3-Year Recalcs for Rookie Seasons, Final Seasons, Missing Seasons, Single-Season Players?

[chucksabr]

The 3-Year Calc function averages the season at hand with the two surrounding seasons, double-weighting the current season. To wit, from the manual:

If you are playing in 3-yr recalc mode in 1988, for example, then player ratings will be based on 1987+1988+1989 stats. If you are playing in 3-yr recalc mode with "double-weight" in 1988, then player ratings will be based on 1987+1988+1988+1989 stats.

Which I assume they divide by four.

This makes sense if you have the totally clean situation of looking at a player's year when they also played the prior and following year. But as we all know, there are many player-seasons that don't comport to that.

So how does OOTP treat the exceptions? Let's game out the 1988 example above for these different scenarios. What does OOTP do when it comes to:

• A player's rookie season? Do they double weight the current season and add in the following season once, e.g., 1988+1988+1989 divided by three?
• A player's final season? Same basic question, do they double weight the current season and add in the prior season once, e.g., 1987+1988+1988 divided by three?
• A player's missing season? Suppose the player in the example played in 1987 and 1988, skipped 1989, and played 1990. Do they just switch out the recalc to be 1987+1988+1988+1990 divided by four? Or do they assign a cipher/replacement value to 1989, which wouldn't be played, and recalc based on that? One tricky exception might be when a player has several missing seasons, such as Steve Avery between 1999 and 2003. Or, do they treat seasons adjacent to missing seasons as though they are rookie seasons or final seasons? (Same question when it comes to a player playing 1986, off in 1987, playing 1988, off in 1989, playing 1990—is it 1986+1988+1988+1990, or do they have cipher values for 1987 and 1989?)
• A single-season player? There is no prior or following season to average with, so do they just take straight average? (Sure, majority of these guys were cup of coffee guys so it hardly matters, but some weren't, like, tragically, Ken Hubbs.)

Appreciate the input, thanks.

[thehef]

First: I don't know for sure. Based upon some very limited testing & other research, I think OOTP just uses available stats and doesn't double up any years unless you select double-weight. But your adjust/weaken settings can very much come in to play.

So if you have three-year recalc enabled and 200 is your threshold for the adjust setting, then you need 600 AB's over those three years (1987, 1988, 1989). If your rookie has 15 AB's in 1987 and 425 AB's in 1988, then he has 440 AB's. To get to 600, then 160 adjusted AB's will be added to his stats and then his ratings will be calculated based upon those 600 AB's. (In this example, if you use double-weight, then you are using those 15 AB's in '87 and 850 AB's in '88, and therefore no adjusted AB's.)

If, OTOH, your rookie had 15 AB's in 1987 and 575 AB's in 1988, then only 10 adjusted AB's will be applied, and he's basically going to be rated to perform like he did in 1988. (For double-weight, you'd have 15 AB's for '87 and 1150 AB's for '88, and no adjusted AB's...)

For a final-year player, his AB's for 1987 and 1988 will be used. If that total falls under 600 AB's, then adjusted AB's will apply. So for a guy who played fulltime in '87 & '88, he's surely going to be well over 600 AB's so his ratings will be based on his final two years... (and double-weight would do exactly what you would think: give double weight to his '88 stats compared to his '87 stats).

For your other examples, let's say a guy - for whatever reason - has only 275 total AB's for years 1987 thru 1989. Doesn't matter what year(s) those 275 AB's occurred, OOTP is going to add 325 adjusted AB's to his stats in order to come up with ratings. (For double-weight, he'd have 550 AB's and therefore only 50 adjusted AB's would be added.)

This is how I think it works. I am not 100% certain on every detail, but I think it's at least pretty close

[Matt Arnold]

I always forget exactly, but I think the adjust settings only apply to the sum of the stats. So if your "adjust" is set to 200 AB, then as long as the player has 200 AB total in the (1, 2, 3, 3+doubled) stats [depending on settings], then he wouldn't get adjusted. He doesn't need 600 (or 800), depending on the settings.

But the calculations always count the years. So if you're calculating the 3-year on a player for 1988, it sums the 1987+1988+1989, whether they played or not.

[chucksabr]

Originally Posted by Matt Arnold

I always forget exactly, but I think the adjust settings only apply to the sum of the stats. So if your "adjust" is set to 200 AB, then as long as the player has 200 AB total in the (1, 2, 3, 3+doubled) stats [depending on settings], then he wouldn't get adjusted. He doesn't need 600 (or 800), depending on the settings.

