Super Hal and Fatigue - The Mystery Continues

[apolivka]

I think it is fairly clear (in a fuzzy way :) ) that @results off a tired or tiring pitcher are NOT automagically "re-rolled" by HAL. We had previous threads about @results _sometimes_ being hits and _sometimes_ being outs which lead to various arguments on that topic.

I'm not sure how I "feel" on this one. I get the feeling that they attempted to create a very realistic gradual yet somewhat random algorithm for pitchers to get worse as the game wears on based on MANY factors (pitch count, POW, previous hitters results, F-Number, etc.) that are probably very hard to explain in 4 or 5 bullet points on a help page. Would I be happier with a much simpler system? I'm not so sure. If Bernie published the, say 50 box flowchart with what "really happens", I'm sure every result that some people didn't want would be endlessly complained about anyway!

A little ambiguity in the game, in some ways, makes it more interesting. I'm interested, and will continue to study the issue and hopefully use it to more expertly set my pitcher's preferences. But, we just got super HAL bullpen logic, so more research is certainly needed.

Based on what I have seen so far F0 and F1 pitchers are REALLY bad.

[Detroit-Tigers]

[quote:485d74e920="The Biomechanical Man"]I wish Bernie would just tell us what is happening. I'm sure he must have learned something when he coded bullpen version2.

It is very helpful that the F-value is now shown on the boxscores. However, when a pitcher fatigues we don't know if:[list:485d74e920]
[*:485d74e920]the chance of @ becoming singles gradually increases (as stated in the [url=http://somonline.bluwiki.com/go/Pitch_Count_Fatigue]Rules[/url] on the online game)
[*:485d74e920]the chance of landing on a pitcher's "bad column" gradually increases (suggested by Marcus Wilby [url=http://forums.sportingnews.com/viewtopic.php?t=563603]here[/url])
[*:485d74e920]the chance of landing on the hitter's card gradually increases (suggested in this current thread)
[*:485d74e920]some combination of the above
[*:485d74e920]something totally else happens[/list:u:485d74e920][/quote:485d74e920]
The reason he can't is because it starts getting close to SOM proprietary information (i.e., how the game engine works). If you can figure it out, he hasn't given anything away.

[MARCPELLETIER]

[quote:bd5628aab4]
Marcus,

Under your hypothesis, a pitcher should never be able to get anyone out when at F0. Have you tested that hypothesis? [/quote:bd5628aab4]

Actually, under my hypothesis, a pitcher should never be able to get anyone out ON HIS CARD (DEF-X excluded, and outs on hitters card continue to be available), and this ONLY AFTER a pitcher gets significantly WORSE than the initial F0 (remember than the downward spiral continues even though a pitcher has a F0 rating, at initial F0, I estimate that only 10%-15% of outs become hits).

And yes, I tested that hypothesis specifically. With the CD-ROM, I played 4 games in a row during which I yanked my starting pitcher for Papelbon after the first out of the first inning, and would leave Papelbon there for the right of the game. In most games, at some point, usually around the sixth inning, Papelbon would not get anymore outs on his card, except for DEF-X. First two games were lost by scores like 21-4 and 26-5.

Funny enough, in the third consecutive game, Papelbon pitched a gem. He had a one or two-hitter up until the 7th inning, and started to be significatively worse only during the 8th inning, and still got the win if I remember correctly.

[MARCPELLETIER]

[quote:83a03af62b]"PCF affects readings on pitcher cards preceded by an "@", but it doesn't automatically turn them all into SINGLE**. Instead, it is a more gradual effect, and the variety of the resulting penalties is greater."[/quote:83a03af62b]

btw the lines of this quote, here is what I read: The @->single** doesn't apply anymore. There is another rule, underwhich a @ can become single**, but it's not automatic anymore. With this other rule, @ can also become walks, single*, double, even homeruns. The variety of the resulting penalties coming from that rule is greater. And it's more gradual because at F8, only 10% of @ are affected by this rule, at F5, 40% of @ are affected by this new rule, and it's scaled like this until F0. I'm not sure about those numbers, of course, but the numbers are probably in the ballpark.

I fully admit it's an interpretation of my own, but I see it coherent with was Bernie posted there.

