|
Post by RS-icsulescu on Jan 29, 2012 22:00:42 GMT 2
Right now, the match engin compares the ratings of the bots teams. What I want to make is a more realistic engine where the match events are played like in real life.
Each match will have 40 attacks. There will be 3 set pieces: scrums, lineouts and kickoffs (for the moment. later I might add mauls). When a team wins a kickoff, scrum or lineout gets the possesion and starts the attack. The player who has the ball (if kickoff can be anyone, scrum no. 8 or no. 9, lineout no. 9) will pass. The first pass of an attack phase will be an easy pass so almost every of this passes will be succesfull. If not, there will be scrum, lineout of loose ball. If the pass is succesfull, the player who recieved the ball will attack. A defender will try to tackle him. If the defender will tackle him there will be: - scrum for the defending team or for the attacking team - lineout for the defending team - penalty for the attacking team - ruck If not, the player will try to pass the ball to another team mate. During an attack maximum 4 players can touch the ball. After the 4th player touches the ball the next attack phase will start.
P.S.: Any suggestions for: KICKOFFS LINEOUTS HOW TO CALCULATE PASS SUCCESS HOW TO CALCULATE TACKLE SUCCESS ?
More details soon...
P.S.: I'm waiting for your suggestions.
|
|
|
Post by gillou76 on Jan 30, 2012 10:34:48 GMT 2
It's a great challenge to try to do a so accurate match engine, but... When going into details, I am very afraid to encounter a lot of frustration, because I'm not sure everybody share the same opinion about what is the key of a successful match. When reading you, I have the feeling your model is the following : - forwards gain the ball - backs gain field advance thanks to individual dueling. I would be very desappointed if gaining the ball in my 22th, with a few advantage within the score, my defensive team plays with hands instead of kicking to gain field. And what about 40 attacks ? It could be less or more... For instance, what do you think of the tactic of keeping the ball with no field advanced, just to prevent the opponent to gain a final attack (and in this case reduce the amount of attack phase) ? Another point is (and this one could be more easily added in your model) : the result of the current individual duel is not the result of attack skill against defense skill. Since forward try to consume opponents before releasing the ball, since half use their intellect to deploy the ball for a positive attack position, since good pass, or good fixation may lead to remove a defenders from the play.... all the previous phase actions have to add some bonus/malus to the current duel. It's all for a first glance
|
|
|
Post by ohdaesu on Jan 30, 2012 10:50:16 GMT 2
As you said first pass should be succesful unless your passer (n.8,shalf or flyhalf) is bad. After then player may pass or penetrate due to his best skills. And should be successful in passing if handling is good or should breakthroug if brth. is good. There should be an algorithm to decide what will be the next movement and what will be the result. There will be several things to effect decision. In case of penetrate brethr and speed skills of offensive player, tackling and strength of defenders, and some luck which shouldnt effect more than %10 to result. There must be a well developed algorithm for that and this may be improved by time. If the tackling is succesful there will be ruck or maul. And this will be an issue to talk about. If offense is succesful then the speed of players is important. Catch him if you can. One important thing to talk is the position of the line. Can it be implemented how far to the try line, and how far the offense team make their way to the line. That will be very helpful. There are more issues about penalties, drop goal attemps e.t.c and there should be more specific tactics also. We will talk step by step.