But the calculations always count the years. So if you're calculating the 3-year on a player for 1988, it sums the 1987+1988+1989, whether they played or not.

So if a player debuted in 1988, it takes what he did in 1987, which was nothing, against 1988 and 1989? Let's pretend we are talking about OPS+. If a rookie gets 80 OPS+ in 1988 and 95 in 1989, what do they give him for 1987? Is it zero? So does that calculation go, (0+80+80+95)/4 = 64? That seems off—does the game apply something else for the 1987 he never played?

If a player plays only a single season, and gets an 80 OPS+ in 1988, what number would they apply for 1987 and 1989, years he did not play?

[Matt Arnold]

Originally Posted by chucksabr

So if a player debuted in 1988, it takes what he did in 1987, which was nothing, against 1988 and 1989? Let's pretend we are talking about OPS+. If a rookie gets 80 OPS+ in 1988 and 95 in 1989, what do they give him for 1987? Is it zero? So does that calculation go, (0+80+80+95)/4 = 64? That seems off—does the game apply something else for the 1987 he never played?

If a player plays only a single season, and gets an 80 OPS+ in 1988, what number would they apply for 1987 and 1989, years he did not play?

Everything essentially is weighted by how much they played each season. We don't import OPS+ directly, but take something like power. If they have 0 HR in in 0 AB in 1987, 20 HR in 600 AB in 1988, and 30 HR in 400 AB in 1989, then counting the 3 year doubled current, they would be rated essentially as a player with 0+20+20+30 = 70 HR in 0+600+600+400 = 1600 AB.

[Rain King]

Originally Posted by Matt Arnold

I always forget exactly, but I think the adjust settings only apply to the sum of the stats. So if your "adjust" is set to 200 AB, then as long as the player has 200 AB total in the (1, 2, 3, 3+doubled) stats [depending on settings], then he wouldn't get adjusted. He doesn't need 600 (or 800), depending on the settings.

But the calculations always count the years. So if you're calculating the 3-year on a player for 1988, it sums the 1987+1988+1989, whether they played or not.

I *think* this is slightly incorrect in that if your "adjust" setting is 200 and you are using 3-year recalc that actually places the "adjust" level at 600 at-bats. It multiplies the adjust number by the number of years being considered in the recalc.

Hopefully Garlon will weight in. I believe he is the resident expert on this stuff.

[thehef]

Originally Posted by Rain King

I *think* this is slightly incorrect in that if your "adjust" setting is 200 and you are using 3-year recalc that actually places the "adjust" level at 600 at-bats. It multiplies the adjust number by the number of years being considered in the recalc.

Hopefully Garlon will weight in. I believe he is the resident expert on this stuff.

That's what I believe Garlon explained elsewhere, and that was the basis for my attempted explanation in the first reply above ... Both he and Matt are aware of this thread so rest assured that one of them will clear this up for us soon

[chucksabr]

Originally Posted by Matt Arnold

Everything essentially is weighted by how much they played each season. We don't import OPS+ directly, but take something like power. If they have 0 HR in in 0 AB in 1987, 20 HR in 600 AB in 1988, and 30 HR in 400 AB in 1989, then counting the 3 year doubled current, they would be rated essentially as a player with 0+20+20+30 = 70 HR in 0+600+600+400 = 1600 AB.

I see. That's helpful, thanks.

[chucksabr]

Originally Posted by Rain King

I *think* this is slightly incorrect in that if your "adjust" setting is 200 and you are using 3-year recalc that actually places the "adjust" level at 600 at-bats. It multiplies the adjust number by the number of years being considered in the recalc.

Hopefully Garlon will weight in. I believe he is the resident expert on this stuff.

So the adjust level would be 600 for the entire 3-year period? So if a batter IRL has 100 at bats each in 1985 and 1986 with no homers either year, and then busts out in 1987 with 40 homers in 600 at bats, would the game then assume the rate of (40/100+100+600)) = 5.0% HR/AB rate across all three years, rather than individual home run rates of 0.0%, 0.0%, and 6.7% year to year? Is that how you understand it?

[thehef]

Originally Posted by chucksabr

So the adjust level would be 600 for the entire 3-year period?

It would be 600 AB's if your adjust setting was 200 AB's, yes.

