1.9.2 stages

[gyboth]

the usual topic telling you what's going on behind the scenes

so, what have i been up to the last days? michal (alefzero) kept complaining that the game is too easy, and he also told us why:

It seems that older (~28 years) players with very large skill and talent difference are too good to buy. They are cheap and they advance much too fast. I will suggest just not to generated such players (it seems improbably to improve from 30 to 50 in such age).
I believe the large changes in player skill are what makes the game so easy to win.

and,

Make players more stable, that is, significant changes can occur only in large span of time (let's say 5 years), unless player is very young. So, players such as Sk=30, Age=27, Tal=70 should be excluded. Some estimation would be to limit Talent to be Skill * 1.1 ^ (MaxPeekAge - Age). 20 years old player, can than achieve 2.6 x Skill in 10 years. On the other hand, he will not improve more than 10% of his skill in a year.

now, i don't want the game to be too easy, of course. and i've got to admit that the above example of a fairly mature player with the given talent/skill values IS unrealistic. so i rewrote the skill/talent system. now it works like this:

• at generation, the players get a peak age value, a peak region value, an age value, and a talent value. the meaning of each is:
  
  • peak age: the age 'around' which the player reaches his best performance (highest skill)
  • peak region: the number of years before and after the peak age the player can hold his performance. this is a kind of skill plateau during which there are only slight random changes in skill
  • age: the player's age at generation time.
  • talent: the player's talent at generation time.
  until now, the average skills were read from the league files and the talents of the players were assigned accordingly at generation. in the new system, the average talent is given in the league definition files.
  
  after these four values -- peak age, peak region, age, talent -- are assigned, the player's skill gets calculated depending on his age relative to his peak 'plateau' and his talent. if he's very young, his skill gets reduced, if he's very old, the same. if he happens to be in his prime, his skill is his talent value.
  
  since the reduction depends very clearly and exactly on his age and talent, improbable combinations like the one alefzero mentioned cannot occur.
• now, during the game, the players get older and develop, just like until now. but their skill develops the same way as during the skill reduction at generation (only the other way round). this ensures that the whole thing is consistent.
• because of these changes, i had to replace the 'average_skill' parts in every league definition with 'average_talent' parts.

i hope the changes work out well. the parameters can be tweaked easily in the constants file, e.g. if the players develop too fast.

i've cvs'ed the changes, you're all welcome to test the new system and give feedback. (i guess i should write a 'how to get a CVS version' soon )

the next thing i'm hoping to add are hot and cold streaks for players.

gyözö

[gyboth]

some thoughts about hot/cold streaks:

• it's unfortunate that vector goes on holidays, i could use a nice fire icon and ice icon (anyone played NHL2003? something like the hot/cold streak icons there).
• i imagine having three new variables in the player structs:
  • streak: the kind of streak the player is on: hot, cold, none.
  • streak_prob: a value between -1 and 1. it expresses the probability to go on a streak; if it tends to 1, the probability of a hot streak is high, if it tends to -1, the probability of a cold streak is high, if it is around 0, there's likely no streak.
  • streak_count: a counter that can be positive and negative. if it's positive, it expresses the number of weeks the player is still on the current streak, if it is negative, it expresses the number of weeks the player may not go on a new streak (to avoid players having too many streaks).
• now the interesting question is: what causes the streak probability to tend towards hot or cold? my thoughts (and i'm sure i'm missing a lot here, you're encouraged to add your ideas).
  
  tends to HOT (the higher the number the bigger the probability increase):
  • if a player plays in the startup formation (5)
  • if a player comes into the match as a substitute (3)
  • if a player scores (10)
  • if a player assists (5)
  • if the team wins (against not-too-weak opponent) (3)
  • if a goalie has a save (2)
  • if a goalie finishes a match without a conceded goal (5)
  tends to COLD (the higher the number the bigger the probability increase):
  • if a player is not in the startup formation (3)
  • if a player gets substituted (2)
  • if a player scores an own goal (10)
  • if the team loses (against not-too-strong opponent) (3)
  • if a player gets injured (7)
  • if a player gets banned (5)
  • if a goalie concedes a goal (not a penalty, though) (3)
• if a player is on a hot streak and gets injured or banned, his streak stops.
• what's the result of a streak? i'd say,
  • skill gets increased/decreased
  • fitness gets reduced less/more if a player plays
  • fitness gets recovered faster/slower if a player doesn't
• what i don't know is, should i make the skill change visible to the user or not? i'm tending towards the latter, as it makes the game more interesting and counteracts the 'i just play with the 11 that have highest skill'-type of management.

gyözö

[gyboth]

first version of player streaks is implemented in cvs. the icons are a bit crude since i had to make them myself

gyözö

[vector]

erm i just cvs and i acnt see any new things, icons or streaks etc ,,, where should i be looking?