|
|
|
Post by gillou76 on Jan 30, 2012 14:33:42 GMT 2
Player has to take a decision according to a state. Then what could be this state ? I imagine at least 4 values : - field position (in our 22, in opponent's 22), in short the remaining distance to the goal zone. - overflow amount (how high is the attack advantage in the width of the field). High value helps wingers to have a long run along the side, or center to break through without early takling. Low value prevents to gain advantage in run. - penetration amount (how deep is the attack in the opponent defense). High value helps mid run, since one or more defensive lines are cancelled. Low value state the the whole first defensive line is there just in front of the poor man with the ball. - forwards distance. Not so evident, but I think a player choice will not be the same if he is near his forwards. Then, according to theses states, player choice could be : - early pass : increase "overflow amount" without distance gain - run and pass : gain few distance without direct oppostion and then give the ball (add a 5th pass, since the player is not tackled). - fix and pass : go to duel (increase "few overflow", mid "penetration amount") and pass. - penetrate : gain the maximum distance (with penetration) till the tackle. If tackle never comes, then - kick over : gain mid distance, gain "penetration amount", but next possesion will be very hazardous. - kick outside : gain long distance, but possession is lost. The state values then can help player actions : - "run and pass" and "penetrate" with high "overflow amount" helps to run long distance. - recovery chance will be higher after "penetrate" is process when "forwards distance" is low. - with high "penetration amount", longer distance may be gain to. Pass may be more difficult, cause in middle of the defense ? Not sure. Edit : these "overflow" and "penetration amount" are the idea that rugby is a collective battle. In my mind, these values are the realizations of previous players work (without any distance gain), that will help last player to gain significant distance.Further with this idea, defender may have an impact on this values Edith 2 : seems I have forgotten the kicking (Drop goal) action. These can suffer a big malus if "penetration amount" is high (defenders are all around the kicker).
|
|
|
Post by gillou76 on Jan 30, 2012 15:08:08 GMT 2
Next ? ;D
Then the choice algorithm is just a multi-dimensional table (that could be easily tuned with feedbacks) :
IN : - field position (not sure it's the right word, not so good in english)... the distance reamining till the goal zone. - overflow amount - penetration amount - forwards distance - player position or skills ? - amount of players remaining in the attack phase.
OUT : - either THE action to process, or % of processing for each action... random or not here ? do not know... perhaps pourcentages are better, since tactics and others could be applied by altering these pourcentages.
|
|
|
Post by RS-icsulescu on Jan 30, 2012 15:48:54 GMT 2
It's a great challenge to try to do a so accurate match engine, but... When going into details, I am very afraid to encounter a lot of frustration, because I'm not sure everybody share the same opinion about what is the key of a successful match. I know it's a challenge, but I'll have plenty of free time in the summer holiday . This is my objective with the new match engine. Right now, you have in the help section skill importance and you see team rating after every match. This is too easy and some users might get bored. In the new match engine it will be more important to have multi-skilled players. And I think this is the key to success. When reading you, I have the feeling your model is the following : - forwards gain the ball - backs gain field advance thanks to individual dueling. My model is real-life . Forwads will usually gain the ball, but in the kickoffs backs could also gain possesion. When talking about attack/defence this is more complex. For example: if your team is playing expansive and your opponents is also playing expansive, your backs should be good in attack but also in defence (although forwards will also tackle). But if your opponent is playing driving then your forwards will need good defence. I would be very desappointed if gaining the ball in my 22th, with a few advantage within the score, my defensive team plays with hands instead of kicking to gain field. Kicking will be used only for dropgoals, convertions and penalties. You can advance about 60 meters in each attack phase (if you have a very good team). So getting out from your own 22 isn't that hard. And what about 40 attacks ? It could be less or more... For instance, what do you think of the tactic of keeping the ball with no field advanced, just to prevent the opponent to gain a final attack (and in this case reduce the amount of attack phase) ? 40 attacks because if there will be a random number some teams may loose the offensive bonus point (in a match will less attacks). Too much random isn't good. I like this tactic. I will make another dropdown menu with a list with some tactics you suggested me. This is the tactic you are talking about? Strategy (closed / def / equal / off / open) Another point is (and this one could be more easily added in your model) : the result of the current individual duel is not the result of attack skill against defense skill. Since forward try to consume opponents before releasing the ball, since half use their intellect to deploy the ball for a positive attack position, since good pass, or good fixation may lead to remove a defenders from the play.... all the previous phase actions have to add some bonus/malus to the current duel. It's too complex what you suggest. I want to start with something simple and then improve it.
