forcing pinch hitter order

[yajeflow]

situation:
halfway through the season. no extra inning wins and 18 one-run losses.

i noticed i had two pitchers that are hitting .400+ on the season. why not give them a chance to pinch hit when not starting, eh?

i use 7-day lineups exclusively. i never even look at the other lineup options, as i want the lineup i want, period. someone pointed out that the vLHP and vRHP lineups offer pinch hitter choices. so why not give that a try, says i.

i tried it with the settings i use at the top of the 7-days:
YES
NO
NO
NO
NO

from what i could see, the manager did not use the guys i wanted in any way that i had wanted.

so i changed it to:
YES
NO
YES (allow use of depth chart)
NO
NO

the pinch hitters were unchanged. nothing i wanted happened, except my backup 2B suddenly started a game at 3B. this is exactly why i do NOT want to ever happen.

then i changed it to:
YES
YES (all use of alternate lineup)
NO
NO
NO

in the second game, there were seven guys who went up to pinch hit. of the four guys i listed on my PH list on the vRHP and vLHP screens, the #1 guy was the third PH off the bench, and the other three were never used.

so i guess my question is this:
is there any way to use a PH order when you use the 7-day lineups?

[NoOne]

the manager's concern about lefty v righty would affect this. i forget the exact label, but it's close to that. it's the last slider in the coach editor, if it isn't shown on the Team Strategy page.

you can also turn GM-legacy mode on (global settings) to have total control over a lineup and bullpen regardless of a micro-managing coach being hired. if i had to set lineups every 7 days it would take me significantly longer to play a season. that may or may not be important to you, but you can get 90-95% of what you want with gm-mode and ignorance is bliss after that.

[coljesep]

If you're talking about starting pitchers, could you change their position pre sim to make them a hitter that day and then they would be eligible in the depth chart?

[RchW]

My issue would be when the inevitable small sample size PH pitchers regress back to hitting 0.120 or they get injured legging out a dribbler to 2B will it have been worth it?

[yajeflow]

Originally Posted by RchW

My issue would be when the inevitable small sample size PH pitchers regress back to hitting 0.120 or they get injured legging out a dribbler to 2B will it have been worth it?

any hit in an extra inning game would be worth it, rich. it is MADDENING.

however, the pitchers have put together some decent stats so far:

17-43, 3 2B, 1 HR, .395/.422/.535
and
6-14, 2 2B, .429/.429/.571

23 for 57 is significant

[NoOne]

either very lucky or they hsould be starters (or traded for a more useful piece, if they are actually that good). if you got a professional hitter on your bench he isn't doing much for you.

if you simulate the games, you will be left to the AI's choices. if you want to ph l/r more often, i'd edit the coach and move that slider to the right.

the only way to manipulate the AI per your team is to edit the team strat and coach strategy (need comissioner mode to edit the coach - may be an ethical problem for some, but my coach is a digital me so i'm jsut being myself :P )

sliders to change for this topic:

pinch hit for position players - this will make it more/less common the coach will ph for a starter.

favor l/r matchup - they'll play the l/r matchup game more religiously, of course. i don't know if this is pitching / batting or both - i'd assume both.

pinch hit for pitchers will help to a lesser extent - if they are being used when you want, don't adjust... timing would probably be the key here; not frequency.

edit: what rich is saying is that they aren't actually as good as the small sample size numbers say they are... and you may actually be reducing your likelihood of a hit by using a pinch hitter - in fact most of the time this will be the case in most leagues.

when i have pinch hitter listed in the depth chart, it's usually a borderline developed everyday player... a little seasoning on the bench because they are too good for AAA. if they have power i pop them into the ph depth chart.

when you have a platoon situation that's an ideal spot to pinch hit often. something extra that makes pinch hitting viable.

[yajeflow]

Originally Posted by NoOne

either very lucky or they hsould be starters (or traded for a more useful piece, if they are actually that good). if you got a professional hitter on your bench he isn't doing much for you.

if you simulate the games, you will be left to the AI's choices. if you want to ph l/r more often, i'd edit the coach and move that slider to the right.

the only way to manipulate the AI per your team is to edit the team strat and coach strategy (need comissioner mode to edit the coach - may be an ethical problem for some, but my coach is a digital me so i'm jsut being myself :P )

sliders to change for this topic:

pinch hit for position players - this will make it more/less common the coach will ph for a starter.

favor l/r matchup - they'll play the l/r matchup game more religiously, of course. i don't know if this is pitching / batting or both - i'd assume both.

pinch hit for pitchers will help to a lesser extent - if they are being used when you want, don't adjust... timing would probably be the key here; not frequency.

