That's right, it is impossible to actually get your 7th pick in a 1v1 game with 3 starts. If both players pick the same 6 in order one of them will get 145 and the other will get 236.

(Also impossible to get your 6th AND either your 4th or your 5th.)

I think he said something about the player who gets 1st pick wins 10% more games.

Was this in games where both players picked the same pick as 1 or in all games? I would suspect the former.

You know, this data could totally change Warlight ladders/CLOTs. Bayeselo (which Fizzer uses or the ladders) does adjust for first-pick advantage using an experimental constant, but the one Coulon (maker of Bayeselo) derived came from chess- and I have no idea how Fizzer got his experimental constant.

So smileyleg's data can be used to get a good constant for first pick advantage. But it gets even better. Now we're even able to adjust for very specific situations (with more data, perhaps)- like getting 1,4,7 when your opponent gets 2,3,5. Or derive different constants for first-pick advantage depending on whether you had the same first pick.

If we had probabilities like that and were able to determine which changes were neglible and then go ahead and derive constants from the ones that weren't, well, that would totally change every CLOT out there and make the results much more accurate in the long run.

DanWL, how in hell do you think there could be a possibility to get the 7th pick when the total territory limit is 6?

Edit: When I had to use 6 of my picks, my first 3 picks were all gone and I was left with picks 4-6.

bullshit, you ALWAYS get at least 1 in your first 2 picks and if you get your 6th pick you DIDN'T get your 4th and 5th

Edited 6/25/2015 19:13:04

@Dan, with 3 picks per player it is impossible to get 7th. Don't attempt to argue this one; just accept it as fact, please.

bullshit, you ALWAYS get at least 1 in your first 2 picks and if you get your 6th pick you DIDN'T get your 4th and 5th

Is there a non-cyclic way to handle picks? I.e., a purely random one where the system decides between P1 and P2 for each picking turn until one of them has 3 starts.

Obviously not on the Strategic 1v1 template or the 1v1 ladder, but does one exist? I haven't seen it used, ever.

EDIT: to clarify, not that I would ever want to use it

Edited 6/29/2015 07:16:04

"a purely random one where the system decides between P1 and P2 for each picking turn until one of them has 3 starts."

Well, there is automatic distribution. and that is the only option to manual :)

Is there a non-cyclic way to handle picks?

There isn't. Even with random move order, once the game has decided priority the turn will happen in this fashion:
P1, P2, ... , Pn-1, Pn, Pn, Pn-1, ... P2, P1, P1, P2, ...

in a 1v1, it is always
P1, P2, P2, P1, P1, P2, ...

Ah. I almost want to try that, just to see how things would change now that I can no longer predict my pick combos (or at least rule out impossible ones like 1-2-6).

(or at least rule out impossible ones like 1-2-6)

But 1-2-6 is possible.

smileyleg how you collect all of this information?

My guess is he just runs a program that checks the data for each of the 8million games, and saves the data from any ladder games. From what I've seen it's a number around 100,000.

That would be my guess anyway :)

Dublin Warrior: It's the map "Medium Earth" and it starts with 7 wastelands. It's the standard set up for 1v1 Ladder games :)

Here's an example game so you can see the starts: https://www.warlight.net/MultiPlayer?GameID=8932688

1,2,6 is a way to easily determine gg.

Either you know you've won or you know you'll lose.

Fleecemaster- it's a bit simpler than that.

WL has a Game ID feed API that gives you the game IDs of any tournament or ladder. Once you get those, you can use the Game Data API to get data from each game (which is *really* comprehensive if you set a parameter called "GetHistory" to true). You can use that to perform all sorts of analyses on the game- what smiley's done so far is just the tip of the iceberg, and I suspect he's got a lot more coming. You can get the team IDs of players, their income each turn, their territories each turn, and even have enough data to reconstruct the map (at least the territory connections- you won't get the images).

Edited 6/30/2015 09:34:26

Right, pretty much exactly what knyte said.

I pull the data into a SQL database and then run queries to produce the analyses.

Oh yeah, I didn't realise there was a list for the IDs, nice :)

Seems you're the right people to ask a question I have then...

Is it possible to pull the history from a game in progress?

I can't find the way to do it, but I was hoping maybe if I can query with my username or something I can get access to that data? Seems it's set up to do it, I think I just don't have the right way? If you can just point me in the right direction or tell me it can't be done that would be great :)

Fleecemaster, try this:

https://www.warlight.net/API/GameFeed?GameID=XXXXXXX&GetHistory=True

Where Game ID is changed to the ID of the game you want to query. Just go to that URL and you can test it. I know it doesn't let you get data from games you weren't a part of but since I don't have API powers (non-Member) I have limited knowledge of the possible outputs of calls made to the API.

I suspect it will be disabled, though, or at least limited by fog- otherwise this would be a pretty easy way to cheat and gain an unfair advantage.


Source:
https://www.warlight.net/wiki/Query_game_API

Edited 6/30/2015 17:49:02

That doesn't give me the history of the game, it only gives me the starting info, the player IDs and such. I feel like I'm maybe missing a command that will give me the turn info of a game while it's running. At least, I hope it's possible...!

Is there somewhere that has all the possible calls that I've not come across?

That URL works fine once the game has completed though, of course :)

So far, the only documentation I've found has been on the wiki and it's been pretty sparse. I don't think there's any parameters you can pass in on that request other than GameID and GetHistory, though.