|
|
|
Post by RS-icsulescu on Jan 30, 2012 15:56:15 GMT 2
As you said first pass should be succesful unless your passer (n.8,shalf or flyhalf) is bad. After then player may pass or penetrate due to his best skills. Usually players will use their best skill: pass or breakthroughs, but they won't always use their best skill. One important thing to talk is the position of the line. Can it be implemented how far to the try line, and how far the offense team make their way to the line. That will be very helpful. Yes. This will be implemented. You will see the position of the ball and team who has possession for every attack phase. This way, it will be more exciting to watch the match live and see what will happent after your team is 5m from the try line .
|
|
|
Post by RS-icsulescu on Jan 30, 2012 16:02:36 GMT 2
Player has to take a decision according to a state. Then what could be this state ? I imagine at least 4 values : - field position (in our 22, in opponent's 22), in short the remaining distance to the goal zone. - overflow amount (how high is the attack advantage in the width of the field). High value helps wingers to have a long run along the side, or center to break through without early takling. Low value prevents to gain advantage in run. - penetration amount (how deep is the attack in the opponent defense). High value helps mid run, since one or more defensive lines are cancelled. Low value state the the whole first defensive line is there just in front of the poor man with the ball. - forwards distance. Not so evident, but I think a player choice will not be the same if he is near his forwards. What if you gain possesion in your own 22, and there is minute 76. There will be only 2 attack phases so your players can score a try only if they attack by passing and don't give the other team possesion by kicking the ball outside the field? This seems very complex to me.
|
|
|
Post by gillou76 on Jan 30, 2012 16:20:02 GMT 2
Kicking will be used only for dropgoals, convertions and penalties. You can advance about 60 meters in each attack phase (if you have a very good team). It seems that this point is closed to you 11-14-15 players would be able to user their kicking ability.... like IRL. 40 attacks because if there will be a random number some teams may loose the offensive bonus point (in a match will less attacks). I rather though at the following : each offensive phase takes 2 mintues.... but, if there are some turnover early in the phase, then the phase has only consumed one minute. Or if the attacking team play carefully, the phase consums three minutes. Too much random isn't good. agree. I like this tactic. I will make another dropdown menu with a list with some tactics you suggested me. This is the tactic you are talking about? Strategy (closed / def / equal / off / open) I'm not sure I understood all I put in the post since it was the translation of a game rule. here is what I understood : - tactics : it's what players have to do with the ball (for instance the new idea to keep the ball without gain, or other tactics already in RS). - strategy : it's quite a state of mind : how do your players share their energy ? closed => full defense (defense bonue, attack malus), open => full attack (the opposite). It's too complex what you suggest. I want to start with something simple and then improve it. Yes, if you want to have such an accurate model (individual duel), it's complex
|
|
|
Post by gillou76 on Jan 30, 2012 16:30:09 GMT 2
What if you gain possesion in your own 22, and there is minute 76. There will be only 2 attack phases so your players can score a try only if they attack by passing and don't give the other team possesion by kicking the ball outside the field? This seems very complex to me. Ok that's the point : rather one simple comportment than a more complex one, which can lead to inconsistency. Yes I agree. Perhaps another variable has to be added, that shows the need to take risky choice to score... Even if you remove this defensive action, I like the principle of counters(overflow amount, penetration amount) that increase attack efficiency thanks to first players work.
