Calculate starting pitching fatigue?

I am trying to adjust card readings for my starting pitchers in an effort to figure out what "dont relieve before" F value I should use. After reading through the Advanced Game pitch count fatigue rules, I appear to have learned that this all happens in a black box and my attempt is essentially an exercise in futility.

If I understand correctly, there is a pitch count set at the beginning of the game. We dont know what this is, but some in-game algorithm combines this initial pitch count before fatigue would set in, the "success" of the pitcher during the game, and the actual in-game pitch count to adjust the F levels. As the pitcher gets fatigued, when a "fatigue roll" comes up (noted by @) there is some other algorithm which then takes that roll and figures out IF it will be a hit, and if it is a hit, it might not just a single but could be extra bases.

Do I have this right? Is there any way to know the odds of a fatigue roll turning out with a particular outcome before the fact, or am I at the mercy of Hal's black box?

It's a part of Hal's black box. Just like real baseball, the manager is never sure how much the pitcher has left in the tank.

If I understand correctly, there is a pitch count set at the beginning of the game. We dont know what this is, but some in-game algorithm combines this initial pitch count before fatigue would set in, the "success" of the pitcher during the game, and the actual in-game pitch count to adjust the F levels. As the pitcher gets fatigued, when a "fatigue roll" comes up (noted by @) there is some other algorithm which then takes that roll and figures out IF it will be a hit, and if it is a hit, it might not just a single but could be extra bases.

Do I have this right? Is there any way to know the odds of a fatigue roll turning out with a particular outcome before the fact, or am I at the mercy of Hal's black box?

What you get wrong is that the "@" is not involved in the SADV (super-advance) pitch count, only in regular face-to-face. Instead, the balance of readings get shifted towards the hitter's card. At F8, it's subtil, perhaps a 3% shift, but at F0, it's almost 66/33 and can get worse.The way I see it, the algorithm probably involves a 1-100 number generator, and if you deal say, a 1-2-3 result at F8 , the result will be read on the offensive card, and the lower the F, the greater the probability of reading the offensive card. As you say, the consequence is that it's not only chances of singles that get raised, but doubles and homeruns too, as found on the offensive card. But I do speculate a bit on the way Strat does it, as you say, it's pretty much a black box of which we can interpret only the results.

What you get wrong is that the "@" is not involved in the SADV (super-advance) pitch count, only in regular face-to-face. Instead, the balance of readings get shifted towards the hitter's card. At F8, it's subtil, perhaps a 3% shift, but at F0, it's almost 66/33 and can get worse.The way I see it, the algorithm probably involves a 1-100 number generator, and if you deal say, a 1-2-3 result at F8 , the result will be read on the offensive card, and the lower the F, the greater the probability of reading the offensive card. As you say, the consequence is that it's not only chances of singles that get raised, but doubles and homeruns too, as found on the offensive card. But I do speculate a bit on the way Strat does it, as you say, it's pretty much a black box of which we can interpret only the results.

That's very interesting, and as you say that I suspect that you are right. Because of this, you would expect a very small but persistent bias towards the hitters rolls. It might not even be noticeable for a single season, but over time there should be a small shift. I always look at the game as a 50/50 thing, but as I learn more I am beginning to feel that it is much more complicated than that.

To get back to my original question/objective though, there really is no way to determine a priori when my starting pitcher will be less effective than my relief pitcher?

It's kind of a "feel" thing. I weigh how good the starter is and how good my bullpen is. I've set stud starters as low as F1. I'd say the most common setting is probably F7

I have played around with F settings a good bit since my first on-line teams (April 2013)

This is my assumptions of how the F-Ratings work on-line, through experience and other long time players threads, though no one knows for sure.

I could easily be mistaken on some of my assumptions though, so anyone that disagrees feel free to chime in.

1st --- ignore the @ ratings the board game uses completely.

2nd -- As the pitcher tires--the F-rating will drop from F9 to F0, but that's not the end of it, if a pitcher stays in at F0 he actually gets worse as the at bats mount--though you can't see it in the play by play, the F ratings kind of continue into negative numbers.

3rd --- If a pitcher starts to tire there is a chance the original dice roll will be changed (transparent to the replay)
at F8 it's a very slight chance---with the first batter at F0 I think it is a 50-50 chance the roll will be changed, and it gets worse from there with each batter to a near 100% chance (probably 95%).

4th -- when the roll is changed, it changes the single die to the best column for that at bat, regardless of whether it is on the hitter or pitcher card. (poor hitters get the worst column on the pitcher) I don't know whether the 2-12 two die combo is changed--I suspect it isn't, but I have seen poor hitters find the lone HR on the pitchers card when he is tired, even if that is not the pitchers worst column.

5th -- F3 seems to be about the equivalent as the old @ tiring rule OPS wise, though with the online rules it's not just singles.

6th -- F settings can result in more rolls on the hitters card, due the single die change, though it isn't that noticeable until about F3. F1 and F0 set pitchers almost always result in more rolls on the hitters card (assuming the hitter has better chances on his card than the pitchers card)

I would say Miller has it about right except for the last point (6). I haven't seen any evidence to suggest that F3 is the "magic barrier" or anything like that. I think it's a more steady slope from F8 to F0 and then of course F0 just gets worse and worse

I set my ace starters to F8. All other starters I leave blank.
I set my best reliever to F8.
I also set my (cheap) backup closer to F8 because too often I've seen HAL pull him after one batter to bring in my fatigued regular closer.

Everyone else I leave blank.

Truthfully, whether you set a pitcher to F8 or say F4 doesn't make that much difference (usually one batter) because once a pitcher's endurance starts to drop, it goes quickly. However, it makes a huge difference between no setting and F8.

I tend to take my aces a bit lower depending on how good my bullpen is, usually F6 or 7. This team I am considering F5 as I went with four stud starters and essentially a 1 man bullpen. That's why I was hoping to get more quantitative.

http://onlinegames.strat-o-matic.com/team/1412252

In a 80M league, with 3 SP having a 9-inning stamina, I even wonder whether you shouldn't go lower in the F-rating and spend otherwise the 1.77 on your lone reliever.

I kinda agree with Milleram that F3 probably represents an impact similar to the "traditional" rule of "swithching a @ for a single**. I think this is so because when a fresh reliever or a fresh starter allows 3 consecutive hits, the rating falls from F9 to F3, and then gets lower. So I suspect that the algorithm used provides at F3 roughly 11 additional on-base to the hitters (as milleram says, instead of the old rule where the 11 on-base chances added were singles**, in SADV, they are a mix of walks , si, do, triples, and homeruns---if you want quantitative numbers, I would guessestimate that a fatigued F3 pitcher will allow against a typical hitter an additional 2.5 walks, 5 singles, 1.75 double, 0.25 triple, and 1.5 homeruns.

I would definitely avoid the F0 state though, again as milleram says, because the readings kinda cluster on the worst column (I don't think the algorithm "switches" to the worst column, contrary to what milleram's #4 suggests, I think it's only a consequence of the algorithm forcing an allowed hit--hence the readings get clustered to the worst columns---but if I am right, I would predict that the outs are still distributed roughly evenly across the columns--or at least out not coming from DEF-X).

So if I go with only studs in the starting rotation and busts in the bullpen, I would probably go with a F1-quick hook for my starters.