Onderwerp: Sokker Asistente

Leave them in blank, it will get the correct ones

is it possible to change the yellows and red cards for the NT team showing?
Now it is a problem as it shows the situation of the official games.
I understand why it is so, because sokker also shows it wrong in squad section, but it shows correct when you click on individual player.
It is really confusing now, maybe you can correct the sokker mistake:))

Yes, fixed

I'm just setting this up, but I just don't see how it estimates talent? it's all defaulting to 3.5.

https://imgur.com/a/X1fJCEj]
(gewijzigd)

It needs some weeks of data. It can't know what did you train last weeks or what skills did the players pop

Oh, you mean it can't read previous data... it's only good for data going forward?

Okay

You can manually add previous weeks if you have the data in other assistant, for example. But it can't read it from Sokker

It can't know what did you train last weeks or what skills did the players pop

You could use https://sokker.org/api/training/<player-id>/report to get training history for each player ;) Only thing missing would be coaches, but you could just use the current ones and tell user to fix if it's wrong for the past :P

the good news is, my coaching has barely changed at all since I started. The assistants have been upgraded slightly, but my HC not at all.

I'm down for manually adding the previous season, just have to figure out how.

Cool, It has all what I need :)

I can only see 7 weeks of training, but I guess it's because I changed all my team 7 weeks ago. How many weeks of training can you see?

@islander hmmm It seems you need to wait 1 week until you have some data. Then a "+" sign will appear in the list of trainings that will let you add weeks. But maybe I can adapt It to the new api before the next Thursday

I can only see 7 weeks of training, but I guess it's because I changed all my team 7 weeks ago. How many weeks of training can you see?

Everything :P The data from the table in the "Graphs" tab comes from this api call.

For example the end of that table for a player I bought in my first week:


Here is full json with his reports

I think you should only use data for the last 33 weeks (so only "week":994 and higher), earlier ones don't have info on the time played (as it was before the training reform):

Maybe even since week 1007. Previous weeks were affected by Rauls touch (slowing down).

Maybe even since week 1007. Previous weeks were affected by Rauls touch (slowing down).

I think the best option would be to have training settings divided into 3 parts:
- before the training reform (100% normal training, 0.10 general training factor)
- broken times after the training reform (~85% normal training, 0.10 general training factor)
- good times after the training reform (100% normal training, 0.15 general training factor)
But I know it might not be worth it to fix the problem that will slowly fade away with time :P

Currently talent calculations are broken for players I have for a very long time, as it applies that general training factor to older times.

Another thing - I can confirm what Raul said about pace training. that it slows down much quicker than training in other skills, for example this 22yo player:

You can see that his today's playmaker and passing pops match the talent I gave him (that was confirmed to be about 3.2-3.3 a long time ago), but in pace he needed much more training for incredible skill than he should (if it was the same 1,094 / year multiplier for each skill).

Btw. if I delete his talent it shows something stupid like "4.89 / 3.14", the "min" number is clearly being impacted by very old pops from general training with the current setting.

Everything depends on whether or not user have PLUS bought. For non-premium users game keeps track of last 10 trainings only (as it visible on graphs) for premium users longer history is kept (and more info can be read back).

For a player I have for at least 3 season I can only check last 10 trainings ;)

First of all, thank you for this awesome app Terrion :D

I was wondering if you have any place, or a piece of code which shows how the sum of training points and estimation of training points are calculated? I was looking at your github page but I'm not sure if it's up to date. If you could give me a hint where to search I could probably figure it out from .jsp files from the sokker.war that's available on your website ;p

Yes, the source code of GitHub is outdated. Maybe I'll work on cleaning It of passwords to be able to use it directly so that It is always up to date, but for the moment my server can't use the variables in the server.xml

It's an easy idea but a conplex explanation. I'll try to simplify it:

To begin, we need to give a constant talent value to a base case. The rest of cases just would need to apply some maths to this base case. I used a pop to weak pace in a 16 yo player as base case, because I had a junior with talent 4 that popped to it in exactly 180 training points, IIRC

In fact it wasn't "exact" because it popped the previous level with a general training, so there was about 10 training points of error margin. But I was happy with such a little error margin

The problem is you don't usually have a so little error margin, so on general you have a bigger margin which gives you a range of talents: min and max

How do I apply the formula to different ages/levels? Just multiply it by the age/level factor. For example: a 17 yo player would need 180 * 0.1 = 198 training points to keep a talent of value 4. The same applies to a 16yo player who pops to regular instead of weak

What happens with other skills? More or less the same. In the beginning I added two levels to calculate It for other skills compared to pace, since that was the approximate difference between them. Now I use the different skills factors to make the comparison

Now that we can compare everything we can also use several consecutive pops to reduce the error margin. For example, if you add the training points needed to get two pops and divide It by 2, you get the average training points needed, but you also have the average error margin, which usually will be less the more consecutive skills you add

Of course this is an approximation. We would probably need some exponential function to get a better approach, but that is all what I could do :P
(gewijzigd)