ps's bot seems to dominates mine. Played him 6 times and he's beat me every time. This is the best I've done: http://theaigames.com/competitions/warlight-ai-challenge/games/53879cca4b5ab20a017292c8
I might have actually tied (or possibly even won) in that game except he had two lucky attacks that gave him the advantage. The other 5 games he's won in 15 to 30 turns. Perhaps I'm just saving my lucky victory over him for later :P

I doubt he feeds his bot cheese so I figure some of these bots will be a bit anemic by the semifinals and I'll be able to have a lucky streak with my well fed bot. Atleast I assume your prior method of feeding your bots was doing it manually and then updating it and since you can't do that during the competition I figure my lone advantage lies in my cheesiness.

Trogatog: mine should have started expanding into europe once the stacks were high on SA, no idea why it didn't, would have to debug it, but there is no motivation to do that now that the competition is closed :)

Ikalgo: does your bot deploy his newly acquired North America income out of my sight on purpose? To "hide" the bonus? Well done.

http://theaigames.com/competitions/warlight-ai-challenge/games/538a38e44b5ab20a0172c687

pedrito: Yes! I was making specific changes on the last day to counter you and GreenTea; he actually immediately rushed me in North America as soon as he found out the income. So I went for 2 changes, hiding income and going for Australia to attack.

First to hold the border region in Asia so I can protect in time (GreenTea was building a large army in Australia in preparation of going to North America). Secondly, obtain North America, and then finally rush Australia. Note that my calculations allow me to nicely expand. I also monitor Europe at the same time; whenever 2 regions are held in Europe at the same time bordering North Africa, I would "smartly" try to break the bonus.

My main problem was that I wasn't able to test building an army in Asia. In the end, the troops just wander off rather than go for Australia, and they've ran into Dalek so many times causing me to lose (which is very frustrating!), and it happens not only against him. I was trying to keep it safe by implementing those 2 changes, but I believe against the current competition I could just hide income and instead simply first capture North America and then rush Australia; no need to hold Asia because you guys won't be checking unless you see my income.

Ah well, lesson learned.

Edited 2014-05-31 22:48:01

Hello

Can you give a link where you have North America + South America and the opponent has Australia and he then rushes North America from Australia? I don't think this is possible. The only way I can see this happen is when you use your extra +5 armies to expand like crazy instead of finishing the opponent off fast. Once you have more armies than him in Africa the game is won imo.

My bot also can tell that you have North America when you deploy more than 7 btw ;)... he just doesen't use that information properly. It's a problem that when you just start coding without specifying first it's quite hard to add completely new thoughts later.

The easiest solution to win is just keep stacking in Africa until you have enough to hit him. I can't remember my bot ever lost a game in such a situation and my bot doesen't even defend North Africa since he always attacks when there is an opportunity.

Hey,

Norman: Here is the example I was basing the assumptions on (http://theaigames.com/competitions/warlight-ai-challenge/games/5380e9f84b5ab2583f241863). There are a couple of more games like this. I was afraid that simply hiding the bonus wouldn't be enough to prevent the enemy from going to North America sometime in the game (also it is possible that the bots not only consider placements, but also transfers "to"; I know bots of a higher level would certainly take it into account). Before I knew we have enemy placement info, I was checking the amount of troops on the enemy regions at the end of the turn. If 2 turns the same increase, then that's what I would predict for the 3rd turn. This led to some great predictions, but somehow after changing placements instead of end-of-turn-counts the prediction didn't work that well anymore (e.g. I used to attack 2 troops with 5 if I was pretty confident it wouldn't be reinforced, but now this never happens anymore and I don't know why).

Oh and wow, it's so interesting to hear you were struggling with the specifics. I went for a general approach but with my split up strategies I couldn't really fit it and just went specific in the end. I still feel going "general" is the way to go. I guess what we would need is a general system that easily allows exceptions; but how :)?

Edited 2014-06-01 08:54:46

First turn 5 win I've ever had I think. Best prior was 7:

http://theaigames.com/competitions/warlight-ai-challenge/games/538a92944b5ab20a0172cd55

Hi, i uploaded my bot on github, so you guys can have a good laugh :D

https://github.com/125ch209/Gadzbot/tree/8e283bb1f30b4693fa0c8dd84317b4e24e49957d

to Blender: about http://pastebin.com/LryLDniy. Now I understand what you mean by simulation.. Im also tried to do this approash in the beggining, but find out that even with big number of simulations dispersion of results was too high. For example for first 1000 simulated attacks chance of capture was 0.76, and for second 1000 attacks - 0.74. Its not good when behaviour of bot depends so much from random numbers.

But it is possible to precise calculate average losts and also chance to capture using iterative calculation by formula of Bernulli
http://mathbits.com/MathBits/TISection/Statistics2/bernoulli.htm

For my bot simulation is little bit different. It is repeated virtual attack on region to give answer on question "coul'd I capture enemey region when from now I starting to attack with all forces and when enemy will use all his income on defence".

