perpetua

One issue that I see with this argument. A newcomer to the game is going to struggle regardless because the game has a pretty big learning curve. However a relatively less sophisticated AI reduces the level of interest of more seasoned FM players if they don't feel challenged. The sweet spot seems to be for those in between, those who have gotten past the learning curve but are yet to decipher the AI's patterns and can experience slight overperformance peppered with occasional disappointment to keep them interested in the game. Those in the sweet spot will eventually move into the seasoned FM player stage and likely fail to find a reason to keep playing. So for the sake of continuity, getting newcomers through the learning curve and into the sweet spot, then keeping them in the sweet spot as long as possible should probably be the objective.

To an extent this is as expected. If you are a better passer, you can attempt more difficult passes and a higher chance of having the pass intercepted. So are you concluding that with no tactical changes or changes to the role that a player is given, a player with a higher passing rating, holding everything else constant, attempts more risky passes? If so, I take this as a big win for the player AI and the ME.

Interesting conclusion. Any observations as to why this occurs? I wonder if it's related to more rigorous defending of the more well rounded player by the opponent? Or perhaps the single tool player (ie. dribble only), being more likely to use their only tool makes the player more effective or the defending less effective.

To the best of my knowledge (which is very limited) the underlying values assigned to each attribute level is linearly related to the attribute level. Using a non-linear approach is interesting but I would question how this would impact lower leagues. Someone with 18 for an attribute will be X better than someone with 15 for the same attribute but someone with 12 for an attribute will only be (X - y) better than someone with a 9 for an attribute. Let's say we want to achieve a league table with a 50 point difference between the top team and the bottom team. Without a change to the cost of attributes in CA (a whole other can of worms in my opinion), the expected CA difference between the top and bottom teams will be less at the top of the football pyramid than one step below, even less than two steps below etc. Research would have to adapt to this change as well. Without relying on CA differences being linear, this will leave us to shoot in the dark for lower levels and giving us very random looking results in the game until we get a handle on the impacts of the change.

The reason I used scale is because I try to think of each attribute as a questionnaire with a likert scale. But your point is well taken. To the best of my knowledge level of attributes are more granular than 1-20 once the game starts, so attributes researchers award are basically starting points. The player attribute progress screen displays this in a little more detail as player attribute graphs start to slightly go up or down over time without the attribute necessarily changing in on the player profile screen. I don't know which of these the match engine uses.

I guess it's a matter of choice of what kind of a game is desired. Reduce the distance between attribute levels and you have everyone travel at approximately the same maximum speed, thus quickly making speed less relevant. Then how do you deal with lower speed differences, leading to fewer attacks/shots on goal, leading to less scoring in the game? Make passes more accurate, make crosses more accurate, make shots on goal more accurate. Will this be done by increasing distance or by shifting the mean? If you shift the mean, what happens to lower leagues? If you increase the distance, won't strikers end up scoring too many goals compared to real life? And how will defending end up looking? No easy answers.