[gyboth]

erm i just cvs and i acnt see any new things, icons or streaks etc ,,, where should i be looking?

do you expect players to be on a streak when you start the game? play a bit, preferably losing a few games in a row or winning them, and you should see your players going on a streak.

i even suspect that with the current settings there are too many streaks, but let's wait and see.

gyözö

[gyboth]

the Bygfoot Team (a.k.a. gyboth a.k.a. gyözö ) proudly presents

The World Cup 2006 Germany! (TM, (C) and (R) etc.).

i've made a definition for the world cup. not only the final with the 32 teams of course, but the whole bunch of qualifying stages; there are 138 teams all in all. it was harder than it might seem because quite a few source code changes were necessary.

in this definition, a lot of things are switched off, like finances, transfers (of course) or youth academy.

what still needs to be done:

• assign average_talent values to the individual teams (at least the better-known ones). it'd be a bit surprising if argentina and brazil don't end up among the first 4 in their qualifying group.
• i've skipped a few stages; some qualifications, like the african one, consist of 2 or three stages. i didn't skip a lot of them, but maybe i should add those i skipped (although i doubt anyone would want to try to become WC with guinea-bissau, but who knows).
• add icons (waiting for vector, i guess).
• test. update your cvs and become WC! and report the bugs you encounter!
• i had to change the european qualification a bit, because round robin stages with different number of teams is not supported currently. maybe i should work a bit on the source code to make it possible.

gyözö

[gyboth]

average_talent values are added to all but a few teams in the WC definition.

i simply took the current fifa ranking and assigned average talent 9500 to the first-placed team (brazil), 9470 to the second-placed etc. the last one got 5000 and something

this change should prevent the Faroe Islands or Cyprus from winning the WC

gyözö

[gyboth]

some updates:

• different numbers of teams in cup round robin groups are possible now, so the european qualification for the WC is like in reality
• the average talents for the WC definition are updated and reflect the fifa points, too, not only the rank (thanks to alefzero for writing the script)
• there are two new translations, both of which are almost complete: spanish and chinese. thanks to benyuan liu and Marcelo Garrone.

gyözö

[gyboth]

i've added some command line switches. currently there are the following ones:

Code: Select all

-s, --support-dir=DIR     Specify additional support directory (takes priority over default ones) 

-d, --debug-level=N       Debug level to use 

-l, --lang=CODE           Language to use (a code like 'de')

if there is an argument without preceding switch, the argument is taken to be a savefile. to make things easy, you can omit the '.zip' extension and also the full path, if the file is in the default saves dir; you can also call bygfoot last_save to load the last savegame.

suggestions for additional switches are welcome.

gyözö

[gyboth]

another one of alefzero's suggestions is implemented: season results (figures->season results). that's just a chronological list of your team's matches and results for the current season:


gyözö

[dylansmrjones]

the Bygfoot Team (a.k.a. gyboth a.k.a. gyözö ) proudly presents

The World Cup 2006 Germany! (TM, (C) and (R) etc.).

i've made a definition for the world cup. not only the final with the 32 teams of course, but the whole bunch of qualifying stages; there are 138 teams all in all. it was harder than it might seem because quite a few source code changes were necessary.

PURRFECT! This means that Denmark can still qualify to the World Cup?

[gyboth]

PURRFECT! This means that Denmark can still qualify to the World Cup?

exactly provided you're a good manager

gyözö

[dylansmrjones]

exactly provided you're a good manager

I'm always good

Just won at home over Ukraine - the real team didn't do that

It's gonna be a tough summer next year - I won't be watching it; not when all of it happens just south of the border.

Revenge in 2008!

[dylansmrjones]

Just won 4-0 against Germany in 2nd match in Germany

And of course was second to none in the qualification... Had expected no less

[gyboth]

2:30 am... you really want to become WC tonight

gyözö