Guide: Understanding the 2016 Training System

[Al Svanberg]

Important Message

In the first table I posted there was an error causing me to speculate on young players gaining Fitness ability faster. That was an error caused by a wrong entry in the excel spreadsheet I had created which has been corrected below.The numbers no longer indicates that young players improve their fitness ability faster than any other ability.

I have lately done a comprehensive study of training sessions I have made and I think the results I have seen have been interesting so I thought I put them in a guide on the topic of how the training system works. I have definitely not found and groundbreaking stuff, but I have maybe realized a few things I thought was pretty unclear before.

As usual, ANY input, questions and what not is greatly appreciated!

1. HOW DOES A PLAYER INCREASE HIS ABILITY BY TRAINING?

1.1 Background

I have done the study by running a training session 21 times and very thoroughly documenting the results of it.

The training session I have run is the following:


The practices used has the following level:
P-G-S Lvl 4
FCA Lvl 4
PTP Lvl 3
SD Lvl 2
Stretch Lvl 3

The fast trainer I used when doing the study was 18 y/o and had 72% when I started it. He is a DMR/MR/AMR so all abilities practiced are highlighted (I wanted to have clean numbers).

1.2 The result of the training session

From the 21 training sessions I run I got the following results:


I hope the table is fairly self-explanatory. But for example, down to the left I have circled the number 6. This is the total gain from session 1, ie 6%. In the upper right I have circled the number 7, this is the total number of % gain in the ability Conditioning from all 21 sessions. In the low right I have circled the number 80, this is the total gain in % all categories from all 21 sessions.

1.3 Why did the training sessions I ran give this result?

This table explains the results:


To explain how the table is constructed I have circled a few numbers to be used as examples. I will start from the left and go to the right when explaining them.

First, to the upper left I have circled the number 1.5. This is the condition loss for the practice Pass-Go-Shoot (for some reason I used the abb. PSR, it should be PGS). This is also the "factor" the game use when calculating the result of a practice. Below it I have underlined the number 130%, since my PGS is lvl 4, the increase from this practice is 30% higher than normal. Further down I have circled the number 0.65. This is the total gain from the practice PGS in the ability Passing. This number comes from the following formula: 1/3 (passing is one of three abilities trained) x 1.5 (the mentioned factor, ie 1.5% condition loss from the PGS practice) x 1,3 (30% increase from PGS being lvl 4) = 0.65. The number 1.95 is the total factor from the PGS drill.

Secondly, the next number from the left circled are 2.69 and 17.35%. This is the total number of increase from the 5 drills used in each practice session. In other words, when I run this practice as a starting point 17.35% of the gain will sort under the ability Passing. The numbers come from adding up the total number all drills add to passing (0.65+1.22+0.825=2.69) and 2.69 divided with the total number for all abilities which is 15.53 (this is basically the total conditioning loss from all drills increased with the bonus from each level for the respective drill).

Thirdly, the numbers circled are 13.88 and 14. 13.88 is what I call the expected gain. Its the gain in the ability Passing that basically is to be expected, at least before a few other factors are taken into account (see below). In this case it comes from taking the total gain, 80, times 17.52% (the number explained in the previous para), which gives you 14.02. The real gain I got in this case was 14 so the numbers added up so to speak.

As known before, there is a random factor involved in practicing and the numbers shown that. However, and I think this was pretty interesting, this random factor is a bit smaller than I expected. From courage down to tackling, almost all numbers are within +/- 10%. The bigger deviations are shown in the abilities that are trained very little, ie. shooting and strength, and are explained by the fact that 1% for them has a much bigger impact and you need a bigger sample size to calibrate them.

1.4 The Training-Efficiency Factor

The above table and numbers -- only --, after the fact, explains why these sessions gave the results I got from them. It would not on the basis of them be possible to beforehand predict the outcome of the session.

To able to do this you need to know -- what I call the -- the Training-Efficiency Factor for this player. In short, this is the pace a player will increase his abilities in relation to the condition loss of a practice/drill.

For the fast trainer I have used, the Training-Efficiency Factor is:


Ie the total gain of 80 divided with the total condition loss of 236. This means that the player trained, with the level of practices used, will gain 0.34% ability per each % lost in conditioning. For 15% worth, a Rest Pack, the gain will be 5.1%.

I have started to create a table with the top Training-Efficiency Factor (TEF) for each categories of fast trainers that I hope I can finnish:



2. FACTORS IMPACTING A TRAINING SESSION

2.1 Highlighted abilities

The below table shows the impact of highlighted abilities. But first some background information. This is the player trained, the so called grey or non-highlighted abilities are circled, crossing and finishing. A player is said to develop grey abilities slower than highlighted abilities-



I have used the following combination of drills, they cover all abilities except heading:



This were the result from 31 sessions:



As you can see, the grey abilities have both developed approximately 30% slower than expected. The sample size is still so small that I wouldn't bet on this number being correct to the decimal, but from running several extensive tests I would bet on the rule being that around 30-40% less gain will be made. Still I would probably have to use up towards 100-150 greens to get a good numbers and I only had a third of that at hand for this test.

So, what conclusion can we draw from this? Will a player gain less if he participates in a drill that train abilities that are grey for him? I have actually found that this is not the case, at least contrary to my believes before doing these tests. The grey abilities for my player Dailly -- Crossing and Finishing -- are covered only by the Fast Counter Attack drill in this practice. As the table show, the decrease in gain in Crossing and Finishing have instead almost to the decimal been attributed the other abilities covered by the Fast Counter Attack drill, ie Passing and Creativity.