As others have alluded to in the thread, there is a lack of fast players in lower levels. We could flip this observation around and suggest there is a lack of slow players in higher levels. By extension, this means higher level players are not as mentally/technically strong as they perhaps should be while the lack of fast players in lower levels means there is sometimes too much technical/mental ability in lower levels. So the database itself separates higher and lower level players by speed. https://pubmed.ncbi.nlm.nih.gov/33261014/ Del Coso et. al. (2020) showed that teams' running speed did not impact final position in the Spanish League. What did impact final position, however, is the frequency of these runs (Baskaya et. al. 2023). So stamina and work rate impact performance but not necessarily maximum pace. When we look at Bundesliga and Bundesliga 2 we see (https://www.bundesliga.com/en/bundesliga/stats/players/top-speed, https://www.bundesliga.com/en/2bundesliga/stats/players/top-speed): Fastest Player observed in Bundesliga: 36.41 km/h Fastest Player observed in Bundesliga 2: 36.58 km/h 10th Fastest Player observed in Bundesliga: 35.96 km/h 10th Fastest Player observed in Bundesliga 2: 35.93 km/h 25th Fastest Player observed in Bundesliga: 35.28 km/h 25th Fastest Player observed in Bundesliga 2: 35.30 km/h 50th Fastest Player observed in Bundesliga: 34.75 km/h 50th Fastest Player observed in Bundesliga2: 34.71 km/h 75th Fastest Player observed in Bundesliga: 34.38 km/h 75th Fastest Player observed in Bundesliga2: 34.24 km/h 100th Fastest Player observed in Bundesliga: 34.08 km/h 100th Fastest Player observed in Bundesliga2: 33.96 km/h 150th Fastest Player observed in Bundesliga: 33.56 km/h 150th Fastest Player observed in Bundesliga2: 33.51 km/h 200th Fastest Player observed in Bundesliga: 33.15 km/h 200th Fastest Player observed in Bundesliga2: 33.03 km/h 250th Fastest Player observed in Bundesliga: 32.78 km/h 250th Fastest Player observed in Bundesliga2: 32.65 km/h 300th Fastest Player observed in Bundesliga: 32.33 km/h 300th Fastest Player observed in Bundesliga2: 32.20 km/h So there doesn't appear to be an important difference in speed between the Bundesliga and Bundesliga 2. Is there a difference between Bundesliga 2 and lower divisions? It would be advisable to make comparisons between the top levels of many nations to get a better handle on this. I conjecture that it's not player/nation level but rather nation sports culture that impacts top speed. Anecdotally, what also appears to impact top speed is player position. Central midfielders and central defenders tend to be the slowest players, followed by strikers and central attacking midfielders; and wingers and full backs being the fastest players on average. This is well documented in academic articles on the topic. Since in database research we tend to award higher speed to higher level players and lower speed to lower level players, when the match engine is balanced to provide realistic results conditional on database attributes, the correlation between player speed and team quality is what ends up getting reflected. Certainly the match engine has improved over the years and anecdotally the effect of speed isn't as prominent as it used to be. However this database practice hinders the development of the ME and the manager AI to deal with scenarios such as those in the video in my opinion. At the end of the day, the match engine has to provide realistic results out of the box. But its development is limited by the data it uses and can't just take the position that the ME is correct but the data is flawed. It's also worth mentioning that technical attributes represent the mean/average of a distribution. So someone with a finishing of 10 attempting the same shot 100 times will score a goal 20 times on average (numbers made up for illustration), while someone with a finishing of 11 attempting the same shot will score more than 20 times or someone with 9 will score fewer than 20 times. Same with decisions. Someone with a 10 will make the correct decision 70 times out of 100, someone with 11 will make the correct decision more often. Same with work rate. On the other hand Pace is defined as the highest running speed of a player. Not an average like other attributes but a maximum. So I don't believe we should treat (most) physical attributes as being dependent on playing level but rather base it on player physiology and nation culture with less emphasis on playing level. The impediment to this is the fact that we are often too timid when awarding mental and technical attributes for top players. For example a central defender in his late 30s with more than 700 top level and international matches getting 15 for decisions and positioning but 13 for pace trickles down to a a central defender in his mid 20s getting 13 for decisions and 15 for pace and the lower CA player in the same position getting 12 and 14 or 11 and 13. This makes players too similar to each other while strengths and weaknesses aren't reflected in the 1-20 scale. So the AI manager or the ME doesn't have to be developed to deal with scenarios which are not like this. So not seeing the AI assemble such squads or assuming that humans would not build such squads (as you put it) is a weakness. On the other hand, the players/researchers are used to a certain way attributes have been rated for years and a change to this is always going to be painful for all involved Attempt to fix the issue highlighted in the video only and realism of results in the game will likely to suffer. Don't fix the issue highlighted in the video and you have (at best a short term) exploit. No easy answers.