edit: what rich is saying is that they aren't actually as good as the small sample size numbers say they are... and you may actually be reducing your likelihood of a hit by using a pinch hitter - in fact most of the time this will be the case in most leagues.

when i have pinch hitter listed in the depth chart, it's usually a borderline developed everyday player... a little seasoning on the bench because they are too good for AAA. if they have power i pop them into the ph depth chart.

when you have a platoon situation that's an ideal spot to pinch hit often. something extra that makes pinch hitting viable.

thanks for the reply. i am a statistician, so am well aware of sample size issues as well as the fact that they are pitchers. but at some point, data does become relevant. at this point, i would love for them to go in there and take some hacks just to see what is what. i actually have another relief pitcher who was a high school catcher. i won't PH him, as he has no recent data to go on.

update:
sure enough. 95 games in with no extra inning wins, and now we have won 2 of the last 3 in extra innings. and that's why they play the games...

[NoOne]

Originally Posted by yajeflow

thanks for the reply. i am a statistician, so am well aware of sample size issues as well as the fact that they are pitchers. but at some point, data does become relevant. at this point, i would love for them to go in there and take some hacks just to see what is what. i actually have another relief pitcher who was a high school catcher. i won't PH him, as he has no recent data to go on.

update:
sure enough. 95 games in with no extra inning wins, and now we have won 2 of the last 3 in extra innings. and that's why they play the games...

well, you say that but then you go on to use small sample data as evidence and completely ignoring the ramifications. like the update that you give is predicated on a small sample of data that cannot tell you anything with confidence. your previous posts used inferences from small samples, too.

alot of this depends on your scouting settings - if you have no ratings on, then you have to resort to trial and error in this situations (pitchers without much of a track record). if you have a good scout and feel comfortable with the ratings, then it's a balck-and-white decision as to when a ph is best for the team and who they should use.

unless those pitchers happen to be the freaky 2-way guys that develop their batting ratings while only pitching in the minors, then they can't consistently hit as pinch htters .. regardless of what the results (good or bad) from a handful of AB tell you. if they were 10-for-50 i'd say the same things if they were 30-50 --you can't know much from that sample.

Chris Shelton - hit ~.400 throgh the first 2-3 months of a season and was virtually never heard of again... way more than 50ab and better resutls. Brennan Bosch hits .340 with power as a rookie for 1/2 the year.. .then bombs... has another 1/2 good year the next and a pitiful other half... soon it's known that he was simply lucky early on while pitchers found his weaknesses. you should be able to think of numerous instances of these types of batters and pitchers from watching your own home team. (i forget the pitchers name started off hot including a no-hitter or nearly and then never pitched well after that - not galerraga (near no-hitter too), but he fits the description - he didn't bomb as fast as the guy i cannot think of the name)

(if they are that competent at hitting, there's a good chance they'd provide greater value in different roles.. especially if the power they are showing is actually real - only a larger sample can clear that up. a couple lucky hits in 50ab can severely impact slugging pct.)

if they don't have better ratings than the player you are pinch hitting for, then you are reducing probabilty of success when you use them. there's no dodging that fact.

if they are successful without good hitting ratings, then you have found a bug in the game - provide data and report it int he bug section.

[yajeflow]

Originally Posted by NoOne

well, you say that but then you go on to use small sample data as evidence and completely ignoring the ramifications. like the update that you give is predicated on a small sample of data that cannot tell you anything with confidence. your previous posts used inferences from small samples, too.

alot of this depends on your scouting settings - if you have no ratings on, then you have to resort to trial and error in this situations (pitchers without much of a track record). if you have a good scout and feel comfortable with the ratings, then it's a balck-and-white decision as to when a ph is best for the team and who they should use.

unless those pitchers happen to be the freaky 2-way guys that develop their batting ratings while only pitching in the minors, then they can't consistently hit as pinch htters .. regardless of what the results (good or bad) from a handful of AB tell you. if they were 10-for-50 i'd say the same things if they were 30-50 --you can't know much from that sample.

