I spent a bit of time over the last couple days collecting the new XML game data ( http://blog.warlight.net/index.php/2011/03/game-data-feed/ - thanks, Fizzer!). The first thing I have done with that data is calculate the most and least valuable bonuses on the 1v1 ladder. Now, I am not going to post the data quite yet. First, I'm curious to hear some thoughts on what you folks think the results will be.

So, lets hear it! What do you think the best and worst starting bonuses are for Stategic 1v1 settings? The only thing I will say ahead of time is that there are no truly dominant starting spots, but there are some very bad ones. At least one of these surprised me.

I have no guesses, but man have I been waiting to hear the results of this! I would still love to see a ladder that has automatched starting spots.

What's your criteria? The spots that the winners of each game chose?

Yes, pretty much, Fizzer. I recorded the number of times each bonus was occupied after territory selection, and the number of times that the player won the game when choosing that territory. Then I just divide times won by times chosen to determine how often a player wins the game when he starts in the bonus. The best have the highest percentage, the worst have the lowest percentage.

It would be so much more helpful if there were not random wastelands. The results while interesting from a trivia perspective, have no actual bearing on the value of the spots in actual games.

|> It would be so much more helpful if there were not random wastelands. The results while interesting from a trivia perspective, have no actual bearing on the value of the spots in actual games.

Sure they do, they're just not the only thing to consider.

The wastelands throw off the data a little, but given that they're random they should throw off everything evenly and still make the data consistent.

Further, players rarely choose spots with wastelands in them. However, the data might be slightly more accurate if you were to exclude starting bonuses with wastelands.

My guess for the best are the small bonuses: Central America, Scandinavia, Antarctica.

The worst is a bit more tricky since I'm sure things like East Russia rarely get picked. But if you're really calculating it that way, if even one person chose East Russia and lost then it's a 100% loss rate, making it impossible to beat as the worst bonus. So I'd have to guess East Russia and Canada.

If it turns out that nobody ever picked East Russia then it's a divide-by-zero which I hereby deem an infinite lose rate - make sure you include that!

Every game where there's a wasteland in an otherwise top spot, you data is skewed. That's pretty much every game I play.

Yes, but it's all skewed evenly, so having enough data still makes the results interesting.

Also now that I think about it, I'm sure every bonus has been picked at least once. I think I may have even picked East Russia before, when the westernmost territory was the available one.

Every bonus has been started in at least 16 times (this is for the first 1187 games in the ladder). I could pretty easily exclude starting spots with a wasteland, since I am already explicitly tagging bonuses as "wastelanded" in my code. But I doubt the few mistakes will make any real change in the numbers.

Since start spots are random within bonuses, players also do not complete bonuses where an opponant starts next to them. That doesn't mean the bonus is not a top pick, just that you couldn't invest armies into taking the bonus while your opp is sitting next to you. This also happens most games.


It's just not that helpful to guid how you'd pick spots. I agree the results would be interesting.

I also think it's Scan/Antartica and Mexico. After that it's the choice 4s: East China/West Russia/E&W Africa.

If you ran this same test on default big world, warlord start spots it would be very predictive and valuable. The start spots are fixed and no wastelands.

This just has too much randomness in it to be predictive. That's what makes the games fun for repetitive play, but it makes data mining mostly unhelpful.

|>"Since start spots are random within bonuses, players also do not complete bonuses where an opponant starts next to them. That doesn't mean the bonus is not a top pick, just that you couldn't invest armies into taking the bonus while your opp is sitting next to you. This also happens most games."

If a there's a spot that would otherwise be powerful, but often has opponents starting right next to it, then that is just another factor affecting the value of the bonus. I don't know why we would want to exclude that information.

I think the wastelands will actually throw the chances of the higher territ bonuses down some, considering that those bonuses have a higher chance of having wastelands in them overall..

ofc' my vote is Mex-Ant-Scand for best, in that order, and worst Being Ant-ME-C.Rus

Leave it to Perrin to put the same bonus in both his "best" list and "worst" list.

I recorded the number of times each bonus was occupied after territory selection, and the number of times that the player won the game when choosing that territory.

Does "occupied" mean they completed the bonus or they picked the spot?

The suspense is killing me.

|>"Does "occupied" mean they completed the bonus or they picked the spot?"

It just means that they occupied a single territory in that bonus after territory selection. I didn't say "picked" because picking a territory does not ensure that you will receive it.

|>"I think the wastelands will actually throw the chances of the higher territ bonuses down some, considering that those bonuses have a higher chance of having wastelands in them overall.."

Since I am ranking by winning percentage *when occupied*, not overall winning percentage, this shouldn't matter.

|> Since I am ranking by winning percentage when occupied, not overall winning percentage, this shouldn't matter.

It should matter a little. The more territories in a bonus, the more likely it is to have a wasteland. Bonuses with wastelands are less likely to be picked. Therefore, bigger bonuses are more affected by wastelands than smaller ones.

Yes, but it will affect how often that bonus is picked -- but it will not affect winning percentage when that bonus is picked, which is what I am using as my ranking criteria.