Where do you get the data from?
The data we use to make our projections is of public record. It comes from various sources, such as ESPN, the NFL, and so on. The projections themselves and the technology to make those predictions are a series of proprietary technologies that were custom built for this site.
You said Player X was going to score 12 points, and he only scored 8! What happened?
We would like to be as accurate as possible, trying to give you every advantage in winning your league. However, there are always occasions that are out of the control of any predictive model (poor weather, an injury, etc.) and of course, any performance is subject to variation due to the mechanics of the game. We do the best that we can to provide an accurate projection, but we're not divine superpowers.
Yahoo! said that Player X was going to score 12 points, you said 6, and they were closer! What happened?
Again, due to the variations of game performance, we sometimes are off with our predictions and other models, scientific or not, end up being closer. With that said, we strive to be closer over the long-haul, over many weeks and many players, and we are confident that are analytical, data-driven approach is the best way to guarantee that.
When you say a player is a 87.3% match for another player, what does that mean?
On the player level, a match means that Player A is similar to Player B based on his statistics, his style of play, and various other factors of similarity. On a game level, we compare both teams to other teams throughout history for similar styles of offense, defense, and game strategy. The predictive matching is the union between those two sets; similar players put into similar situations featuring similar teams.
Why do some players have predictions, and some not?
Some players may not have enough historical similarities for us to make a qualified prediction. This is often the case with players who are third or fourth-string.
Who is going to win the Super Bowl in the 2012-2013 season?
The Pittsburgh Steelers.