BTW, if I were correct, the overall impact on pitching would be roughly 1% at F8, 5% at F5, and roughly 10% at initial state of F0. I think these numbers are in line with the little experiments done here.

[MARCPELLETIER]

[quote:cf7bb1130f]I think it is fairly clear (in a fuzzy way Smile ) that @results off a tired or tiring pitcher are NOT automagically "re-rolled" by HAL. We had previous threads about @results _sometimes_ being hits and _sometimes_ being outs which lead to various arguments on that topic. [/quote:cf7bb1130f]

Exactly. Under my hypothesis, at F8, only 10% of @ are "re-rolled".

But at the lowest possible state, an extremely-abused pitcher, I believe all outs on the pitcher's card, except for DEF_X, are re-rolled.

[The Biomechanical Man]

If, indeed, fatigue means more-and-more rolls on the hitters' cards and less rolls on the pitcher's card, then fatigue will have a much bigger impact in ATG than in 200x. This is because in 200x the typical hitter cards have slightly more hits than the pitcher cards, but in ATG there is a huge difference between the mashing on the hitters' cards and the near perfection of the stud pitchers' cards.

[apolivka]

I started a new team recently and thought I would run another quick study on the results of pitcher fatigue. Here what I saw for for my team during the first 1500 or so rolls:

F0 25-55 OB 32-55 rolls
F1 3-5 OB 5-5 rolls
F2 2-2 OB 0-2 rolls
F3 8-11 OB 7-11 rolls
F4 4-6 OB 3-6 rolls
F5 5-6 OB 2-6 rolls
F6 3-4 OB 3-4 rolls
F7 2-8 OB 4-8 rolls
F8 10-27 OB 14-27 rolls
F9 1477 rolls and it was almost exactly 50/50 for rolls

The first element is the Fatigue number in the box score. The next numbers are the batter's result of getting on base. For example, on F0 25 out of 55 rolls ended up with the batter getting on base via hit or walk. The next set of numbers shows how often the batter rolled his result off his own card.

This set of data points seems to suggest a big drop off after about F7. But, it's pretty much unrefutable that an F0 through F3 pitcher is SUPER ineffective. The results show many more rolls off the batters card and a hitter on base percentage of around .500. F4 through F6 certainly isn't likely to be much better, since it was actually worse in this small sample of 16 rolls.

[MARCPELLETIER]

Apolivka,

Thanks for the effort. Just to make sure, are you certain that you took the result FOLLOWING the F-state.

For example, let say that a pitcher starts the inning at F9

single F4
double F3

This means that a single was hit while the pitcher was at state F9,
and a double was hit while the pitcher was at state F4.

Also I find extremely strange that you have the same number of rolls on the hitter card as on the pitcher card for every single state.

[ergie63]

I think you're on the right track marcus_wilby.

I would refine the various hypotheses as follows:

At F9 a roll on the pitchers card is taken at face value.
At F8, 19/20 times a roll on the pitchers card is taken at face value.
At F7, 18/20...
and so on well past F0.

The penalty (what happens 1/20 times at F8) is not a re-roll that is read off the batter's card, rather it is a hit or walk. Whether that ends up being a walk, single or home run is a proportionate random selection of what is available on the batter's card.

This keeps a batter's stats in line with his card rather than piling up singles on a slugger's card for example.

If we assume batters are retired at the rate of 66% from the pitcher's card then at...

F9 (1.0 * .66) + (0 * 0) = .66 chance of an out
F8( .95 * .66) + ( .05 * 0) = .627
...
F0 (0.5 * .66) = .333
F0-1 (.45 * .66) = .297
F0-9 (0. *.66) = 0

[apolivka]

Well the above stats are for the results where the F# shows on the box score. In other words, when F3 was shown at the end of a line in the box score, 8 out of 11 hitters made got on, and 7 out of the 11 rolls were on the batters card. What is was before that is not shown in my stats. I might change my program to see if there is any significant difference if I take that into account.

As for your other comment--only at F9 were there the same number of rolls on the pitchers card vs. the hitter's card. At the other F numbers, it was typically skewed towards the hitter's card.