but it’s hard.
Another related issue is then how to set the criteria. What counts as stopped when you don’t have a match baseline? You would have to work out their stats per game, which is doable but again more work.
I’m avoiding man marking because I also don’t have an outright goon/enforcer to do the work, and need to be responsive to where on the pitch the threat is. No point telling my LB to two foot their DMC really. I don’t want to pull my team’s shape apart for the sake of this.
And why not more than one player? Because at the moment I want to see if it works. If it works then I’ll be ramping it up. My exceptions to this are if it’s a former player that left my club or if a player is carrying an injury. In both those cases they will also get some extra attention. Because I’m petty.
Responses and Examples
What’s the process then? It’s nice and simple, and I don’t want to take any credit for it. The use of opposition instructions is common across the FM community. FMStag has done great work in the past targetting players based on key passes for example, and was one of the first I saw do it in depth. I’ve done similar targeting players before too. I’m not reinventing the wheel by any means, but here’s how I’m doing it.
Match View at 30 mins
I’m using the fantastic statman skin, and I’m using a panel for the opposition passing stats, as well as a panel with their formation so I can better understand who is playing where. In this case we are at 30 minutes and there’s no clear threat in terms of key passes alone. I’d be tempted here to wait until half time.
Player Analytics at 45 mins
If you get to half time you can use the player analytics tab before the team talk to check in on potential targets.
As mentioned above I’m focused on key passes as I want to stop strings being pulled but if we have multiple players on the same number of key stats I’ll take into consideration dribbles made, CCC’s and pass completion. I’ll also take into account match condition, and single out weaker/tired players.
In the half time example above Connor has tied with two other players on key passes but has more dribbles, a high(ish) average rating and low condition. So he becomes the victim. Being selected at half time means I’m allowing no more key passes from him or just one additional dribble if I’m to consider him stopped.
Once you have your victim then roll on the opposition instructions. I’ve ticked a few options here so you can see but in reality I’m only going to use tackle harder and trigger press.
Results of the Reducer
For 20 glorious games in the North Counties East Premier League and NCEL Cup I have tried this approach. Route One Rovers have been promoted with me twice back to back, so we were in a league that is tier 9 and still fairly amateur. Non and part-time contracts are around, but a lot of the teams are very close ability wise.
All matches were 3 subs, and all changes were made at the 30 or 45 minute mark. With most taking place at the 45.
These are the raw results by fixture. If nothing jumps out at you and you want to avoid any mental arithmatic take a look below.
I’ve summarised the key points below, but please keep in a mind a few things. These are just observations based on a small sample size at lower league level. It might not have the same effect elsewhere. However, the results are quite striking, which is why I’ve not continued with an even larger sample size. It doesn’t look like a borderline or small effect. It seems fairly consistent.
The Stopping Power of Reducers
It is fairly clear. In 90% of cases the player targetted for our reducer treatment was ‘stopped’. Again by our criteria that means they weren’t able to maintain their performance at the same rate as pre-reducer. There are many reasons why that might be but the consistency of the effect seems to suggest it was our doing.
As a stat that strongly suggests there’s value in taking the 30 seconds to pick a target and set the trigger and tackling. I’m pleased with that. 9 times out of 10 a player with a reducer was not able to maintain their pre-reducer performance.
Escaping a Reducer
You can’t outrun the violence but you can be subbed. 45% of our victims were subbed. Interestingly none of them were actually injured by the reducers so these subs must have been tactical, performance or condition related. As a percentage that is much lower than the percentage stopped but what we really want to compare it against is the chance of being subbed anyway. With our reducer were our targets being subbed more or less often than you would expect by chance?
We’ve assumed keepers aren’t being subbed, and in all competitions there were 3 subs. If all subs were to be used then each player by chance alone might expect to be subbed in 30% of cases. With this in mind we should have seen 5-6 players out of 20 subbed.
We actually saw 9 subbed, 15% more than we would expect by chance alone. Our if you want to play with the numbers more that’s 50% increase compared to the base rate. Again strongly suggesting what we were doing wasn’t just impact the performance of the player but forcing the opposing manager to make changes to accommodate what we were doing. It sounds odd but it’s one of the more obvious and direct examples of the AI reacting to a tactical change you can make. I think sometimes FM can leave you wondering if what you are doing has any impact at all. Clearly it is here.
Did that translate into a change in average rating? Not really. The average change pre to post reducer in the match rating was 0.09. That’s positive. So if anything the average rating was going up. If you look at table 1 in the raw results you can see one match in particular where a player went on a spree and was not stopped, getting an insane 1.8 boost. That has skewed the results in a sample that size a little. Typically there was little shift in the average rating, up or down. Without that one match taken into account the average shifts to be a small negative change but nothing to shout home about.
I think that reflects how bad the match and average ratings are. They don’t pick up on the bread and butter of playing and instead represent values that are skewed by memorable ‘big’ events. For me it’s just another reason to ignore it as an in game stat.
Will the leg breaking continue?
Without a doubt. Machiavelli was right. Brutality is holy and the oppositions abilities and opportunities have been wasted. We are flying high with Route One Rovers, with playoffs looking likely despite our tiny budget and limited players. I think we owe that in part to putting the fear of God into at least one opposing player a game.
It’s also really fun.