How does the winning percentage (white win/draw/black win) in chess 960 (Fischer random) compare to that of regular chess? Are there more white wins, fewer draws,...?
The CCRL 40/4 FRC engine competition that has run through almost 200,000 games running night and day come up with these stats:
Chess960 FRC based off 200,000 or so games: White wins: 80'914 (41.6%) Black wins: 70'840 (36.5%) Draws: 42'546 (21.9%) White score: 52.6% http://www.computerchess.org.uk/ccrl/404FRC/rating_list_all.html
Standard chess based off 1.4 million games: White wins: 547'582 (39.2%) Black wins: 447'252 (32.0%) Draws: 403'047 (28.8%) White score: 53.6%
As you can see, there are significantly fewer draws among the engines and slightly more black wins. These days, the top engines are playing the Chess960 opening fairly well so the stats aren't completely useless. As for top quality human games, there are less than 2000 on the records which is way too small a sample.
A rudimentary case study was carried out to find out if there were any really unfair start positions in Chess960, but none were found. Each of the 960 starts offers players something reasonable. Here is where it was done: http://chess960jungle.blogspot.com.au/search/label/SP%20analysis
I can't post more links as references because the site won't let me. The problem with Chess960 right now is that most of it is being played online at blitz time controls which is useless for statistics.