Players perform better or worse than their actual stats depending on how the year they are coming from stacks up against the average year. Hitters from 1927 are going to perform worse, because hitters dominated that year. Hitters from 1918 are going to preform better, because hitters sucked snowballs that year.
League average hitters from those year don't perform quite the same though. A league average hitter normalizing downward from 1927, will perform a bit better than a league average hitter normalizing upward from 1918. They seem to take into acount RL stats as well to some degree. Their exact method of normalization is hard to put your finger on (I've never read where anybody had it down pat), but you can get the general idea pretty easy.
The 'Performance History' link we now have for every player, is an excellent tool for predicting what you will get from a player, especially hitters. Someone will come along and contradict this statement, but you can take my word for it, its quite accurate. A relief pitcher with 8 uses, including one where some bozo let him get a 27.63 ERA, will of course be skewed, but other than that it is there for you to take advantage of.