Jack Joyce

Don't get much time to respond to things on the forums recently - but just wanted to say this is a really interesting discussion so thanks for all your thoughts on the CB roles. It's great food for thought.

From what I remember it factors into the speed at which they're able to identify the roles the opposition are playing. So you'll be left in the dark for longer if your staff are worse.

AI managers can use the Libero role, but currently only if it's set as a preferred role by the researcher. This is essentially how it's designed yeah, your staff are watching the match and feeding back to your on what roles the opposition are using. We believe info like this is really important because where FM differs to real life quite significantly is, in FM you're watching highlights. If you're on key/extended highlights then you don't see the full picture like a real manager would, so you need a bit more information to fill in the blanks.

The partnership lines are reflective of the player's past performances when used together, and don't impact performance directly. So a good link just means that the two players tend to perform well when used in a partnership (get good ratings).

No worries! This clip from Luke Williams was really interesting and about a similar topic https://x.com/LowerTiers/status/1727315641747886161?s=20

This is actually normal and how most clubs and data companies track corner goals - a corner goal isn't just a goal scored directly from the taker's kick, it's a goal from the corner phase of play. Otherwise no short corner or 2nd phase play would ever be considered a corner goal.

If it's just those two leagues then it's likely because there's something specific about teams/players in those leagues that makes them different to other leagues. We'd need to identify that really to know what to change, but it's likely a research/data thing causing it.

The ME uses the full range of attributes, which can vary over time depending on tons of factors such as consistency, big matches, morale/body language etc. What this means is that your 18 finishing striker won't always be 18 finishing 100% of the time, for example if they're inconsistent they may sometimes be playing like a 13 finishing striker for portions of the match. In terms of playing styles though, there's nothing stopping you from playing any style really based on level of play. The football you're playing is relative and that's important to remember, for example Notts County in the National League last season played a very Pep-esque possession system that received a lot of plaudits, and Barrow did before them at the same level as well. Now, that doesn't mean they're as good as Man City! But relative to the opposition they're playing, they were extremely capable of keeping the ball and building attacks from back-to-front.

The changelog tends to just be a summary of the key changes, but there's nothing major left out of the list. There's often fixes for knock ons to other fixes that hadn't been public yet etc. That we don't list or just combine into a single line. We didn't make any changes to attacking movement, most of the changes were to do with set pieces tbh. The only change between beta and the next ME was reducing injuries to GKs.

I have no idea sorry, I'm all about the Match Engine so when it comes to this sort of thing I know as much as you do. Best to report it as a bug in the bugs forum and someone will let you know.

It's certainly unique but we feel it creates the best metric - if we based our xG numbers on real life finishing then it's never going to be as accurate within an FM world as when trained on FM data, even though we obviously always strive to make our finishing as close to real life as possible anyway, there'll always be slight differences. But it also gives us a lot of added benefits when covering leagues where data coverage isn't the best, for example in lower leagues or in womens football it's very difficult or nigh on impossible to get the level of real-life data our xG models require. But when training using FM data we can play all the matches we like, as many times as we like with all the data that we could possibly ask for.

We run a ton of simulated matches via an automated tool which updates our xG data. We then push that data into the next update so all your xG will update immediately for the next match you play. Old matches played on an older ME will use the old xG data still, any new matches will use the new xG data. What it means though is that if finishing was better in a new engine, the xG would be higher per shot to reflect that, so you can't really use xG to identify better or worse finishing from a game balance POV since it's baked into the data in the first place.

You can believe what you like, but I'd suggest that you need a lot more than one or two matches to draw any sort of conclusion over there being a lack of goals in the new engine or not. You don't have to go back very far to see real-life matches with these exact sort of situations - PSG v Newcastle only a couple of days ago had an xG of 4.54 for PSG, and they needed a very questionable penalty to even score once. If a team never performed under their xG then football would be very boring!

Just so people are aware, FM's xG model is trained on FM match data from the latest engine. So the xG stat will adapt to the finishing of the current engine automatically and is not likely to be representative of any changes to finishing or chance conversion between versions.

Yes, goals are lower following the ME changes as a result of defensive improvements and tweaks to heading accuracy when players are being challenged directly by a defender. For example in the Premier League, total goals per game went from around 3.1 per game to about 2.84 which should hopefully be a pretty good balance.