125ch209: wow. ~3000 lines of spaghetti code in one file xDD

1 Germany SupremeDalek-v358 Norman Java WarLight 2025
2 Ukraine GreenTea-v63 GreenTea Java WarLight 2022
3 Netherlands AdsRiskbot-v187 AdsRiskbot C# ExactFamily vdH 2019
4 Spain pedrito-v173 pedrito PHP WarLight 1961
5 United States Trogabot-v625 Trogatog C# The NerderyKansas CityWarLight 1927
6 Canada WisdomsPower-v138 Garrett C# WarLight 1917
7 China Herz-v229 herz Java UCLA 1905
8 Netherlands TheeYo-v246 Thee?Yo! C++ ASML 1899
9 Netherlands Blender-v875 Dualinity Python 1899
10 France Gadzbot-v1336 125ch209 Java WarLightFrench CommunityArts et Métiers Paristech 1890
11 Portugal psWarlighter-v101 psenough C# WarLight 1888
12 United States RaptorBot-v406 AllianceBot Java Ravenwood High SchoolWarLight 1860
13 Dbagbot-v85 Doushibag C# WarLight 1851
14 France KewaKareban-v182 kendou Python WarLightAm I the only one around here who's not in an university? 1841
15 Germany mweb-v478 mweb Java mweb 1837
16 Germany Webmartino-v204 Webmartino Java 1809
17 Germany Speedy2014-v191 Speedy2014 Java 1803
18 Czech Republic HouplessBot-v41 Houp C# Am I the only one around here who's not in an university? 1803
19 Poland Damian_Warlight-v228 Damian Java Wroclaw 1797
20 Switzerland Muli-v196 Muli Java WarLight 1786
21 Netherlands Assimilator-v49 mjjw C# Exact 1772
22 Poland SupremeLeader-v48 Scony PHP Poznan University of TechnologyWarLight 1752
23 Canada Alleirbag-v150 alleirbag Python WarLight 1750
24 Netherlands Toldea-v71 Toldea 1730

http://www.theaigames.com/competitions/warlight-ai-challenge/semi-finals/show

heuuuuu.... pedrito and green tea on the same bracket?? really?

my bracket : 10-12-22-24
green tea: 2-4-14-16

does that makes sens? i mean i'm not complaining but...

Edited 2014-06-01 22:25:26

Being lucky number 13 in the ranks I get the pleasure of playing #1 SupremeDalek first! Fantastic, he'll give me a quick delivery to the loser's bracket and relieve me of any hope of lucking my way into the finals.

The first rounds are set to +12 it seemed like... but the second matches don't match that theme.
IE 1 vs 13, 2 vs 14, 3 vs 15, etc. However the second round would be (assuming no upsets for simplicity) 1 vs 3, 2 vs 4, 5 vs 7, 6 vs 8. If their aim is for the first round is to have 1 through 12 have best odds of winning, then to maintain that theme you'd want to have in the second round 1 vs 7, 2 vs 8, 3 vs 9, 4 vs 10, 5 vs 11, 6 vs 12. But they didn't do this.
That bracket would have looked like...
1 vs 13 <> 7 vs 19 == SupremeDalek vs Dbagbot <> Herz vs DamianWarlight
2 vs 14 <> 8 vs 20 == GreenTea vs KewaKareban <> TheeYo vs Muli
3 vs 15 <> 9 vs 21 == AdsRiskbot vs mweb <> Blender vs Assimilator
4 vs 16 <> 10 vs 22 == pedrito vs Webmartino <> Gadzbot vs SupremeLeader
5 vs 17 <> 11 vs 23 == Trogabot vs Speedy2014 <> psWarlighter vs Alleirbag
6 vs 18 <> 12 vs 24 == WisdomsPowers vs Houplessbot <> Raptorbot vs ToIdea

But then I wouldn't get to be in the same bracket as #1 and #3 and what fun would that be for me? I might get hopeful or something.

Edited 2014-06-02 04:30:56

I looked up my game record history against the 3 bots in my bracket (qualifier games only):

2/2 against SupremeDalek (even, not terrible odds!)
0/3 against AdsRiskbot (not good for me)
6/6 against mweb (that's a solid toss up)

Anyone else done this for their winners bracket?

Edit: Just checked mweb vs AdsRiskbot, 1/6 in Ads favor. AdsRiskbot will beat us all! Maybe. Unless the Warlight Gods decide to shake things up and have powers that extend beyond this humble site. Wonder how many first round 'upsets' there will be. Some people may be reasonably matched despite the rank difference.

Edited 2014-06-02 05:19:17

2/2 against SupremeDalek (even, not terrible odds!)

This ain't no cheap hollywood movie. Here the bad boys win. SupremeDalek will crush douchebag-bot. He will rip you apart. You have hope? Daleks have no concept of hope. Your bot feeds off cheese? SupremeDalek feeds off his own pure hated. The ultimate destiny of you lesser bots is extermination.

EXTERMINATE

"I might get hopeful or something. " == currently without hope.

Yeah my bot is going down, but atleast it won't die hungry. Although if your bot rips mine apart and eats the cheese from its stomach then I guess it won't go hungry either, eek!

vs pedrito we have 18 games. from which:
13 my wins
4 loses
1 draw

But anyway its too low count of games. To get robost statistics even 100 games is not enough with such big luck modifier in game. At least 500 games.

Edited 2014-06-02 08:24:46

groups changed!
http://theaigames.com/competitions/warlight-ai-challenge/semi-finals/show