Originally Posted by chucksabr

So if a batter IRL has 100 at bats each in 1985 and 1986 with no homers either year, and then busts out in 1987 with 40 homers in 600 at bats, would the game then assume the rate of (40/100+100+600)) = 5.0% HR/AB rate across all three years, rather than individual home run rates of 0.0%, 0.0%, and 6.7% year to year? Is that how you understand it?

So if you are in 1986 in OOTP, then, yes, you would be looking at 800 AB's with 40 HR's. And then OOTP turns that in to 550 AB's with 27.5 HR's...

[Garlon]

If you use 200 AB for your adjust setting and use 1yr it looks for a total of 200 or more AB, if you use 3yr it looks for 600 AB over those seasons, if you use 5yr it will look for a total of 1000 AB over those seasons. If you are using 3yr double weight then it will be 800 AB, and if you are using 5yr double weight then it will be 1200 AB.

If you use 3yr recalc with 200 AB, for rookies and players in their final year it will only look for 400 AB because the player did not play before or after the given season.

If there are gaps in the middle of the career for a player it does not account for that. A good example of this is Hank Greenberg who missed 1942, 1943, and 1944 during WW II. If you are using 3yr recalc with 200 AB he is fine in 1941 because he has a total of 640 AB between 1940 and 1942. When the game moves to 1942 it sees he has only 67 AB between 1941 and 1943 and his ratings tank. In 1943 he has 0 AB between 1942 and 1944, so the game just holds over his previously tanked ratings from 1943. When you get to 1944, he only has 270 AB between 1943 and 1945 so he gets a new set of tanked ratings. When you get to 1945, he is fine because he has 793 AB between 1944 and 1946.

An option that I have always thought would be beneficial could be to have the game use "most recent 550 AB", so if in the focus season a player has 550 AB then the game uses that season, if the player is below 550 AB it will try 3yr recalc, and if 3yr recalc is still below 550 AB it will try 5yr recalc, and so forth until it has at least 550 AB. This could be the same for pitchers. If we added this to the game, it should only be an option because this will add a significant amount of talent to the league every season and reduce big seasons. There could be options for 1yr recalc, 3yr recalc, 5yr recalc, and "variable multiyear recalc" which is this model.

With variable multiyear recalc Greenberg and anyone else gets a good import every season. There are definitely problems with this though. There are players who broke into the majors but did not really play much for 5 or more seasons, and then when they did play they had a big season. Under this process this player could get an extra 5 seasons or more of great seasons. So perhaps we will need to just have variable multiyear recalc limited to using 1yr, 3yr, or 5yr but not further out than that to find 550 AB. There is no perfect solution to the import settings, but with 150 AB and 5yr recalc the only significant player I have seen with some issues is Hank Greenberg.

For historical games I use 5yr recalc and use either 150 or 200 AB for the adjust setting.

[chucksabr]

Originally Posted by Garlon

If you use 200 AB for your adjust setting and use 1yr it looks for a total of 200 or more AB, if you use 3yr it looks for 600 AB over those seasons, if you use 5yr it will look for a total of 1000 AB over those seasons. If you are using 3yr double weight then it will be 800 AB, and if you are using 5yr double weight then it will be 1200 AB.

If you use 3yr recalc with 200 AB, for rookies and players in their final year it will only look for 400 AB because the player did not play before or after the given season.

If there are gaps in the middle of the career for a player it does not account for that. A good example of this is Hank Greenberg who missed 1942, 1943, and 1944 during WW II. If you are using 3yr recalc with 200 AB he is fine in 1941 because he has a total of 640 AB between 1940 and 1942. When the game moves to 1942 it sees he has only 67 AB between 1941 and 1943 and his ratings tank. In 1943 he has 0 AB between 1942 and 1944, so the game just holds over his previously tanked ratings from 1943. When you get to 1944, he only has 270 AB between 1943 and 1945 so he gets a new set of tanked ratings. When you get to 1945, he is fine because he has 793 AB between 1944 and 1946.

Tremendous overview, thanks. I’m in a league that uses 3-year double weight, and I have no opportunity to change it since I am just a team owner.

I might have an additional question or two as I work with your overview against my info, but one immediate question I have is about pitchers: Our league is 100 IP for starters and 25 IP for relievers. If that is based on IP across three years and the pitcher must be either a starter or a reliever, how does the game determine that?

I assume it is based on games started vs total games pitched across the 3-year period. Would it be, if it’s over 50% of games appeared in that the guy started, he’s a starter, and under 50% started, he’s a reliever? If it’s exactly 50%, which is he? Or is it some other percent or metric?

