|
Forum: Games Topic: RTS stats and theory started by: Malcolm Posted by Malcolm on Aug. 16 2013,10:46
< Kind of interesting >.
Posted by Malcolm on Aug. 16 2013,10:49
Goddamnit. Someone move this to "Games != UO" forum.
Posted by thibodeaux on Aug. 16 2013,10:51
QUOTE Researchers at North Carolina State University WOOO! Wolfpack! Posted by Leisher on Aug. 16 2013,11:09
Moved.
Posted by TheCatt on Aug. 16 2013,13:14
Ive always wondered about approaching it this way.
Posted by Malcolm on Aug. 16 2013,18:49
(TheCatt @ Aug. 16 2013,15:14) QUOTE Ive always wondered about approaching it this way. I've been thinking about this since the days of Civilization I and Warcraft II. Although, Civ is more turn-based. Posted by GORDON on Aug. 16 2013,18:56
Yeah, most coders think like that, I bet. In UO we used to think that way, more or less, adopting tactics that were much ridiculed but very successful. The problem with LOL is that there are too many variables. Champ selection is just the first stage. How you build them during the game makes too much difference. I'd love to be able to code a program that scrapes data from a particular match and tells your team the most likely successful way to thwart their team, but again, too many variables as the game goes on. And, the good players adapt their tactics to account for your build and tactics. Too much data. Posted by Malcolm on Aug. 19 2013,16:27
QUOTE The problem with LOL is that there are too many variables. Pshaw. QUOTE Too much data. Bull to the shit. No one knows everything. No one is James Bond. Everyone has a weak point. It's merely a matter of finding and hitting it repeatedly. Questions are ... 1. What does this data look like? 2. Data for who? You, your opponents, your team? 3. Do you have a way to pipe in data from the current game into a program somewhere machine? 4. How much data do you have access to from a historical perspective? With enough time and targeted history, the real time analysis doesn't have to be god-like. |