<div>
(colby2152 @ May 22 2008,8:55)</div><div id="QUOTEHEAD">QUOTE</div><div id="QUOTE">Have you assumed log-normal, polynomial, exponential or power from another base?</div>
For one exercise, I tried taking three data points and deriving an exponential function from those three. Didn't really come out that well either. My guess as to why it worked so bad is that I did two of the sets at the same workout, so the results from the second set were likely skewed by the fatigue from the first set, hence rendering the function useless.
Anyway, since there are so many factors in play, it's probably easier to just take the maxes the hard way - take a week at the end of the cycle to check how much you can actually lift for every rep range.