EPT
Starcraft Game Theory
| Blogs > motbob |
motbob
United States12546 Posts
| ||
Aesop
Hungary11233 Posts
If I see it correctly, you want to talk about strategy around starcraft, that isn't actually part of the game itself? If so, you could go deep into psychology. Choose a "shocker" strategy in a first set of Bo5/Bo3, to set the mind for the rest of it, as July did to Best in the OSL finals, or as by.great did against Bisu with the aid of July (all-in first game to make him cautious in the macro game 2). | ||
|
Abydos1
United States832 Posts
| ||
|
d_so
Korea (South)3262 Posts
perhaps one aspect of pro gaming that could work is who a coach picks for Ace matches or for the upcoming Winner's League. For example: 1.) The map is andromeda 2.) The following are winning percentages for the map and the top players for each team: Winning percentages Team A B Zerg B Terran B Protoss A Zerg 50 75 25 wants zvt, avoid zvp A Terran 60 40 80 wants tvp and tvz, avoid tvt A Protoss25 40 40 wants to avoid pvz, pvt, pvp Team B A Zerg A Terran A Protoss B Zerg 50 40 75 wants zvp, avoid zvt B Terran 25 60 60 wamts tvt amd tvp, avoid tvz B Protoss 75 20 60 wants pvz, pvp, avoid pvt based on these percentages, coach A picks ________ and coach b picks ____________ something like that edit: actually, i could see it for BO now too. for example, regarding 14cc player 1 believes if he goes 14 cc, he will have an advantage if the counterpart T goes siege expand or facport, and that they will be tied if the opponent goes 14 cc as well. However, if the opponent goes BBS or fast academy --> MnM + SCV all in, then player 1 believes he will lose. player 2 believes he will lose with all build orders unless he uses BBS. so the question is, will player 1 use 14cc? | ||
|
ondik
Czech Republic2908 Posts
| ||
|
ondik
Czech Republic2908 Posts
| ||
|
d_so
Korea (South)3262 Posts
| ||
|
thunk
United States6233 Posts
Flash versus Hero on Proleague, watch a lot of SKT1 games. They get sniped like every time; Bisu usually gets a solid PvP'er or a PvT'er and almost never a PvZ; it seems that Best always gets put out PvZ these days. But I think you'd really want Best PvZ'ing instead of Bisu right now. OGN made a spectacular run at Proleague playoffs last year through solid Ace snipes and good match-up making. | ||
|
Abydos1
United States832 Posts
On January 16 2009 09:36 ondik wrote: I mean.. in prisoners dilema if you finch you are better of in both other prisoner's reactions. Whereas in, as d_so mentinoed, PvT game it's not that clear. There is no pure strategy equilibrium, there may be mixed strategy equilibria, though i'm not sure how far did u go with GT, if further than bayesian games, imperfect information and so.. Part of the problem is that your strategy is based on how current opponents would be playing it, this defines the current meta game. For example mech is seeing a lot of use lately TvZ, if zerg all the sudden come up with a good counter it is going to see a lot of use leading to the decline of mech. FE protoss is just better in the current macro heavy meta game so its used almost exclusively. You could examine coach's strategies for the all-kill winners proleague this round. For example, is it better to start off with a solid mid level player or an S-Class programmer who might be vulnerable to a snipe because of a poor match-up. | ||
|
t_co
United States702 Posts
On January 16 2009 12:42 Abydos1 wrote: Part of the problem is that your strategy is based on how current opponents would be playing it, this defines the current meta game. For example mech is seeing a lot of use lately TvZ, if zerg all the sudden come up with a good counter it is going to see a lot of use leading to the decline of mech. FE protoss is just better in the current macro heavy meta game so its used almost exclusively. You could examine coach's strategies for the all-kill winners proleague this round. For example, is it better to start off with a solid mid level player or an S-Class programmer who might be vulnerable to a snipe because of a poor match-up. It seems that PvT and even mirror matchups with non-mirror BOs could be modeled by a decision tree which has set decision points at certain timings. E.g. in PvT terran faces a decision after FE, Protoss faces a decision after 2gates are up, etc etc Each decision tree could have certain payoffs at certain timings (e.g. obviously expansion timings but also tech payoffs) So you could actually make it quite complex, nonlinear (as decisions can loop) and also nondiscrete (because you could split the decisions down as far as you wanted to). | ||
|
Savio
United States1850 Posts
I'm interested to see where you go with this. Keep us up to date. Those of us who have taken game theory can answer any specific questions and give advice, but mostly I wanna see what you do with this. | ||
| ||