Having played this game for far too long, the results in the video aren't really all that surprising to me. It's a low CA team, playing against stronger opponents. The stronger opponents under AI control are more likely to come out with a more attacking setup as a result. This gives the pacy, low CA team exactly the scenario that they need to succeed, a team that leaves a lot of space to exploit and lots of opportunities to play 1v1 against slower players. If the AI manager would look at the type of team they are facing (ie. a bunch of very pacy but unskilled players) instead of CA when devising the strategy, it would opt for a different strategy. At the very least that's what I would do. So this is a joint test of not only attributes but also the AI's ability to nullify the very real threat of pace by deploying tactics designed to do just that. A more salient test would be to play human vs. human against this pacy team and discover the conditions under which the pacy team thrives or suffers in order to pinpoint the underlying issue. Play a more risky, more high tempo game against the pacier team and you're playing to their hand. Play a more cautious style that forces them to use their weak skills more frequently and you're playing to their weakness. The real question is, does the AI manager know to do that?

@sidbets Yes. These were reflected in the last version of FM 2024 that was released.

Yildiz is a second name entered by the German and Norwegian research teams to be used for newgens. I suggest raising the issue with them. I suspect your physio has German or Norwegian nationality. All Turkish newgen names should be using the correct lettering.

Regarding the first issue, we deal with the incorrect letters in names if and when players play in Turkey. However there are a lot of players with Turkish names playing abroad and we don't necessarily have concrete information as to how their names should be. A player may indeed have the name Kilic or Kiliç or Kılıc or Kılıç. It's probably best not to make an assumption until we see him appear in the Turkish FA website. For example Samet Akaydin's name is spelled this way but if I were to make an assumption, I would have assumed (incorrectly) his last name would be Akaydın. Regarding the second issue. Turkish players are typically known by their first name. It's very rare for Turkish players to be known by their last name. If a player happens to play at a club where more than one player shares the same name then they are known by their first and second name. If we added the first name as common name, then we could potentially end up with 3-4 players with the same name at the same club and it would be impossible to distinguish them from each other on the tactics board or in the match engine. So SI have coded for us a special rule where players' first name is shown unless the club happens to have two players with the same first name, in which case both players are shown by their first and second name.

Thanks. Already reported and replied in the rhread above.

Your statement is predicated upon the assumption that these players are rated correctly. 40 year old Pepe's pace, for example, is 13. Based on your argument, Szymanski should be rated higher. Perhaps 14 or 15. If that's the case, where do we put a truly fast players like Yusuf Ozdemir or Baris Alper Yilmaz or Osayi-Samuel? 17-18-19? Then what about players who are even faster? In my opinion, there is an irrational desire not to rate any player with less than 10 for speed related attributes. So who should be slow in this database? Nobody? If everyone is fast, then speed isn't a factor that makes a difference between teams, making speed an irrelevant attribute. Suppose every player in the game had tackling of 10 or higher. Then every player can defend well. Even worse is if only very few players had tackling of 15 or higher. Then the entire population of footballers, regardless of position, are stuck in a 6 point range for tackling. Everyone can tackle at a similar level, making it meaningless to have someone who can tackle. Same goes for finishing, first touch, passing, decisions etc. etc. etc. The game guide/manual has always indicated that even an attribute level of 1 is meant to represent a professional player who is weakest in this attribute. Key word here is professional. Not someone playing among friends in a five a side game on the weekends. So that player who has a 1 rating is meant to be miles better than the average person. It's the same with pace. Someone with pace = 1 is a slow professional player but is still an athlete who trains every single day. 10.5 is supposed to be about average speed while 20 is supposed to be extraordinary, once in a generation type of speed. So it makes no sense whatsoever to restrict attribute ratings to the 10-20 range like your argument suggests. It's not a formula. We are able to observe players' top speed as they display it in matches. Not every player displays their top speed in matches, in fact they show their top speed perhaps once a season. Bundesliga publish these statistics and can be seen by anyone. Feel free to peruse those lists and see the differences between players. After taking a look at these, I would suggest the following thought exercise. Is there any logical or empirical reason for us to believe that the slowest Bundesliga players should be faster than approximately 50% of the world's professional footballer population (assuming that speed is normally distributed - I would actually suggest that the distribution of attributes is half-normal, meaning that in theory there should be far more players with a 1 rating than a 20 rating). So there is the challenge. How do you prove that the slowest Bundesliga player is faster than half of all professional footballers. Or if I am right and there are far more players in the world with a 1 rating than players with a 20 rating, then really how much better is Bonucci's 11 from the average player worldwide? Here is a hint, Bundesliga 2 also publish players' top speeds. You'll notice that those are not very different than Bundesliga. That is, despite the decline in playing level the distribution of players' top speed doesn't change. There are plenty of players who have a top speed of 36 km/h in Bundesliga 2, just like Bundesliga. And the lowest outfield players run at approximately 30 km/h. Go down another level and perhaps you'll see a slight change in the distribution but not as drastic as what we see in the game database. So the assumption that Bonucci is faster than at least 50% of the footballer population that the Pace rating of 11 implies doesn't appear to have support in this instance. So why do we have this assumption in the database? I do not know.

