Why are Starcraft 2 tournaments so unpredictable? - Page 15

On February 04 2012 04:20 Primadog wrote:
Eight years and 27 (combined) seasons later, Liquibet remains a cherished tradition amongst the best oddmakers of TeamLiquid. Each Liquibet season, a champion is crowned by picking the most winners out of upwards of 500 matches played that season. Despite format changes, metagame shifts, and racial imbalance claims, winning percentage has remained surprisingly consistent:



It's a reminder that there's no sure things in StarCraft (BW or the sequel), an attribute of the game that made it a serious spectator sport.

On February 04 2012 05:20 Itsmedudeman wrote:

I'm not gonna argue one way or another, but SC2 liquibet is far safer in my experience due to the level of competition in foreign tournaments. When you have A team koreans going against B team foreigners in NASL and other foreign tournys it's pretty easy to predict. GSL is far more difficult to predict when it comes to discerning between great players, and BW typically only has matches between the real top players in the world.

On February 03 2012 21:01 Teddyman wrote:
Let's look at unpredictability by analyzing how well viewers actually predict the results, shall we? Fortunately we have a lot of data in the form of past Liquibets. The hypothesis would be that if SC2 really is more unpredictable in ways that BW or RTS games in general are not, the rate that one predicts the result correctly would be smaller. If the game was totally random and not skill-based the rate would be 50%. So I took the average of the Liquibet points percentage of the top 10 players in some liquibet seasons and charted them against the maximum number of points that were available that season. The logic here being that a smaller amount of maximum points means less games to predict and more chances that someone gets a high number of predictions correct with luck. I included the last 4 seasons of Liquibet for BW and SC2 and also seasons 14 and 16 for BW to get a season with a very low and a season with a very high number of games. Here are the results.

You could say that this points towards BW being more unpredictable that SC2 but the sample size isn't huge. A lot of assumptions are made that aren't necessarily true: that people are as proficient in predicting both games, only looking at the top 10 for a season, and looking at points predicted instead of matches. It doesn't really prove anything but does give some clue that we aren't playing rock-paper-scissors here.

EDIT: Another factor to take into consideration would be that there are more people predicting SC2 matches, resulting in a higher chance of someone getting lucky with their predictions. This could be examined by checking whether the same people are placing high every Liquibet or if it's always changing. Anyway, 2^8th post bitches!

On February 04 2012 05:40 IseeFish wrote:


Did you take the top 10 in the Liquibet ranking? If so, your results are going to be skewed because of the different number of liquibet participants. As you mentioned in your edit, there are more people predicting SC2 matches, so more will get lucky. If you have enough people tossing coins, you will find ten who get heads a hundred times. So you need to either analyze the whole Liquibet dataset or choose a fixed percentage.

On February 04 2012 05:20 Itsmedudeman wrote:

I'm not gonna argue one way or another, but SC2 liquibet is far safer in my experience due to the level of competition in foreign tournaments. When you have A team koreans going against B team foreigners in NASL and other foreign tournys it's pretty easy to predict. GSL is far more difficult to predict when it comes to discerning between great players, and BW typically only has matches between the real top players in the world.

On February 04 2012 06:09 Primadog wrote:
Can you link to threads discussing this classifications system for us? Very curious on how things were done in BW.

On February 04 2012 06:13 MCDayC wrote:
There is zero statistical evidence for SC2 being more volatile. I know that statistical evidence is not everything, but when almost every conceivable measure of volatility (Liquibet for peoples ability to predict, Winrates for players consistency, MVP getting top 4 in every tournament since MLG Anaheim bar AOL 1 and the latest GSL) is put up against... feelings? It's seems like this is a ridiculous arguemtn

On February 04 2012 06:28 StarStruck wrote:


Looking at the Liquibet isn't a good measure as a few people pointed out already and like I said earlier. We're already seeing plenty of division between the pro's in SC2 as well. There's a wide range of them.

On February 04 2012 07:03 EienShinwa wrote:
Because the game is still very young. People make new builds, try new things, and nothing is set in stone. We see pros floating into the 4 digit numbers in games and meta-late game builds are not really the best or concrete in stone. We just need more time is all.

On February 04 2012 07:00 MattBarry wrote:
Rofl at all the people calling sc2 terrible. If we accept bw as the greatest rts, then it's likelyyoull agree that sc2 is the second best. If the second beat rts evr is shit then the rts genre as a whole is shit. Youcouldsay "it's disappointing that it didn't quite live up to it's predecessor" but it's audacious and downright stupid to call it shit.