I have been thinking about fatigue levels lately (not enough to do?) and I want to throw out my thoughts and see if they appear to reflect reality at all.  I have a team with a bench I rather like.  As such I have my pinch hitter/defensive replacements/pinch runner settings turned up so that they occur at the drop of a hat. . .any hat.  Also, I have been trying to just flat out rest one starter a game.  My hypothesis is that though it may say "100", clearly not all 100's are created equal.  Rather, each player starts out at a number higher than 100, but that 100 is the highest number than can be displayed.  Every at bat or inning of defense eats in to number on a fractional basis and every inning they are not participating increases that number.  Therefore, an inning or a game off here or there slows down the speed by which the player reaches that 100 and starts to fall below it.  Therefore, the more they are rested early in the season, they more they will be able to be used late in the season where some owners have lost interest in winning.  This, combined with playing against players more fatigued then themselves will make long winning streaks more likely. 

My 88-74 finished the season 60-30 last season using this strategy, but I couldn't have a smaller sample size than 1 season.

Am I thinking about it in the right way or am I resting players before they reach 99 and below for nothing?

I use the same strategy you state where I have one "Super Sub" rotate through all position but C & SS. When I have an injury & the Supersub stops rotating I notice that players fall below 100% much sooner. So I think early rest delays the below 100% drop

Posted by boogerlips on 7/16/2011 2:11:00 PM (view original):
Just to be clear, fatigue is based on PAs for hitters and Pitches Thrown for pitchers.

I've never bothered to look too hard into how durability translates into PAs at 100% a guy can have in a season, but you should be able to figure it out easy enough if you really want to. Whats possible is that you could be substituting too much for one particular player when he could have played more...

Posted by csherwood on 7/18/2011 2:13:00 PM (view original):
I have never noticed any significant disadvantage for playing guys with their percentages all the way down to around 90 (position players).  I do rest guys who fall below 98 or 97 during the season, but if I am in the stretch run and in a playoff battle, I play them until their arms and legs fall off.  I usually mark Roster Expansion Day as the day in which fatigue starts getting ignored if I am in a battle.

This makes sense.  Say a guy is .300 hitter.  If he's at 99% I would still expect him to hit .297 (.99*.3) which is 2 hits over entire season.  Basically indistinguishable.