Data Mining for Ratings

Recently we've been randomly sampling race results from various serious from TORA and TURN, and you can also help us by either posting your results, or pointing us to the site that has them posted, and we will have the computer rate those drivers, and incorporate their ratings with the world rankings list.

Very soon there will be a massive increase in the total number of rated drivers on the rankings list.  You will see many familiar names appear on the list that haven't been on the list before. Keep in mind that until a driver reaches at least 10 rated races his rating and therefore world ranking is only provisional.  It takes at least 10-15 clean rated races in a row for a driver rating to be relatively accurate.  This is because lag outs and unforeseen crashes can skew the ratings.  This in turn means that occasionally you will have a high skill driver who may be under rated, or a low skill driver over rated.

When there are a lot of provisionally rated drivers (less than 10 races) in a given lobby with a high rating, that is when an established rated driver can get some easy points from them.  This is because when you are first rated your ratings tend to be inflated by around 4%.  This is on purpose and is done, so that drivers are not made to feel bad about their initial world ranking, and so that the veterans can have some movement with their ratings.

We will be spot rating the "Forza Championship" results as they begin the quarter finals too.  These are one vs one match races best of 3 contest and will be a true tale of the tap when it comes to skill.

So if you are in a series and you'd like to get it rated, just send me a PM or post it here and I will plug the results into the ratings computer.

AAR GTDon