Etta, I looked at your PBP from your last game. From your previous comments I infer that Thomas Morris is the 4th stringer in question, coming in at SG 4:22 into the game. Based on the PBP I'm also guessing your depth chart is something like:
SG - Houseknecht, Doe, Caudillo, Morris
SF - Lewis, Doe, Caudillo, Bellamy
PF - Moore, Bellamy (didn't note past Bellamy)
After a couple of minutes the game puts five subs in, putting in Doe at SG, Caudillo at SF and Bellamy at PF. So in this situation your third SG is in the game at another position.
You must have Caudillo targeted to play only a few minutes a game, so at the 4:22 mark the sim engine probably determined that he had enough minutes for the time being (I doubt it was fatigue-based given how long he was in and his high stamina rating). At that point, I guess Lewis wasn't ready to re-enter the game, so next on your SF depth chart would be Doe. In order to slide Doe to the 3, Morris had to enter as the SG. (Houseknecht probably wasn't fully rested). I may not have all of the details exactly right, but the scenario I have outlined above would be a specific instance of why the game engine might go to your 4th stringer so quickly. I agree with others that the sub model sometimes produces strange results - my explanation should not be interpreted as an endorsement of said process. However, with experience any coach can tweak their rosters so that this would happen less often (though probably not eliminate it entirely).
I also noticed that Morris came in again at SG very early into the second half, which I found more odd since there were people above him in the chart that were rested. I'm guessing that might have been a case where Morris may not have hit his minutes total in the first half and the game engine was trying to get some time for him early in the second to make up the deficit.