|
Post by rawcookiedoh on Nov 5, 2009 19:31:24 GMT -5
First off Zack(Travice), you do have and ass for a face, lol. Secondly, I think that we should also be scored by our stay percentage and our win percentage.
I've seen peeps with uber high scores and ridiculously low stay rates and win percentages. What this means to me is that they are either 1) farming their asses off all game and ignoring their buddies, 2) purposely delaying the game and in effect losing due to not pushing or 3) farming, pwning, dying once or twice, and ultimately leaving.
If this is already in effect then I'll just give myself an enima with tabasco sauce.
|
|
|
Post by Travice on Nov 6, 2009 1:20:47 GMT -5
I agree with this 100%, perhaps we can change the way the bot weighs certain stats to the score.
|
|
|
Post by Deege on Nov 6, 2009 13:12:22 GMT -5
Sounds good, I'm going to ask the bot creator (or look through the source files) to see exactly what variables I can use with the bot's scoring formula. I'd really like to add stay rate in there along with make score based off of time instead of games. So instead of having an average of 7.1 kills per game, I'd rather have a formula that translates it into kills per hour of game play (since the bot has to record how long each player has played on the bots in order to calculate the stay rate).
At the very least, I'm going to make it so that wins weigh at least as heavily (if not more heavily) than KDA.
This is the current formula, all based off of averages per game:
((wins-losses)/total scored games)+(kills-deaths+assists/2) +creepkills/100 +creepdenies/50 +neutralkills/50 +raxkills/6 +towerkills/11
The way it's set up right now, kills/deaths/assists make up for the bulk of your score, your win percent can only add or subtract a max of 1 point, and everything else adds anywhere from a tenth of a point to a point.
Here's a quick make over until I figure out what variables I can actually use, if you have suggestions to the weights, let me know:
(kills-deaths+assists/2) * win% + (the rest of the stuff, creeps, towers, rax, etc.)
This formula will bring everyone's score down slightly because nobody has a 100% win rate. But remember, everyone's score will go down a little, so it evens out.
|
|
|
Post by beastly on Jan 10, 2010 8:09:03 GMT -5
I'm wondering if multiplying the current formula with stay rate would be good:
[ (wins/totalgames)*(kills-deaths+assists/2)+creepstuff ] * (games completed/games played)
So with your banning system (85% stay rate at least I believe), players with an exceptionally high score but low stay rate will have their score drastically lowered compared to people will people with a low score but high stay rate.
|
|
|
Post by Travice on Jan 10, 2010 19:55:44 GMT -5
The problem with this though is it will kill the score of anybody who is new to the bot system, and this will further cause our bots to be less populated.
|
|
|
Post by Deege on Jan 11, 2010 17:17:42 GMT -5
The other problem is that I do not have the programming skills to make that a possibility. Right now the scoring formula is not coded to accept stay rate as a variable. I have posted requests on the bot forums (www.codelain.com/forum) to change this, but they said it would take an enormous amount of rewriting in order to accommodate new variables into the scoring formula. Right now the only variables that are accepted are all the ones in the current formula.
|
|