Scoresheet might occasionally make substitutions not the way you were counting on. It is a complicated process and there are a few quirks. Here are a few I have found:

1. Instead of filling a position with the top man on the bench, the computer first checks to see if the DH is listed at another position which can fill the open position. If so, it moves him there. Then after filling all the other positions, it takes the top pinch hitter remaining on the bench and makes him the DH. For example, let’s say  your outfield is Vernon Wells, TJ Matthews, and David DeJesus. Many Ramirez is your DH, Rocco Baldelli is listed no. 1 off your bench, Kevin Youkilis is your top ranking PH, and Joey Gathright is on your bench purely for defensive purposes – you hope. If DeJesus is injured, you want Baldelli to sub before Ramirez is moved to the outfield, but you would rather have Youkilis DHing with Ramirez in the outfield than have Gathright starting if both DeJesus and Matthews can’t play. So, you list Ramirez as DH-LF and on the bench list Youkilis behind Baldelli, but ahead of Gathright. Sorry, this won’t work. When one of your starting outfielders goes down, instead of taking Baldelli, Ramirez will go to the outfield before the system even looks at the top sub (Baldelli), then finds the top PH available (Youkilis) and puts him in the DH spot.

I haven’t tested this, but it should work if you listed both Ramirez and Gathright as DHs (keeping Gathright as your defensive sub). This will certainly result in Baldelli coming in DeJesus’s place instead of Ramirez-Youkilis. If Matthews is out, too, once the Scoresheet computer passes down the rest of the line-up bench and finds no more players listed in the outfield, it would be consistent if it then moved Ramirez to the outfield and put Youkilis at DH before it reached Gathright on the second pass looking for a player eligible in the outfield.

2. The Scoresheet computer cannot seem to handle too many fielding possibilities for checking the best alignment. Once I had Jose Valentin for a starting and defensive sub shortstop. He was listed at all his positions SS-3B-OF. Shane Halter was on the bench as SS-3B. He was my defensive sub thirdbaseman. Yet, when my starting thirdbaseman missed some games, Halter came in at shortstop and Valentin was moved to third. This was easily fixable by taking away Valentin`s -3B. However, if I had another shortstop who could not play third, I would miss that extra flexibility.

I would say Scoresheet should fix that by comparing the ranges of players before moving them  around like that – and, perhaps, they have. In the early 90s, my top pinch hitter and defensive substitute firstbaseman (John Jaha) came in to DH, while my starting firstbaseman (Jack Clark) remained at first. That particular quirk was fixed. Since then I noticed a case where the substitute firstbaseman with a superior range than my starter’s got the start at first while the starting firstbaseman was moved to DH.

3. Remember, unless you specify otherwise, your top ranged outfielder will be placed in centerfield and your worse ranged outfielder will be placed in leftfield. So, if you list a better ranged outfielder as your defensive sub in leftfield than the outfielder you list as the defensive replacement in rightfield, and they are both playing, they will switch positions at the time your defensive substitutes come in.

Dave Barton, March 20, 2007 in Scoresheet-talk:

Our rationale is that if the world's fastest man who was also great at SS tried an entirely new position, it's unlikely he'd be immediately better than the average major leaguer at that position, since that's a very high standard. He wouldn't have years of experience turning the pivot at 2B, picking bad throws at 1B, reading bounces off the walls in LF and RF, knowing instantly whether to move forward or backward in the OF when the ball leaves the bat, etc. In CF there's more room to use his speed to compensate than in LF or RF, so I can believe he'd be better than 2.07 there but not in LF or RF. Fielding range is all approximate anyway. I agree we could have just used 2.10 as a cap at all three positions, but actually I think the way I programmed it is a little more accurate, even though it's also a little more complicated.