This can be explained with the following example. Lets say we run a large number of Fast Counter Attack drills (this drill improves Creativity, Crossing, Finishing and Passing) on a AMR (Creativity, Crossing, Finishing and Passing are all highlighted on a AMR), we could get the following result (plus minus only a few percent):
Creativity +50%
Crossing +50%
Finishing +50%
Passing +50%

But if we instead runs the same amount of drills on a MC (Only Creativity and Passing are highlighted while Crossing and Finnsihing are grey). we should roughly get the following result:
Creativity +65%
Crossing +35%
Finishing +35%
Passing +65%

2.2 Mix of drills

I have found that the mix of drills does -- not -- impact the Training Efficiency Factor at all. As is well known, you get a higher bonus (Offense, Defense, Conditioning and Possession) from a well constructed practice. But the individual gain in ability is not impacted by the mix of drills in a session.

This is for example the result from 10 straight sessions with 6 x Press the Play drills. The differences between the gains where very marginal and occurred at exactly the same pace as the gain from a school-book mix of drills.

[LeManiaque]

Best post around. We think a lot alike and I'm excited to see how this develops.

[Rautz]

Great work Al, you're very meticulous and have a good mindset for experimenting and extracting information


(Just a small correction, for the condition the gain is not 44% higher, it is 144% higher, like your table says. +144% or *244%)

[Al Svanberg]

Thanks so much guys!

Great work Al, you're very meticulous and have a good mindset for experimenting and extracting information


(Just a small correction, for the condition the gain is not 44% higher, it is 144% higher, like your table says. +144% or *244%)

Thanks! I had a total brain block when I tried to figure that out late last night. Lol I still can't, something is odd with that number. You can take it straight off and get the actual gain. But all other positive numbers, you need to add 1 (ie 100%) to get the actual number, like normally is the case when you talk about something is as being x% higher than something else. Ie something that is 50% higher than 100 is 150% times 100 = 150. But the formula used in Excel should be the same for all numbers so I don't understand why it isn't the same for all numbers. Need to look into the spreadsheet and see what has happened.

[Rautz]

something is odd with that number. You can take it straight off and get the actual gain. But all other positive numbers, you need to add 1 (ie 100%) to get the actual number

I see no problem with the number.
2.86 + (144/100)*2.86 = 7
3.75 + (6.77/100)*3.75 = 4

The interesting part is where all the numbers behave exactly as expected over a small sample size. That may mean the gain is not random (or very small random), it's linear and it can be calculated precisely.
The other interesting part are the 2 stats that behave abnormally (Condition and Aggressivness) and if they can be manipulated for higher gains

[Al Svanberg]

I see no problem with the number.
2.86 + (144/100)*2.86 = 7
3.75 + (6.77/100)*3.75 = 4

The interesting part is where all the numbers behave exactly as expected over a small sample size. That may mean the gain is not random (or very small random), it's linear and it can be calculated precisely.
The other interesting part are the 2 stats that behave abnormally (Condition and Aggressivness) and if they can be manipulated for higher gains

Thanks!

Yeah exactly, that was the big thing I got out of this too, like its fairly random with a very small sample size, but it evens out over just 4-5 practices basically. Like you say, in essence you can predict the outcome.

I definitely know that I have read that younger kids improve conditioning faster. Like the Gym drill seem very interesting in this perspective. If you have a world class gym drill it should be amazing at adding %. A big question is, does the premium for Conditioning carry over into training a new position or special ability? If that is the case, a gym or long run drill or a combo of them could be very interesting.

BTW, does anyone know where the info that kids train conditioning faster comes from?

[Al Svanberg]

Here comes some more Data.

I ran the following drills 30 times with the same kid, but this time after he just became a 5 star player. Here are the data, please ask if anyone have on questions on the numbers:

1. Press the play: 2. Fast-Counter Attack: 3. Press the play: 4 Slalom Dribble: 5 Shooting: 6 Stretch







The TEF is down from 0.34 to 0.26 after he got to 5 stars. The diffs are smaller with a bigger sample size, but more conditioning and less aggressiveness is still a clear result.

[Rautz]

Thanks!

Yeah exactly, that was the big thing I got out of this too, like its fairly random with a very small sample size, but it evens out over just 4-5 practices basically. Like you say, in essence you can predict the outcome.

I definitely know that I have read that younger kids improve conditioning faster. Like the Gym drill seem very interesting in this perspective. If you have a world class gym drill it should be amazing at adding %. A big question is, does the premium for Conditioning carry over into training a new position or special ability? If that is the case, a gym or long run drill or a combo of them could be very interesting.

BTW, does anyone know where the info that kids train conditioning faster comes from?

On the first point, it seems random at first, because you don't know the initial state of the abilities, some could be at x.9, others at x.0. But over time (and it seems not even a long time) that tranlates to those tiny differences of 5-10%, and probably 0-1% over a bigger sample.

On the second point, it's time to upgrade our gym drill
It would be wonderful to add faster %, or even NP/SA, but if that's the case, once we find out it may be nerfed

It definitely seem like the mix of drills used does not impact the individual result, only the bonuses. Or what do you guys think?

My tests show the same thing, I'm confident that the mix doesn't matter, thus I always use my most upgraded one.

[Duncton]

Nice work.

[ibangali]

Nice work AI Svanberg.

Just a quick question please what is your training ground lvl.

I'm not sure but I notice when I lvl up my training ground players gaining little more than before, I'll try to check again when I lvl up my training ground again.