- Updated Saturday, September 5th 2015 @ 18:31:43
So now that we have one week left to make changes, Who is gonna win? What strategies are you guys coming up with?
I think this is a good time to start some trash talk.... I can get it started: I think my bot is going to kick ass.
- Updated Saturday, September 5th 2015 @ 20:33:43
On a more serious note I am very excited by all the activity the game is getting, everyone is updating their bot a lot.
On the region drafting side what strikes out to me is Anila8's drafting strategy. Most of the games I play against Anila8 I probably lose just in the drafting stage. I am still working on my strategy.
- Created Sunday, September 6th 2015 @ 01:48:08
With the current versions of the bots, I think the winner will be one of these 4: GreeTea, Zoulander, LuckyBot or Anila8. Given the win/loose rates between these 4, I'd put my money on GreenTea. Also GreenTea seems to have the best knowledge of what his opponent is going to do.
I'm still working on my bot Anila8.
@cowzow, thanks for the compliment, but... what do you mean with "the drafting stage"?
- Created Sunday, September 6th 2015 @ 02:23:34
Oh by drafting stage I mean at the very beginning where each player drafts their starting regions. I prioritize choosing the starting region in a SuperRegion with size 3 and bonus 3, but I often get trapped by you lol.
- Created Monday, September 7th 2015 @ 07:49:32
Ah, that one. Yes, I call it my honeypot strategy. It's indeed quite successful:) Anyway, you have still 6 days left...
- Updated Saturday, September 12th 2015 @ 22:31:05
I think now top 4 bots have almost equal skill. Zoulander and Anila8 are greatly improved in last few days, im impressed! you really got me neurvous in this Saturday and im coding all the day like a maniac to fix all most critical vulnerabilities.. Hope this helps.
Current format of finals (best of 5 or 7?) will be quite random. In my local tests when I fight new version with old one I could easily got 10 wins in a row, and then 15 losses in a row. Even with 500 games set, I can got for example 56% win rate of new version against old one in first 200 games, and finish 500 games with 51% win rate. So you can estimate how big dispersion could be :)
- Created Saturday, September 12th 2015 @ 23:49:38
I agree: it's so close that it's almost random who is going to win, but it will almost certainly be one of GreenTea, Zoulander, LuckyBot or Anila8. Only one day to go. Not sure if I dare to upload a new version at the very last day...
- Updated Sunday, September 13th 2015 @ 00:18:50
@GreenTea: I just looked over a couple of your games. Your board understanding was (as expected) very strong but I believe you are missing some very easy stuff with your move order:
http://theaigames.com/competitions/warlight-ai-challenge-2/games/55f49e1c1c687b0946a827a4 --> Turn 10 wrong move order with your attacks, would have costed you your bonus if you went first. (though nothing bad happened and game was already over at this point).
http://theaigames.com/competitions/warlight-ai-challenge-2/games/55f48f5c35ec1d06d15dfee1 --> Again move order turn 2 with your expansion. A bot with a very strong board understanding would have moved into Super Region 4 this turn. My easy bott just calculates the distance between the opponent and myself and does his 3v2 expansion steps first where he calculates the most distance.
http://theaigames.com/competitions/warlight-ai-challenge-2/games/55f4a2b135ec1d06d15dffe0 I challenged you this game. As expected very strong board understanding turn 5 with the 4v2 preventing me showing up at your SuperRegion while you showing up at mine.. However I expect that a better bot (like GreenTea) would have attacked there also with a 4v2. The bot who gets the first 4v2 ends up with the territory.
- Updated Sunday, September 13th 2015 @ 01:19:26
@Norman:Hi, about games
1) Agree.. Fixed. Attacks from my bonuses which could be broken should go with max delay.
2) Yes.. But too complex for my bot to implement at this point. Mechanism of preventing enemy expand working only in one turn ahead.. However, thanks for idea, maby I could invent something.
3) Its mindgames about how much armies enemy will use to come closer to my bonuses :) if he wants to 4v2 then maby better delay and attack 5v3, which will break his 3 left armies.. Moot point.