Alex: Data Please!

[Syvan]

(08-01-2012 08:58 AM)stevewastaken Wrote:  I was a mouse click away from creating a thread to request a sort of Ultimate League with only 1 division, and players are ranked by their hidden matchmaking rating, rather than points. (though it could be masked somehow to look like points, even though it isn't. thus avoiding confusion from the noobs).

Then I figured there are much bigger issues at hand, and I would save the suggestion for later.

In Starcraft 2, MMR can be fairly accurately inferred at the highest levels by comparing ladder points and the available bonus pool. If we get a visible and limited bonus pool in a future season, we can quickly enough compare skill ranks.

[CombatEX]

(08-01-2012 10:54 AM)MattPowell Wrote:  Eh. I don't see any reason to get tied up in knots over this. You move second 50% of the time in matchmaking play and any community tournament should be designed to balance who plays first.

Imagine if you flipped a coin in sports to decide which team would have a disadvantage over the other. 50% of the time you have the advantage, 50%, a disadvantage.

Say in basketball 50% of the games you play you don't have a backboard.
At a track meet each contestant in the '100 yard' event has a 50% that they actually have to run 110 yards instead.
In the NFL each game one of the teams gets to start with 3 points.

etc.

Sure, it evens out in the long run, but it's not good for any competitive event to have a significant advantage given to one of the teams/competitors.

Maybe FTA isn't a huge problem, but the point of this thread is to see if it is (and if it's a major problem on any of the maps).

As for the super 1 division league (like SC2 Grandmasters), I'd like to see this. Players can still be ranked by points (again like SC2) as when a bonus pool is implemented (Alex mentioned he's tentatively planning on adding seasons and bp in the future) points will be fairly indicative of MMR at the highest levels. The reason I ask for this extremely select league for the top ~100-200 players is to give Super-Titan members something to fight for. Without a hardcap on players, Super-Titan, while a good goal, will only ever have the prestige of something like SC2's Masters league. It's cool being in the top ~2% of players but 2% is still a huge range from rank 1 to the bottom of the 2%. I'm mid-masters league in Starcraft, but I'm *TERRIBLE* compared to Grandmaster players. The same will surely apply to Outwitters as more and more players join and Super-Titan expands in population.

I was thinking Super-Titan can be renamed 'Titan' while this elite hardcapped league can be named 'Super-Titan'. I've always found it weird that we have a *Super*-Titan when there isn't even a normal Titan =P

Anyway, this is a 'way down the road' request =) No rush. Just hope this sticks in the back of Alex and Adam's minds as something to do implement in the far future.

[MattPowell]

(08-01-2012 12:50 PM)CombatEX Wrote:  

(08-01-2012 10:54 AM)MattPowell Wrote:  Eh. I don't see any reason to get tied up in knots over this. You move second 50% of the time in matchmaking play and any community tournament should be designed to balance who plays first.

Imagine if you flipped a coin in sports to decide which team would have a disadvantage over the other. 50% of the time you have the advantage, 50%, a disadvantage.

Say in basketball 50% of the games you play you don't have a backboard.
At a track meet each contestant in the '100 yard' event has a 50% that they actually have to run 110 yards instead.
In the NFL each game one of the teams gets to start with 3 points.

etc.

Sure, it evens out in the long run, but it's not good for any competitive event to have a significant advantage given to one of the teams/competitors.

Maybe FTA isn't a huge problem, but the point of this thread is to see if it is (and if it's a major problem on any of the maps).

As for the super 1 division league (like SC2 Grandmasters), I'd like to see this. Players can still be ranked by points (again like SC2) as when a bonus pool is implemented (Alex mentioned he's tentatively planning on adding seasons and bp in the future) points will be fairly indicative of MMR at the highest levels. The reason I ask for this extremely select league for the top ~100-200 players is to give Super-Titan members something to fight for. Without a hardcap on players, Super-Titan, while a good goal, will only ever have the prestige of something like SC2's Masters league. It's cool being in the top ~2% of players but 2% is still a huge range from rank 1 to the bottom of the 2%. I'm mid-masters league in Starcraft, but I'm *TERRIBLE* compared to Grandmaster players. The same will surely apply to Outwitters as more and more players join and Super-Titan expands in population.

I was thinking Super-Titan can be renamed 'Titan' while this elite hardcapped league can be named 'Super-Titan'. I've always found it weird that we have a *Super*-Titan when there isn't even a normal Titan =P

Anyway, this is a 'way down the road' request =) No rush. Just hope this sticks in the back of Alex and Adam's minds as something to do implement in the far future.

Let's say when facing a player named Bob I have an 80% chance of winning when playing first and 40% going second. That seems like a pretty severe FTA.

Best 2/3, me first in two: I have a 76.8% chance
Best 2/3, Bob first in two: I have a 54.4%
Best 3/5, me first in three: I have a 77.3% chance
Best 3/5, Bob first in three: I have a 61.7% chance

That's with a large asymmetry. If it's smaller they'll converge faster. I don't see this being an issue unless you want individual matches to be ble to hang your hat on the outcome of a single match, but I don't think that can ever be a reasonable expectation.

[onealexleft]

(07-31-2012 01:20 AM)ArtNJ Wrote:  Alex,

So the community can more intelligently talk about the first turn advantage, can we get some data? Specifically:

1. Are you able to look at the % of time player 1 wins the game across all League games?

In beta yes, right now in production no. But our analysis tools need to be re-worked as the scale of data is huge and none of our current tools work. I'm really hoping to get back to this, but with more pressing issues of crashes, non-expiring games, and league matches not counting, it's at the bottom of the list right now.

(07-31-2012 01:20 AM)ArtNJ Wrote:  2. Are you able to look at the % of time player 1 wins in Super Titan v Super Titan match ups?

Answer to #1 applies here.

(07-31-2012 01:20 AM)ArtNJ Wrote:  3. If you are able to look at this by map as well, does it look like any maps have a higher percentage of player 1 wins?

Our initial tools weren't that specific. But I'll keep that in mind when re-implementing the analytics. Breakdown by maps/leagues will be helpful in pinpointing issues.

(07-31-2012 01:20 AM)ArtNJ Wrote:  Also, switching sub-topics, are you able and willing to share the top 50 players by hidden matchmaking rating? I think that a small list like that would let hardcore players compete, but not discourage the casuals.

This I *can* do, and it's ready to go. I've just been sitting on it to let the leagues settle out a bit. Look for it in the news feed and forums soon.