No longer able to use a certain year as base year in Random Debut.

[actionjackson]

Originally Posted by pstrickert

I’ve also seen questionable results in historicals in OOTP18. On the beta forum, I went round and round with Markus and Matt about defensive ratings, after I noticed some significant differences compared to previous versions. Since I could document the differences, they finally agreed that something had changed — what it was, they weren’t sure. But they contended, IIRC, that the changes must’ve made the results more, not less, accurate. I still beg to differ. But whatev.

This might be part of it for sure. I've heard something about defensive ratings being "different" from what they had been in OOTP16. This is unfortunate because it makes employing players like Mark Belanger, Ozzie Smith (pencil in your favourite defensive wizard from history here) et al less optimal than it could be. Baseball is not all about hitting. There are definitely different strata of defensive play IRL, so why have they suddenly stopped being as pronounced as they once were in the best sim money can buy?

[actionjackson]

Originally Posted by pstrickert

I’ll add, sadly, that I trust historical results less than I used to. (It pains me to say it.)

Sadly, I'm right there with you, and I will only use OOTP16 for the historicals that I invest any time or "work" into.

[David Watts]

Wish I still had the league so I could do a screenshot. I did a fast sim using real lineups with 18. Rogers Hornsby hit .485 one season.

[pstrickert]

Originally Posted by actionjackson

Sadly, I'm right there with you, and I will only use OOTP16 for the historicals that I invest any time or "work" into.

I know some people appreciate having all kinds of non-MLB leagues to choose from. Personally, I’d like to see more attention paid to tightening up the historical MLB experience. Seems to me it’s gotten short-shrift for a while now.

[Markus Heinsohn]

Originally Posted by pstrickert

I’ve also seen questionable results in historicals in OOTP18. On the beta forum, I went round and round with Markus and Matt about defensive ratings, after I noticed some significant differences compared to previous versions. Since I could document the differences, they finally agreed that something had changed — what it was, they weren’t sure. But they contended, IIRC, that the changes must’ve made the results more, not less, accurate. I still beg to differ. But whatev.

What changed were the defensive ratings in our database, they used to be buggy, but now aren't The results are now a bit different, but overall more realistic in terms of defense, at least in theory, lol.

Anyway, the topic here is a completely different one, as this is about random debut league. I will run a few tests, maybe there is a ratings calculation bug for some players.

[Markus Heinsohn]

Originally Posted by actionjackson

As I've been saying throughout this, what I'm seeing is that league totals are relatively fine. Great in fact. Using 1984 as a base year for every season, I should be seeing something close to .260/.323/.385/.708 with 4.26 R/G. I believe in the five year test sim I ran, it was something like a .262/.325/.383/.708 slash line with 4.32 R/G, which is absolutely within an acceptable range.

What I find unacceptable, and what I am objecting to is the number of "superhero" seasons (as David Watts calls them) that I'm seeing relative to what I was seeing in OOTP16. A lot has changed between OOTP16 and OOTP18 in historical games. One of the main things that has changed is the addition of the full minor league database. I'm wondering what the connection could be between that and having a surge in super-duper seasons. Is it affecting the ratings of the better players in the MLB database, and spiking them, causing too many superseasons?

The thing is, with random debut, there are no players from the minor league database to influence this. I don't pretend to know the answers to this stuff, or to be smart enough to figure out how to fix it. All I can say is, I don't like what I'm seeing in 18, and I'm pretty sure it was in 17 as well. If I had to point a finger at anything, it's going to be the introduction of the minor league database. That was a major overhaul that could potentially create a disruption in the distribution of the league-wide stats, which, as explained above, appear perfectly normal.

Random debut leagues should be played without minor leagues since it only imports major league players.

And in general, the minor league databas ehas no real influence on the results of the MLB simulation because very few players get from the minors to the majors unexpectedly and change the league in some way.

If the problem in general is some extreme seasons from some rare players while league averages are fine, then this is either a) bad luck or b) a ratings calculation bug.

