I was watching EBFM's Training schedule tests from last year and although he has a lot of great data to look through, I was curious if using the metric of number of attributes that have grew .5 is the best way to decide the best training sessions and CA for training schedule. This led me to 3 thoughts
1. Some Attributes aren't as equally desired as others but are given equal weight in the excel findings
For example in this machine learning post. Long shots aren't desirable unless you're a DM lurking outside the box or a ST who sometimes takes 1v1 shots from those locations. None of these weightings are above 50 in the machine learning post.
Some other examples are Leadership, Flair, and Natural Fitness are where you don't really need them for them to play well on the squad but they are since Match Practice gives them a .5 boost they are counted towards best training session.
For the measure of CA, in this chart Decisions are very highly weighted but in both machine learning and fm-arena tests, they are very overweighted compared to the important attributes. Attacking, Defending, and Match Practice trainings are very good for decisions, so when they are combined together they could grow decisions to a point where it's really good for CA, but the CA is being used inefficiently.
2. Some attributes you want more of compared to others
Obviously Pace and Acceleration are king as well as for some Balance, Stamina, and Jumping Reach. Even though the Physical Training overall is the best for all Physical growth, Quickness trains Pace and Acceleration quicker and Endurance grows Work Rate, Stamina and Strength quicker. He doesn't fill in all the slots in his training regime so maybe with a more rigorous regime you add these more specialized trainings.
3. Training Units: Can they be optimized for better efficiency? If you go into Training >> Units, you'll see that your squad is split into 3 different units. Does putting some Defensive players into the Attacking Unit for a session help them progress attacking attributes like Crossing and Dribbling? In his spreadsheet, Trainings like Attacking Direct and Chance Conversion are extremely high in technical attribute growth up but are under things like Attacking training, however while the Attacking training is equal in progress for all players, those in the defensive unit get 20% focus vs the attacking sides' 60%. If attacking players such as the DM and FB were moved to the attacking unit before the testing started, would the results be better for those respective training sessions?
TL;DR The general point is that, I believe the training schedule and sessions that focus on the growth of the important attributes that fm-arena/chinese machine learning experts have found should be aim of the testing. Maybe using more specialized trainings or optimizing rest days or training units would help bring our players to their peaks better than before.
Was wondering if my thought process is correct and if there are ways to really juice training up?
In one of his other experiments he has shown a connection between units and development, to the extent where dumping every player into one unit and focusing entirely on one type of training schedule should in theory lead to more development.
Surely there are ways to optimize for specific kinds of attribute growth. Every schedule distribution will come with tradeoffs, so what those optimizations will look like depend on what you want to get from your players long-term.
I think EBTS is a great starting point for overall well-rounded players. If you want more speed demons, are willing to accept more injuries/exhaustion that you'd compensate with better rotation, etc. then focusing more on physical training rather than general training would provide better results.
We could do more testing to aim towards idealized stat weightings rather than even weight. For the meta tactics, speed and technique seem more important than mentals, so steering towards those would be a good start.
There is also the question of whether the training schedules work exactly the same in FM24 as they did in FM23. There was some difference between 23 and 22...
Also, just to note, long shots may not be as desirable, but they develop really well regardless of what you train, so there's not much point optimizing away from them I think.
TL;DR below
I was watching EBFM's Training schedule tests from last year and although he has a lot of great data to look through, I was curious if using the metric of number of attributes that have grew .5 is the best way to decide the best training sessions and CA for training schedule. This led me to 3 thoughts
1. Some Attributes aren't as equally desired as others but are given equal weight in the excel findings
For example in this machine learning post.
Long shots aren't desirable unless you're a DM lurking outside the box or a ST who sometimes takes 1v1 shots from those locations. None of these weightings are above 50 in the machine learning post.
Some other examples are Leadership, Flair, and Natural Fitness are where you don't really need them for them to play well on the squad but they are since Match Practice gives them a .5 boost they are counted towards best training session.
For the measure of CA, in this chart Decisions are very highly weighted but in both machine learning and fm-arena tests, they are very overweighted compared to the important attributes. Attacking, Defending, and Match Practice trainings are very good for decisions, so when they are combined together they could grow decisions to a point where it's really good for CA, but the CA is being used inefficiently.
2. Some attributes you want more of compared to others
Obviously Pace and Acceleration are king as well as for some Balance, Stamina, and Jumping Reach.
Even though the Physical Training overall is the best for all Physical growth, Quickness trains Pace and Acceleration quicker and Endurance grows Work Rate, Stamina and Strength quicker. He doesn't fill in all the slots in his training regime so maybe with a more rigorous regime you add these more specialized trainings.
3. Training Units: Can they be optimized for better efficiency?
If you go into Training >> Units, you'll see that your squad is split into 3 different units. Does putting some Defensive players into the Attacking Unit for a session help them progress attacking attributes like Crossing and Dribbling? In his spreadsheet, Trainings like Attacking Direct and Chance Conversion are extremely high in technical attribute growth up but are under things like Attacking training, however while the Attacking training is equal in progress for all players, those in the defensive unit get 20% focus vs the attacking sides' 60%. If attacking players such as the DM and FB were moved to the attacking unit before the testing started, would the results be better for those respective training sessions?
TL;DR The general point is that, I believe the training schedule and sessions that focus on the growth of the important attributes that fm-arena/chinese machine learning experts have found should be aim of the testing. Maybe using more specialized trainings or optimizing rest days or training units would help bring our players to their peaks better than before.
Was wondering if my thought process is correct and if there are ways to really juice training up?
Your thoughts are correct.
In one of his other experiments he has shown a connection between units and development, to the extent where dumping every player into one unit and focusing entirely on one type of training schedule should in theory lead to more development.
Surely there are ways to optimize for specific kinds of attribute growth. Every schedule distribution will come with tradeoffs, so what those optimizations will look like depend on what you want to get from your players long-term.
I think EBTS is a great starting point for overall well-rounded players. If you want more speed demons, are willing to accept more injuries/exhaustion that you'd compensate with better rotation, etc. then focusing more on physical training rather than general training would provide better results.
We could do more testing to aim towards idealized stat weightings rather than even weight. For the meta tactics, speed and technique seem more important than mentals, so steering towards those would be a good start.
There is also the question of whether the training schedules work exactly the same in FM24 as they did in FM23. There was some difference between 23 and 22...
Also, just to note, long shots may not be as desirable, but they develop really well regardless of what you train, so there's not much point optimizing away from them I think.
I'm really interested in this, and might start playing around a little bit myself with Max's findings and calculations.
At the moment, I'm really only interested in trying to max out Pace and Acceleration for as many players as possible.
There's got to be an even better way to min/max this.
Lol, sometimes I think the metagame behind the game is more interesting than the game itself...