tactics

[phtrivier]

Hey guy ...

First of all, kudos for the game, it is as simple as it gets, but is great enough to have such a thing available under our beloved OS ..
Now I am going to complain a bit ; I am pretty sure you came to coding Bygfoot after playing nights of Ch****** Mana**** (I am not saying, it, see !) ... Well, anyway, I did, and what I prefered, strangely as it may sound, was the tactical part of it (I am still sure that given the proper tactic, and with a bit of luck, you don't need to cheat and give 1000,00€ par day to each players on your team ... anyway) . Now, is it on purpose that your tactical system (meaning , the way games are solved based on the players are positionned) is so ... simple ? I looked a (too) tiny bit at the code ... and it looks a lot like complicated dice rolling (which has its avandages : it is pretty simple and codable ).
Now my question is : are you
a) planning to change / improve the fixture resolution // tactical system
b) planning to include some AI
c) needing help ( )

( c) is some very subtle attempt to offer my help in case .. you are free to refuse, since that is pretty agressive, and I don't have any real previous work to show my skills .. except that I have some academic knowledge of AI , and a bit of will to test it on a practical problem -- and tired of the "taquin" thing ... )

Anyway, goot luck for the rest of the project.

Regards
PH

By the way, were does your signature comes from ? A very brilliant french comedian used this quote a few times, but I guess it is hard to find the original author ...

-- Life has a meaning. I found a very elegant proof for this, unfortunately this Internet is too small to contain it ...

[gyboth]

First of all, kudos for the game, it is as simple as it gets, but is great enough to have such a thing available under our beloved OS ..

simple on purpose...

Now I am going to complain a bit ; I am pretty sure you came to coding Bygfoot after playing nights of Ch****** Mana****

nope. have a look at the links on the home page.

Now, is it on purpose that your tactical system (meaning , the way games are solved based on the players are positionned) is so ... simple ? I looked a (too) tiny bit at the code ... and it looks a lot like complicated dice rolling (which has its avandages : it is pretty simple and codable ).

hm... good question. in part it is intentional that the tactics don't play a big role in match outcomes. one of my goals was to write a game that's simple to play and understand, without sophisticated and subtle influences. this is also why players have so few attributes. on the other hand i simply wrote algorithms that seemed not too complex but more or less realistic and delivered good and credible match outcomes.

a) planning to change / improve the fixture resolution // tactical system

i wasn't. but i'd be glad if you gave me some hints on what you'd like to improve. i didn't play the game you mentioned, so i don't have a clue what these 'tactics' you wrote about were.

b) planning to include some AI
c) needing help ( )

that depends on what you'd like to offer

( c) is some very subtle attempt to offer my help in case .. you are free to refuse, since that is pretty agressive, and I don't have any real previous work to show my skills .. except that I have some academic knowledge of AI , and a bit of will to test it on a practical problem -- and tired of the "taquin" thing ... )

let's do it like this: you write me some improvement propositions, and i tell you whether they fit into Bygfoot or not. if they do, then go ahead. but before you start, have a look at my first post here.

By the way, were does your signature comes from ? A very brilliant french comedian used this quote a few times, but I guess it is hard to find the original author ...

i really don't know myself. i found it in the personal info page of a FICS user.

-- Life has a meaning. I found a very elegant proof for this, unfortunately this Internet is too small to contain it ...

your's isn't bad either... a fermat-like signature

gyözö

[vector]

Thought it might help if I added my 2 cents here. This is how I started playing the game and what I expected would effect results. I first must recognise, Gyozo's wish to keep it a simple game and I dont believe what Im about to say would contradict this.

Before start of week:
I first check my players health and remove or substitute any inj, banned and unhealthy players.
I generally try to keep as many at 100% as I can (for eg i tend to rotate the goalie every week.

I then, check the transfer list, wether Im after players or not, to help get an idea whats "floating about" and maybe pick up that once in a lifetime player.

Using the TL I take a peek at my next oppenents team. I look at the 3 areas;
defence, midfield and forwards. While the average skill levels give an overall impression of the team I look for weakness in one of the areas. Say defence and I may not only choose "attack" on this find but also weight my team with more fwds.

a team with a heavy defence for instance I may decide that attacking will just wear my men down so i heap up the midfield In attempt to "keep the ball" in their half and hope for a break.