[Garlon]

The game decides if a pitcher is a SP it RP based on percentage of game started to total games pitched. This percentage is about 35% I think. There us also so minimum amount of GS as a criteria in there. I have not seen any issues identifying RP or SP, but I set up the game to base Stamina on career values.

If your league is set up with 25 IP for relievers and 100 for SP, then for 3yr it will be 75 innings for relievers and 300 for SP. The same rules for batter AB apply to pitcher IP here.

[thehef]

Speaking of pitchers, I was wondering how OOTP handles their batting ratings. Obviously, there have been some good-hitting pitchers. Most of them get fewer than 100 AB's over the course of a year (well, in past years when starters were workhorses, they might get 130 or so..), and thus fall under most gamers' "adjust" setting for AB's.

Several versions ago, there was no problem with this. But I think in v22 I had done some testing and found that all pitchers, regardless of IRL performance, sucked as hitters...

I do have one example from a recent v 23 (for other purposes)... Don Drysdale, for the period 1963-68:

Real Life: 195 avg, 10 HR's in 614 AB's, with his top season being 1965 (.300, 7 HR's in 130 AB's).
OOTP: .197 avg, 14 HR's in 554 AB's, with top season being 1965 (.279, 9 HR's in 172 AB's).

Based on the above - which is using 1-year recalc - I'd say it's working as it should... Very small sample size, though... More testing needed...

[chucksabr]

Originally Posted by Garlon

The game decides if a pitcher is a SP it RP based on percentage of game started to total games pitched. This percentage is about 35% I think. There us also so minimum amount of GS as a criteria in there. I have not seen any issues identifying RP or SP, but I set up the game to base Stamina on career values.

If your league is set up with 25 IP for relievers and 100 for SP, then for 3yr it will be 75 innings for relievers and 300 for SP. The same rules for batter AB apply to pitcher IP here.

Thanks again for checking back in. Do you mean that if a player pitches 35% or more of his games as a starter, meaning up to 65% of his games as a reliever, OOTP determines he is a starter for the purposes of recalc?

Beyond that, if you can help me with these questions, I think I can acquire a complete-enough understanding of it for my purposes:

• If a player does fall short of the minimum and must be adjusted, what numbers (or factors) do they use to adjust it? I had read somewhere that their IRL numbers are applied to the IP or AB they did achieve, then some replacement level (or maybe even lower than that) factor is applied to the remaining IP/AB to bring them up to the minimum for the recalc. If that's how it's done, any way you can confirm what those factors might be? Seems to me it might be even lower than replacement level ...
• Does a single season career year count twice for the player in a double-weight situation? Meaning, if a single-season player pitched just 40 innings in relief in his only season, does he get "double credit" up to 80 IP, and thus meet the 3-year reliever threshold of 75 IP to avoid adjustment with just 40 actual innings?
• If a guy like Jose Parra pitches in 2000, has a gap year, then 2002, gap year, then 2004, does the game treat that 2002 season like a single-season-career year?
• If I try to force a player to play in a year during a series of gap years, how would recalc treat that? Steve Avery is a good example: he pitched in 1999, then didn't pitch again until 2003. If I forced him to pitch in 2001, how would recalc treat that?
• My example of the player above, the guy with 0/0/40 HR in 100/100/600 AB across three seasons, when he busted out in the third year: even though the HR rate applies across all three seasons, including the first two when he got none IRL, does the fact that he played only 100 AB his first two seasons affect how I can use him in OOTP? Meaning, if I play him 600 AB all three years, will he have as much seasonal stamina in the first two years as the third, even though IRL he played way less in those first two?

[Garlon]

1. The game will always use their real stats then add in sone adjusted amount. So if you use 200 AB as the setting and a batter had 190 AB, the band will add in an extra 10 AB at sone level of performance to get to 200. I do. It know how the nudging AB are handled but they are not counted as 0 hits in those AB. There is sone replacement level used. Hopefully that level is tied to the given season.

2. If a pitcher only played a single season and you are using 3yr double weight, then that season should count twice, but will also have an expectation of 2x the setting. In your example of using 25 as the setting, the pitcher in question already had 40 innings. In 3yr double weight it will look for 50 innings because there is no previous or subsequent year. Remember with 3 yr double weight the doubling also counts as an extra season. With 25 innings 3yr double weight the game will be looking for 100 innings. For a player in their only season it will be poking for 50 innings.