I appreciate your comment. What he has been at Fenerbahce is a second striker who takes good advantage of the space created for him by Dzeko. As a result, he has scored quite a few nice goals. Perhaps the Fenerbahce researcher is a bit optimistic on him but I don't really think he's too far off in terms of overall ability (ie. CA). Attributes can always be rated differently through different eyes.

This really isn't a shocker. You have an underpowered (ie. low CA) team full of players who are excellent athletes that can anticipate and have great in match consistency (ie. concentration). The underpowered team is going to be expected to lose most matches, so they will likely face teams that are taking a lot of risks against them. Fortunately the physical ability of the players at their disposal along with dribbling are very suitable to succeed against teams taking risks and leaving a lot of space undefended. On the other hand you have a powerful (ie. high CA) team full of players who are mediocre athletes. The strong team is going to be expected to win most matches, so they will likely come out attacking and taking a lot of risks. In the meanwhile, their opponents are going to consistently park the bus. The lack of athleticism (across the entire team rather than some players) as well as anticipation and concentration (in addition to dribbling) is going to make it difficult for the strong team to break down their opponents. One of the most effective methods of forcing teams parking the bus to open up is goals from set pieces. However the lack of size (jumping/strength) will be an issue for set pieces, further weakened by relatively low anticipation. So they will be very likely to drop points against weaker clubs who will sit back deep and defend against them since the lack of athleticism and anticipation will also leave them vulnerable on the counter. You don't need a whole team of big/strong players for set pieces but a few will certainly help. None means this isn't an option, further handicapping the stronger team. What we would see, in real life, is that opponents would sit deep and force the athletes to play the ball and consistently lose possession. Simply because the team isn't built for that style of play. As for the powerful team, they would recognize that athleticism isn't their strong point but they can keep the ball and not give it away easily. The opposing team, recognizing that the stronger team isn't a threat to beat them with speed, will likely try to pressure high in order to win the ball. So the strong team would play a very patient game, dominating possession to catch the opponent napping and/or exhausting them after a prolonged press. So this experiment is a joint test of AI management sophistication as well as attributes. The fact that the OP went on holiday and delegated all responsibilities to staff is the flaw that, at least partially, contributes to the results in my opinion. In my experience the AI managers are still quite unsophisticated in FM. So any result comparison oriented test leaving the tactical choices to an AI manager will be flawed unless tactical choices were restricted in a way to help these clubs and opponents play as they would in real life. Blaming and calling for attribute rebalancing alone will likely break the game further, leading to more unexpected results for normal teams while trying to fix the fringes. History doesn't repeat but it rhymes. Similar discussions were going on here 20 years ago. As the game evolved, tactical flexibility for the user was restricted in order to minimize odd results arising from users taking tactical choices to extremes. Now it's player attributes. At the very least, we won't see these kinds of player attributes very often. The issue, in my mind, has always been AI manager not having sufficient sophistication to counter a human player and AI tactics being a little too scripted. This game is at its weakest when you observe what your (AI) opponent does and do the exact things to counter it. The above test has done not much more than show this with the fringe squad makeup without realizing. We are in the age of AI now so there is certainly hope for the improvement of this aspect of the game. Just my 2c.