Of course this thinking could go on and on and while describing it may look complicated it takes the briefest of moments to check once your in the swing and can be ignored by the non geek managers without vastly effecting the result but it would allow "great" managers the ability to tweak a not so good team, with little capital, etc to just get one thru

Of course I have no idea how to code this , dreaming is easy, codeing not so fast or even how close it may or may not be to the above already.

[arnaudus]

The tactics remain unclear anyway. I think (but I know Gyozo doesn't agree) that the interface should display clearly the attack and defense scores, and those of the opponent. Something like that:

Code: Select all

                           My team                    Opponent
                            All out attack              Defense
Attack                ======++                  +===            Defense
Defense            ====                              ====            Attack

Then, no more help section is necessary : everybody can see that attack scores are compared with the opponent's defense, and what are the effect of the chosen strategy, etc.

[vector]

I think what Im saying is that I like the way I "have to go fishing for info" ie look at my opponnets side to find week spots.

I dont think the def and attack info and game calculations/tactics should be "on the front page" so to speak.Youd soon learn how to play the game and loose interest.

Good managers are those that go looking.
If Im a rank 2 and the next opponent is a 21 I generally dont bother to check (but im lazy and could/should get punished occasionaly for this by an upset loss). Reverse this, say Im 21 and I have checked my opponent ranked 2 (and he is lazy He hasnt bothered to ballance his side properly and I can see a weak spot. I should be given an advantage somehow for exploiting this.

[gyboth]

I dont think the def and attack info and game calculations/tactics should be "on the front page" so to speak.Youd soon learn how to play the game and loose interest.

exactly my point

gyözö

[gyboth]

hi phtrivier,

i'm beginning work on the new match calculation engine in 1.9. you still haven't backed up your first post with details. if you'd like to have some more tactics in 1.9, you should provide them now.

and another thing that came to my mind: this game you mentioned, ch** m** sounds like some old dos/win game. and those used to be closed-source programs. so how do you know that what you think are 'tactics' aren't 'complicated dice rolling' in reality? just because there are 100 tactics options in a game it doesn't have to mean they have any other purpose than making the dice rolling more complicated.

gyözö

[phtrivier]

hy guboth,

I don't have anything sufficient for the moment (not that I deveoted as much time as I could to it ; not that I claim to have , neither ).

You're right, c**m** is a series of dos win game, for which I have always been quite surprise to find such a huge fan base, but no free software equivalent ; until I discovered bygfoot, and wrongly assumed that you had at some point played the game.

It is true that the game was closed source, obviously, however I had the change to dive into some of its guts, though very superficiously.

The thing is that in c** m**, the "tactics" are description of both were players should be, and of what they should do. So, as a complement to such things as "1 goalie, 4 defs, 4 mids, 2 forwards" (actually, a more precise description also including the side of the pitch), you also add such refinements such as where every player should be when the team is attacking or defending, whether a player should keep the ball or shoot from far away, etc ...). The tactic descibed a playing pattern, that was then applied to a specific set of players (saying that, for instance, Becks is the one player that should stick to the right line, hold the ball and cross as much as possible to the box).
The intereset of such tactics is that it made it, theorically, possible to use "as well as possible" the various players in your team. So, just because the average value of your player is lower than the one of the adverse team, it doesn't mean you cannot design a clever scheme that averages the game.

This naturally relies on the "fairness" of the game engine. Your hopes that spending some time tuning the tactics will have some effect on the game should NOT be deceived, or you will simply go back to the "I need more money to buy stronger player" option. As you mentionned it, this requires either a full-fledge soccer simulation to play the games including all parameters (the one you who excepct to find for the AI of players in Fif* XXXX or Internationnal S* Soccer), or a very-very-very complicated an luckey dice-rolling equation. As you said, I have no way to know for sure, but I dare to think that the "proffi" games we are talking about go for the first option. (Later versions happen to exacly display the game being played, which only makes my presumption stronger). The second option is all the more dangerous as it jeopardize the very essence of the tactic scheme : provide more realism to the game. I completely agree that the "100 tactics options" missed the point at some stage, as it was virtually impossible to detect whether a particular option had any effect on the game !!
It was also, to say the least, pretty frustrating. You always lacked the option you needed ! Forging the tactical files was an option I tried at some point, but the file format was the hell of an hell (I understand you're to use xml for your datafiles : be blessed by all further mainteners for this). A free version of the game should keep the computation free of dice rolling, and open to customisation. (who said "shut up and code it, bloke ?" )

