Yep, what schwarze said is right. There is a great post by elbirdo from several years ago that tells you have to convert real life stats into number of SIM allocated pitches. I've bumped it again in the 'Sim Baseball' forum so it is easy to find. The short version is
- Pitcher fatigue is calculated by pitches thrown in SIM, not innings pitched. (So in that respect it is very different to hitter fatigue)
- How many pitches a pitcher is allocated, before they fatigue, is a function both of their IP, and of their component stats.
- The more H/IP, K/IP and BB/IP a pitcher has in real life (not normalised, just real life), the more pitches/IP they are allocated in SIM.
- And the factor for K's makes a big difference.
The actual formula for allocated pitches per IP/162 is : 3.406*BFIP + 3.762*BBIP + 1.964*KIP, where BF is Batters Faced, i.e., WHIP + 3.
So compare these two pitchers:
A has 200 real life IP, a WHIP of 1, 0.2 BB/9, and 0.4 K/9.
B has 200 real life IP, a WHIP of 1, 0.2 BB/9 and 1 K/9.
Pitcher A will get allocated 3032 pitches, pitcher B will get allocated 3268. If you have 8 of pitcher A, you will get 24259 pitches, or 149/game. If you have 8 of pitcher B, you will get 26144 pitches, or 161/game. The 8 B's will cover how the leagues are playing so far; the 8 of A will be getting fatigued.
The complication is that the B's should be throwing more pitches in the SIM than the A's. But I don't think they throw much more. Roughly, you allocated get 2 extra pitches per K, but you only end up using 1-1.5 of them (given the low K hitters that usually play in the SIM). So there is a bit of sneaky value in high K pitches; this makes the common all deadball strategy a little less good than is sometimes thought. But that's a separate matter, and I'm not 100% sure that what I've said in this paragraph is all true.
And all this is really owing to elbirdo's great work - apart from the guesses in the last paragraph I'm just reporting what he found out.