Chris Shelton - hit ~.400 throgh the first 2-3 months of a season and was virtually never heard of again... way more than 50ab and better resutls. Brennan Bosch hits .340 with power as a rookie for 1/2 the year.. .then bombs... has another 1/2 good year the next and a pitiful other half... soon it's known that he was simply lucky early on while pitchers found his weaknesses. you should be able to think of numerous instances of these types of batters and pitchers from watching your own home team. (i forget the pitchers name started off hot including a no-hitter or nearly and then never pitched well after that - not galerraga (near no-hitter too), but he fits the description - he didn't bomb as fast as the guy i cannot think of the name)

(if they are that competent at hitting, there's a good chance they'd provide greater value in different roles.. especially if the power they are showing is actually real - only a larger sample can clear that up. a couple lucky hits in 50ab can severely impact slugging pct.)

if they don't have better ratings than the player you are pinch hitting for, then you are reducing probabilty of success when you use them. there's no dodging that fact.

if they are successful without good hitting ratings, then you have found a bug in the game - provide data and report it int he bug section.

then you and i have a different definition of the word 'significant'. let's take the kid who is 17 for 43. for simplicity, let's assume that a pitcher usually hits .100. one hit every 10 at bats.

in 10 at bats, it would be perfectly normal to get 1, 2 or 3 hits. heck, do that twice - 20 at bats. let's say a guy was completely freakishly 'lucky' and got four hits after having gotten 3 hits in the previous 10 at bats. so he is now 7-20. at this point, he is already at the extremes of probability. but it is a small sample size...but one that is getting bigger.

now let's say that this guy who is already extremely 'lucky' then goes 10 for his next 23. the larger the sample, the more confidence you have that the data is significant, that it is an actual trend. unusual would be as pitcher getting three times the number of hits as a typical pitcher. this pitcher is currently hitting FOUR times a typical pitcher and is getting near 50 at bats. the extra base hits also suggest something is different.

do i think that this player is a missed batting star? no. do i suspect that he would regress to the mean? of course he would. but this data sample is suggesting that his mean will not be a typical .100 for batting average.

i do not use anything that shows 'real' ratings. if i did, then i wouldn't be here. so reality unfolds slowly. as far as this being a possible bug, you may be right there. i really hope it isn't. this player is only 21 years old, so maybe there is a bug where a younger player has not yet lost his 'special' attributes. who knows?

prediction:
in his next 43 at bats, it is almost impossible for him to get 17 hits. that is a stretch of success that any position could hardly manage if you looked at the best 43 at bat stretch for a whole season (of course, maybe a whole season does not fit your definition of significant, i suppose). even 10 hits would be well above an average pitcher. but i would wager that the combined hits will be above 21. 21 hits would be the current 17 plus 4 more for a typical .100-hitting pitcher in 43 at bats. why is that? because i CAN infer something from this small sample size. it DOES show something. would you like to wager one million eTacos? 21 hits or less in the first 83 at bats and you win.

[NoOne]

side note: no ratings at all? or just scouting inaccuracy? (but, not about overall that's for the ai's use mostly) - without ratings all you got is stats... so yeah that's all you are left to use.. but the +/- % error relative to sample size doesn't change

the difference:

you see 17 / 53

i see: (complete guess without doing the math on actual values but you'll get hte point)

10 to 20 out of 53 (or somethign similar in nature)... (.400 hitter, very unlikey, lol, but i chose a value a bit more relative to the debate, not skewed toward my argument -- it just has to include lower values than what would be useful as a pinch hitter to be relevant to the discussion - ~.200-.400 is fair to both sides and conveys everythign important - i wouldn't doubt that the actual math dips well below the mendoza line at 50ab for the minimum value - that range covers nearly all players in the game - not slimming down the possibilities very much at 50ab.)

i doubt he's a ".100" hitter, because i 'think' the reults you are seeing are too much of an outlier to be likely to occur (just guessing, might actually be within range, i wouldn't be surprised either way). i'd also assume that his current resutls are closer to the top than the bottom for obvious reasons. so i'd also safely assume the actual mean is quite a bit lower and work around that value (a bit of art mixed with science... art as in common sense). that brings a much lower range into the picture than working at .400 and subtracting. be optimistic and pick .280-.300+ (top ~quartile talent as a ph? someone is being misused if that's the actual case) and the range still will cover very poor batting averages, possibly as low as .100. -- it can be as far as 2x the +/- %, if you think the results are at the high end, which a .400 average relative to a normal mlb league average certain is.

but... it doens't tell me anythign definitively about the guy. like i said in real life a player can hit .400 for 2-3 months and be out of the majors in less than a year or two after that streak. that's alot more than 50ab, and look at how decieved you can be by 2-3 months... let alone as few as 50ab. the game is likely to be similar proportion /scope however you want to describe it.