3. Gap years are not ignored in the process. The only seasons that are ignored is the season before their debut season and the season after their final season.

4. For Avery in 2001 it will not have any stats across 3 seasons to use. In this situation the game will hold over his last set if ratings, which will be from 2000.

5. Position player stamina is in the settings menu. In the d smoke with the 0/0/40 player he will be getting new ratings every season. In the first season of that set the game will. It even see the third season of 40 HR because 3yr RF SK only looks one season in each direction. The fact that the player only had 100 AB in the 2nd season does not weaken them because they still have 800 AB total. This is the point of the recalculate, to get a good sample size for the ratings.

[thehef]

Originally Posted by Garlon

... [it] will add in an extra 10 AB at some level of performance to get to 200. I don't know how the nudging AB are handled but they are not counted as 0 hits in those AB. There is some replacement level used. Hopefully that level is tied to the given season.

From the man himself a few years ago:

Originally Posted by Markus Heinsohn

When a player is below the "adjust" limit, his ratings will be adjusted towards the league average, the less playing time he had the more the ratings get adjusted. If a player is below the "weaken" limit, the ratings will also be adjusted, but not towards league average but rather towards replacement level, so that these players end up with rather low ratings.

So I take that to mean that to fill out AB's to adjust a player get to 200 for a player who had 150 AB's, those 50 AB's would skew towards league average.

And to fill out AB's to weaken a player to get to 40 AB's, who only had 20 AB's, those 20 AB's would skew towards replacement level.

I also assume, but do not know, that either league average or replacement level AB's applied to a player could result in actually improving the player... For example, a guy who in real life went 0-for-20, striking out 18 times would probably be improved by having 20 replacement-level AB's applied to him... I don't know what for sure, though.

[chucksabr]

A guy I know has his league on adjust, but not weaken. What is the implication of this?

According to Markus, players under the 300 AB and 100/25 IP thresholds are adjusted to league average, which I would assume is 100 OPS+ and 100 ERA+, but players under 50 AB and 40/10 IP are weakened to replacement level, perhaps to something like 80 OPS+ and 80 ERA+. Makes sense.

What happens with a cup of coffee player under this scenario? A guy named Ricky Pickett faced 9 batters in his career, gave up 3 hits, 4 walks, 6 earned runs in 0.2 innings, and ended up with an 8 ERA+. (That's eight, not eighty.)

If Ricky Pickett is weakened in the game for the 98.2 innings he doesn't pitch (his 0.2 being doubled), I would guess he comes in as a replacement level pitcher of around 80 ERA+. But if the league operator decides to remove the "50" default from weaken while setting up the league, does that mean Ricky Pickett would get adjusted for the 98.2 innings he doesn't pitch, and thus comes in as a more or less average pitcher around 100 ERA+? (Or maybe 99 ERA+?) That would make him a pretty valuable guy in that case, right? Or does the game make further adjustments in such a case?

[thehef]

Interesting. Although I can't speak to the ERA+ stats you'd expect, I suspect that the way you describe it is accurate. (If I'm understanding the "weaken" numbers for AB & IP would be set to 0, correct?)

This is definitely something to test to be sure...

(And I'll just note that I've been spending waaaay too much time testing stuff lately. I really don't need you coming up with interesting things like this...)

[chucksabr]

Originally Posted by thehef

Interesting. Although I can't speak to the ERA+ stats you'd expect, I suspect that the way you describe it is accurate. (If I'm understanding the "weaken" numbers for AB & IP would be set to 0, correct?)

This is definitely something to test to be sure...

(And I'll just note that I've been spending waaaay too much time testing stuff lately. I really don't need you coming up with interesting things like this...)

I put it in terms of ERA+ and OPS+ because those are already rate stats, and the whole 3-yr recalc double season and all that strikes me as more or less a way to convert counting stats across three years to rate stats so they can be applied to player projections in the game. If that's the case, then might as well talk in terms of rate stats that are already there.

Anyhoo—if the game does what Markus says and a league operator decides to change the "weaken" default from 50 to zero, then what would be the alternative except for the game to simply adjust everyone to league average, including Ricky Pickett, Jim Paciorek, and the rest of them? The only alternative would be that the game ignores Markus's directive to do so on its own and decides to make cup of coffee guys replacement players anyway. I wouldn't think the game is artificially intelligent enough to make that decision on its own.

Unless there's a completely other thing neither of us are considering?