|
|
|
Post by LA-wezegaga on Jan 30, 2012 18:25:24 GMT 2
I think we ned to re-think offensive and defensive tactics. You want users not to get bored ? Players should "use" different skills whether the team attacks this way, or this one. For example there should be an offensive tactic which would mainly use breakthroughs and speed, another mainly passing between backs (so handling and speed), and so on. There also should be more defense styles too, (I don't know about names, juste bringing the possibility to change your style of attacks from a game to another). This way training will become more interesting : some players would be optimised for passing, others for breakthroughs, ... About salaries, I ask again for something using the best 3 skills instead of the best one. icsulescu, if you agree I can make a proposition to you. I already have an idea, which would lead to lower salaries, and lower salaries for multiskilled players
|
|
|
Post by gillou76 on Jan 30, 2012 18:53:21 GMT 2
It let me think about Charazay :
it's up to the manager to give individual order for each player (so that Match Engine can not process bad choice ^^) :
For each player, one has to choose :
- offensive/defensive priority. - offensive order : overflow (speed+handling), normal, penetration (breakthrough+speed),... kicking (to score ^^) (stamina+kicking). - defensive order : interception (speed+handling), normal (tackling+speed), secured (tackling+stamina?),...
perhaps offensive and defensive order has to be different for forwards ?
|
|
|
Post by gillou76 on Jan 31, 2012 9:22:28 GMT 2
Hello,
another idea ;D ;D.
KICKOFFS Here is how I think about this :
1) Select potential receivers according to kicker skills. For the kicking team : - if the kicker kicks very well, it may be the best player for receiving - if the kicker kicks well, it may be random player among backs - if the kicker kicks normal, it may be random player - if the kickers kicks bad, no players of kicking team are selected
For the receiving team : - the direct opponent of kicking team selected player is selected - adding one good receiver, if receiving team is more likely to recover.
2) All selected players try to receive the ball : the best one got it ! (then it could be 2 players for receiving team and 1 for the kicking team). If more than 1 player for a team, 2nd one could have a malus ?
3) As there is no well organized team line when the ball is catch up, it seems to be, that the player with the ball can act as it was the 3th or 4th pass of an attack.
|
|
|
Post by gillou76 on Feb 1, 2012 14:49:56 GMT 2
I'm back ^^.
what about pass and tackle... here are some ideas :
This is the way I imagine a duel (if you remark bonus given from an attacker to another, or usage of kick, it's just an illusion ;D).
attacker has the ball, then duel will be resolved within 4 or less steps :
1) gain distance and avoid duel as long as possible - it's mainly a speed/vs speed duel - here is the distance gain
2) attacker and defender are closed together : it's duel time - it's mainly breakthrough against tackle - according to a table the result could lead to : 3a) attacker win duel 3b) defenser win duel
3a) attacker wins duel according to the best skill for the attacker, there is an extra before the attacker is catch up : - speed : few extra distance gain, then pass - handling : next player will gain twice distance (good if next player as lot of speed), then pass - strength : gain a maul - kicking : may kick for a drop goal otherwise pass - breakthrough : few extra distance gain and pass.... or gain a penalty ? - tackle : nothing (go to pass) - stamina : gain an extra pass in the phase (only one per phase) and pass 3b) defenser win duel according to the best skill for the defenser, there is an extra : - speed : nothing (go to pass) - handling : try to steal the ball, if unsuccessfull, go to pass - strength : part of attacker distance is lost, then go to pass (or ruck ?) - kicking : lot of attacker distance is lost (foot clearance), then attacker begin with another attack phase - breakthrough : nothing (go to pass) - tackle : attacker falls without pass : it's either a ruck, or defenser gain a penalty - stamina : remove a pass in the phase (only one per phase), and then go to pass
4) The pass (if no other conflicting event) - it's mainly handling vs... experience >if it succeeds, go to next attacker >if it do not succeed, ball lay down : go to recover the ball (like a kickoff ?) >if there is a fumble, then there is a turnover
|
|
|
Post by RS-icsulescu on Feb 1, 2012 15:12:47 GMT 2
I like your ideas. I'll update the 'Match engine' topic (the which is closed) in the weekend. I also like the bonus ideas, but I'll first implement without bonus, because If there will be a lot of variables in the game, the ME could become messed up. So my idea is to make something simple, then improve.
|
|