17 / 43 can easily be a poor hitter parading as a good hitter. how crappy can he be? probably not .100, but crappy enough to be a bad choice as a ph relative to other options - including not pinch hitting.

btw, i never used .100 as an example, you picked an extreme tailored to your argument, but it doesn't fit the broader context. that certainly isn't the liekly threshold of to ph or not to ph - whether they guy is above .100 with certainty. i.e. just because the results would be an outlier for a .100 average in 50ab, doesn't mean they are a good hitter with any confidence...

you shouldn't feel safe about anything 50ab tells you, because it's +/- some grossly high percent error. when the percent error is 30-50-80% it's not too useful.

yeah, you can use that percent error relative to sample size and 'normal' volatility to figure out min max range that it can be off from the actual mean - then you apply the % error to a conservative guess of the mean. in this example, it's clearly not going to be anywhere near ~.400 like the results (17/43). it's very safe to knock .050 or possibly even .100 off that value and work aorund that with the perent error relative to a normal mlb batting average range for players. (the percent error that is calculated is essentially normal volatility plus additional randmness (probably bad vocab) from a small sample - same beasts relative to this topic, different causality.)

when that range covers significant portions of 'suck' you can't ignore it. eventually it will reduce that window as you get more ab... that was never an argument from me..

but, i do argue that 50ab is nowhere near a point where the range is small enough to use... even 100ab. i'd wager a pinch hitter is fairly unlikely to get enough of a useful sample size during their career AND while they maintain some consistency in talent (=the actual ratings you don't see). you don't get many ab per year, and they are likely to change significantly by the time you get enough AB to have some confidence in the results. if their ratings are signifcantly different after 5-10 years, they are not the same player as when those stats were accumulated. they are now data sets of apples and oranges even though it's from the same player.

i do agree it's not likely a .100 hitter.. not sure what you can do with that morsel of information, though.

17/43, 17/50, whatever/50ab.. even from the opposite perspective - if someone starts off poorly in 50ab i don't think they are likely to be terrible for hitting very poorly.

[yajeflow]

thanks for response and thoughts.

i used .100 as a simple AVG and fairly typical for a pitcher. a pitcher is who we are talking about, so that stands to reason.

it was frustration at blowing extra inning games that led me to even think about this. and when you see the averages of my bench players, you will see why a guy hitting 17 for 43 is appealing.

here is my bench:
.229
.213
.256
.238
.245
.222

so here is an update:
since i could not get him the top kid a spot as a PH in OOTP, i had a rare spot with my 1B needing a day of rest. so i went WAY outside the box and started the kid pitcher there. both pitchers went into the 7th with one-hitters. the kid went 0-3.

fail? not so fast. he started the next game on the mound and went 2 for 4 with a double and drove in three runs.

here is the 'problem' with your thinking. no offense, mind you. at what point in a season do YOU say "what the heck - give this guy a shot"?

a good example to support you, and the reason why i reach 'small sample sizes are a problem' is dusty rhodes. in 1954 as a PH for the giants, he hit .327/.365/.551. a real 'clutch' player, right? two years later in about the same number of at bats, he was a .174/.269/.239 pinch hitter. clutch is a myth.

but in 1954 i would've pinch hit him. i just would've. you are looking at reviewing a player after the fact. after everything is known. i am talking about making a call on a guy in the heat of battle. i dunno, man - 17 for 43 says something to me, and it says something to any statistician.

[NoOne]

i don't discount possibilities. 43-50AB is not a reliable sample. that is the only argument; not whether or not this pitcher is actually a good hitter. (you simply cannot tell from a small sample with great confidence - you have optimism, not confidence)

that sample size has too large of a +/- error to be considered representative of the player's talent - whether good ro bad. he may look good so far, but to infer anything about the future confidently from 43ab is blasphemous.

if you believe that... evolution is made up, the world is only 5000 years old, when you cut your hair it grows back faster, and playing in the rain really does get you sick. (none of that is true, just in case someone misread) oh and because it snowed in washington today, the world isn't actually getting warmer... nothing to see here, move along.

quote: " the kid went 0-3. fail? not so fast. he started the next game on the mound and went 2 for 4 with a double and drove in three runs."

i would never react that way to 0-3, or 0-10... or even 10-10 in the MLB. to think that way is objectively flawed. that's the essence of what i've been saying for the last couple posts.

that is exactly how you (plural -you, anyone, or me) get fooled by randomness when you look at things this way. 3ab, 7ab, or 43 ab tells you very little about the player. when you get to 2-3 years worht of data you can start saying things with certainty about a player, until that point it's a debate no matter how positive the signs look initially.