bye
PH

-- Life has a meaning. I found a very elegant proof for this, unfortunately this Internet is too small to contain it ...

[gyboth]

so, let me try to sum up your suggestions:

1. players should have individual tactics assignments (and there are about 4-6 different assignments)
2. the game result should not involve chance (ie, it should be deterministic)
3. the way the results are calculated should be in xml files, not in the source code

now let me gather some thoughts on these:

1. the result computation would be a simulation of a game, basically a sequence of 'who's got the ball, what does he do with it'.
2. each action has a deterministic result, e.g. 'defender A has the ball, forward B attacks him, if A has higher skill he retains the ball, otherwise he loses it to B' (a bit simplified, of course)
3. each such rule is stored in a file and gets loaded at the beginning of the game

finally, let me tell you why i don't like either of the above (of course i could've misunderstood you completely):

1. such a way of calculating match outcomes would be pure hell to balance. ie, you'd probably have to tweak the settings for months to get a realistic simulation with reasonable match outcomes
2. deterministic result calculation would lead in my opinion to boredom in the end. once you've found out how you have to set up your tactics to win, where's the excitement? and even if you lose, you know it beforehand if you've got enough experience. in my eyes, dice-rolling isn't a bad thing. i like things being random.
3. i like Bygfoot being customizable. but i write the game for users, not programmers (or would-be programmers). if someone would like to change the behaviour of the game, he's welcome to dive into the source code. it's not as big a mess as it used to be in earlier versions.

however, you're welcome to take any part of Bygfoot and modify it according to your wishes to create a manager of your own (considering that our opinions on the game internals seem to differ quite a bit). that's what open source is about.

gyözö

[Guest]

so, let metry to sum up your suggestions:

1. players should have individual tactics assignments (and there are
about 4-6 different assignments)

Well, I have no preconception about the number, but yes, players would have individual assignment to a certain extend.

2. the game result should not involve chance (ie, it should be
deterministic)

I never meant to say that (sorry if I did). Game result would of course involve some random elements (to preview your further example : if player A attack players B and he has such a s skill and he is in a certain position and such a role dice is positive (saying that no UFO is landing on the field, or anything), then he gets the ball).

3. the way the results are calculated should be in xml files, not in the
source code

I can't really see any reason why the code to compute the results would not be in the core of the program ; however storing as much of the parameters as possible in external files would sound sane to me (to allow easier tweaking of the system).

now let me gather some thoughts on these:

1. the result computation would be a simulation of a game, basically a sequence of 'who's got the ball, what does he do with it'.

Unfortunately, that could lead to this.

2. each action has a deterministic result, e.g. 'defender A has the ball, forward B attacks him, if A has higher skill he retains the ball, otherwise he loses it to B' (a bit simplified, of course)

With some randomness involved, but details of the computation might sound like this at some point -- if you want to simulate the game, that is situation you have to handle.

3. each such rule is stored in a file and gets loaded at the beginning of the game

If you have decided that rules are stored in files, obviously. otherwise, it is just loading the parameters of the engine, and running it.

finally, let me tell you why i don't like either of the above (of course i could've misunderstood you completely):

1. such a way of calculating match outcomes would be pure hell to balance. ie, you'd probably have to tweak the settings for months to get a realistic simulation with reasonable match outcomes

Sure. You might use statistical automatic testing for some games where you "know" the result, but that would not be the easiest part. (God, i am taking too much AI classes, i am almost talking about reinformaent learning whereas I can't get to add a bloody row to a GTK table ... )

2. deterministic result calculation would lead in my opinion to boredom in the end. once you've found out how you have to set up your tactics to win, where's the excitement? and even if you lose, you know it beforehand if you've got enough experience. in my eyes, dice-rolling isn't a bad thing. i like things being random.

