StarCraft II: The Balancing Act

At BlizzCon 2010 some of our developers, including StarCraft II Game Director Dustin Browder, Battle.net Project Director Greg Canessa, Game Balance Designer David Kim, Senior Game Designer Josh Menke and Associate Game Balance Designer Matt Cooper participated in panels that answered a lot of burning questions that players have about StarCraft II, Battle.net and the ladder system, how to become a better player, and how matchmaking is balanced.  One of the questions that has emerged from these discussions involves the term 'adjusted win percentage', and we hope to shed a little light on just what that means here.  

One of the many tools that are used to assess balance in StarCraft II are the relative win rates for each race versus the others.  However, if the matchmaking system is working properly, then a player of any race should be so well matched versus their opponents that their win/loss ratio should be very close to 50%, even if the races weren't particularly well balanced.  In effect, the matchmaking system could be hiding balance issues -- if we didn't use an adjusted win percentage which takes the matchmaker's effects into account.  

What's an adjusted win percentage?  Well, while the math behind calculating an adjusted win percentage is extremely complex, an adjusted win percentage can be summed up as the 'true' win percentage of a given race, produced by removing the skewing effects of the matchmaker and factoring in player skill.  Combining information from millions of games being played, hundreds of thousands of players, their hidden skill rating, and a little math, we can generate reasonably accurate figures to compare how successful each race really is versus the others.

These stats can then be categorized by league, region, specific stretches of time, or any other way we might want to arrange the data for analysis. The end result is the information that we use (in combination with many other resources) to piece together a picture of what current StarCraft II balance looks like.

Now that you might have a clearer idea regarding what they are, we'd like to share some recent adjusted win percentages from November 11th for several regions around the world.  Before you review these numbers, we'd like to prevent some common misconceptions.  First, these numbers shift fairly rapidly as newly discovered strategies spread through the community and they're changing all the time.  Also,  due to the way the numbers shake out, we expect a variance of +/- 5% in these results; win/loss ratios that are +/-  5% suggest balance between those races.  So, if a win/loss ratio is approximately 55%:45%, then that's still within acceptable boundaries.  By contrast, win/loss ratios exceeding 60%:40% could indicate that a small imbalance might exist and merit further investigation.

In any case, these numbers aren't the last word in, nor the only component of, our balance analysis.  It's best to consider them a point of interest and one step along the path to fine tuning balance, rather than the final destination.  It's also interesting to note how the races fare versus one another from region to region as well.