yes, it does become more clearer over time - beacuse the sample is growing to a suitable size! sample size is inversely ralated to likely %error. 43 ab is hardly threshold when things merely start to become clear. it takes ~990ab for BA to level off, i don't have to do the math to recognize 43 is going to be so far off that it's mostly unuseable data for predicting future BA. (diminishing returns the larger the sample, so half that sample will not be "half" as accurate - helps your argument a slight amount until you compare ~43 to ~990... 5-10% of a suitable sample size is going to be just terribly useless information)

while i don't encounter this situation much, i do encounter decisions with small samples to work with. i am just not nearly as optimistic about good results from small samples (as with bad results in small samples). i work with what is given, and in the case of 43ab being important, there'd have to be virtually no toher information to work with and even then i wouldn't have high expectations based on any results from 43ab.

no idea why "clutch" is being brought up. that example jut proves how fooled anyone can be by results from a small sample size. he wan't a .300+ba and .500+slugging talent - in spite of his anamolous 54-55 seasons with only 350ab... otherwise he would have continued to perform like that. sounds like a possible result for your very situation to me... and possibly not... that actual result isnt the point.. the point is that it's empirically uncertain

that guy never started because he wan't good.. .the manager knew it despite his uptic in production over a short period of time... unfortuantely we have nothing similar to this in the video game, if you turn off scouting. (he wasn't "crappy" but he wasn't good, either)

results don't make the batter... the batter has certain ability and if given enough opportunity it will reveal it's nature through stats from suitable sample sizes. that guy didn't becoem a crappy hitter after '55, he was a crappy hitter during that time period, too. he had 350 ab where he looked like an all-star... but he wasnt = fooled by randomness, unless you thought he was magically a better player in 54/55 and at no other time in his career.

if it can happen in ~350ab, then it can happen to you in 43 ab. that's the only thing i've described with certainty. what actually results in the future isn't relevant to this debate. the means (talent), not the ends(resulting stats), is important. that's the whole point of having enough data - for it to be actually representative of reality and therefore useful information to base decisions on. without that, you need to lean on the whatever other information you have access to. if you don't use scouts you are somewhat forced to use small sample data a bit more, i'd assume, but it doesn't change the facts that it's information that is unlikely to be representative of actual ability.

also you show #'s from your bench, this highlights exactly the problem.. those #'s are useless information with so few ab in a season for a ph. anyone of those guys could actually be a .300 hitter, the data simply doesn't tell you enough, yet. you can never judge a pinch hitter by any season in his career... there are too many factors that can mask reality in a small sample of results... unless you have magic eyes! do they glow? i want some glowing eyes that can see the future, too.

[yajeflow]

i think i have it now.

you feel confident of data after three years of results. congratulations. everyone has 100% hindsight.

if you were my manager in-game, and a situation arose, you would be unable to make a move. not enough data. but maybe you would PH a guy with a .300 career AVG. but what you would NOT know is that he was just starting a three-year period of decline in which he would bat only .200, so you would have to wait until THOSE three years were up, too.

meanwhile, i need to send someone up to bat. ME? the hottest hand right now. that is who i would send up.

2 for 3? fluke. anything can happen in three ABs.
5 for 10? pretty good, but still a fluke. doesn't tell me much at all.
9 for 21? well, now this is interesting. 21 at bats is about a week of games, and nine hits is pretty good.
19 for 50? well, 50 at bats is nearly 10% of a season's worth of at bats. and this guy is clearly grabbing hits right now. this is a hot bat. i would definitely use him as a PH.
you? um...nobody.

this is all kind of unfair to you, as you do not know the daily goings on of the roster. like pooshemup skarda. terrible recent performance despite adequate power numbers. dear negrete has been awful, too. muggsy leclair looks cool with those homers, but he hasn't done anything since april. hence my desperation and desire to give ANYBODY who can hit a shot at doing so.

Basic Batting Stats Report - Split: Overall

if you want to manage my team, you had better do more than tell me what you would've done three years ago with the knowledge you have today. i need to send someone up NOW.

[Number4]

Why are you so intent on not giving pitchers any kind of chance? What is the risk of them getting some pinch hit at bats in an extra inning game?

That isn't even an unorthodox move, managers do it all the time. In those cases, you're not comparing them to bench players batting, you're comparing them to relievers batting.

And in anyway, it isn't inconcievable that a righty batting starting pitcher raking for his profession might be a better option than a lefty .200 glove-first backup shortstop against a lefty pitcher. The Giants did let Bumgarner hit for himself instead of using a DH one time. The Cubs sometimes have pitchers even adventuring in the outfield.

What is the biggest loss that could happen? "Oh my pitcher struck out". Guess what, the marginal batters that a pitcher might actually pinch hit for do the same thing all the time. The opportunity cost of letting the pitcher bat isn't high.