Fair enough. Even if you change deterministic with "rather deterministic", that is really a question of taste. I am just afraid that randomness can, also , lead to boredom ; since there is virtually nothing you can do to improve your team but buy better players and pray harder than the CPU.
(I think this has to do with the history of the side I am supporting, which always managed to survive by cleverly using cheaper player rather than buying stars )

3. i like Bygfoot being customizable. but i write the game for users, not programmers (or would-be programmers).

(ok, message received. harsh, but fair )

if someone would like to change the behaviour of the game, he's welcome to dive into the source code. it's not as big a mess as it used to be in earlier versions.

Looks fine, as far as I have dived in it.

however, you're welcome to take any part of Bygfoot and modify it according to your wishes to create a manager of your own (considering that our opinions on the game internals seem to differ quite a bit). that's what open source is about.

You don't have to believe me (and you probably won't), but I tried. Unfortunately I lacks time, knowledge of the basics of GTK, and actual motivation (which is why I am not going to bother you anymore, but will keep looking at the game ).

Sorry for disturbing.
Keep up the good work ...

Bye
Pierre Henri

[gyboth]

first of all, thank you for being polite and understanding.

second, i have some good news. i've followed your suggestion and moved almost all game-behaviour-influencing constants to a text file (not xml, though, a simple key-value pair list). i didn't realize i was probably doing exactly what you suggested until after i've done it have a look at this post. the constants file looks more or less like this.

third: if the live game in 1.9 works out like i hope it will, a good manager should be able to get good results even with a mediocre team by using boost, style changes and substitutions intelligently. to see how the 1.9 live game works, get a cvs version of the 1.9 branch (easiest done with the bygfoot_update script).

now let me see if there are things i should comment

I can't really see any reason why the code to compute the results would not be in the core of the program ; however storing as much of the parameters as possible in external files would sound sane to me (to allow easier tweaking of the system).

this is exactly what i've done, thanks for your hint. indeed this is much better than having all the constants in the code.

If you have decided that rules are stored in files, obviously. otherwise, it is just loading the parameters of the engine, and running it.

i'm taking the second possibility.

Sure. You might use statistical automatic testing for some games where you "know" the result, but that would not be the easiest part. (God, i am taking too much AI classes, i am almost talking about reinformaent learning whereas I can't get to add a bloody row to a GTK table ... )

definitely too much AI classes i don't want to write a new program that helps me write the original program. could lead to a vicious circle

Fair enough. Even if you change deterministic with "rather deterministic", that is really a question of taste. I am just afraid that randomness can, also , lead to boredom ; since there is virtually nothing you can do to improve your team but buy better players and pray harder than the CPU.

now, i hope that the new live game will be a good medicine against boredom. though it is a lot more difficult to balance than the old concept and i'll have to find out how to do things without making it too easy or hard to win games.

(ok, message received. harsh, but fair )

sorry for being a bit rude...

You don't have to believe me (and you probably won't), but I tried. Unfortunately I lacks time, knowledge of the basics of GTK, and actual motivation (which is why I am not going to bother you anymore, but will keep looking at the game ).

of course i believe you. and believe me, GTK isn't hard. after some initial experimenting, you don't need any information besides the api reference. (of course i'm using glade, so i don't really build the interface myself.)

Sorry for disturbing.
Keep up the good work ...

sorry for bashing most of your suggestions. and i really hope you have a look at newer Bygfoot versions and cvs versions and the forum from time to time because your ideas are mostly rather insightful, even if they're not appropriate for the game.

gyözö

[vector]

third: if the live game in 1.9 works out like i hope it will, a good manager should be able to get good results even with a mediocre team by using boost, style changes and substitutions intelligently

THIS ROCKS!!
Seriously heaps of fun. I have indeed played with this 1.9 live game feature.. its terrific. As the game prgresses you can really play tatctics like having a flurry of attack with boost just b4 half time or seeing your winning easy.. rest good players by subbing them off etc. Exploit opposing weaknesess by changing structure.... its great