[David Watts]

Originally Posted by pstrickert

I know some people appreciate having all kinds of non-MLB leagues to choose from. Personally, I’d like to see more attention paid to tightening up the historical MLB experience. Seems to me it’s gotten short-shrift for a while now.

I thought I would enjoy historical minors, but no matter how hard I try I just can't enjoy it. As an example, I started a season in the early 70's in which I was going to follow the Detroit Tigers. 3rd or 4th game of the season, Al Kaline went down with an injury. So, I expect to see Northrup in RF and Stanley in CF. Instead, the AI calls up some guy that never even sniffed the bigs and inserts him in RF and bats him 2nd in the order. I'm also not a fan of rotations being manned by guys that never made it out of the minors. I guess my main beef is, the guys in the minors are just that and the AI should see them that way. Instead, since the AI doesn't see names or real playing time, it promotes guys that clearly don't belong in a big league lineup. For me, this takes all the fun out of playing historical.

As you pointed out in another thread, real lineups need to have some form of roster limits. It would be great if the game could have real injuries and real promotions/demotions, but if that's not possible, a method of determining the top 25 players by playing time might be the way to go. Trying to maneuver through a 25 man pitching staff is not fun.

I've long been a proponent of the real transactions feature having it's very own injury setting option. This option would completely eliminate career ending injuries and limit in season injuries to 2 weeks or less. I know, (because someone will always tell me) that I could edit the injury file myself. I don't want to mess with a OOTP file. I play other ways as well. When I play random debuts, I use the high(realistic) setting for injuries. So, I'm not editing a file. I don't even care if the injury the player receives is an injury that couldn't heal in 2 weeks. By only having injuries last 2 weeks or less, I've already stepped into the twilight zone, so my imagination can change a torn labrum to a blister if need be. I just think it would be cool to play with real transactions, but still have to deal with not being able to pencil in Ty Cobb 154 games a season.

[David Watts]

Originally Posted by Markus Heinsohn

Random debut leagues should be played without minor leagues since it only imports major league players.

And in general, the minor league databas ehas no real influence on the results of the MLB simulation because very few players get from the minors to the majors unexpectedly and change the league in some way.

If the problem in general is some extreme seasons from some rare players while league averages are fine, then this is either a) bad luck or b) a ratings calculation bug.

Neither Action or myself are using minor leagues. We both use reserve rosters. I wouldn't even dream of using the minor league database for random debuts.

As for the extreme seasons, I beg to differ. Its not bad luck. If it's a ratings calculation bug, then it's something that was introduced post 16. This method of using one season as a base year, worked like a charm in the past. I can't test using 17 because it's the first version since 12 that I've completely removed from my computer, but its easily reproducible with 18.

Then if we move out of using the one season as a base year method, would you say Rogers Hornsby hitting .485 in a season using real lineups and 1 year recalc is bad luck? How about Melky Cabrera hitting .417 in a random debut league(at work so I may be off a little on Cabrera, but I know he hit over .400). I admit even with 16 you get guys that out perform real life. I just don't see guys doing so at such an extreme level. A level that rivals the best seasons ever put together in the history of baseball. .

[David Watts]

Originally Posted by Markus Heinsohn

What changed were the defensive ratings in our database, they used to be buggy, but now aren't The results are now a bit different, but overall more realistic in terms of defense, at least in theory, lol.

Anyway, the topic here is a completely different one, as this is about random debut league. I will run a few tests, maybe there is a ratings calculation bug for some players.

I know this is a different topic, but it was brought up, so.... If you play using the MLB Quickstart, you will see guys rated 100(or 80 if that scale is being used) at a position. If you play fictional you will see the same. Play historical and it's pretty much unheard of. A guy like Mark Belanger ends up on his teams bench because his glove is not valued enough to make him get the starting nod. I question whether the ratings in the database were changed to make the game more realistic or whether they were changed because time wise it took too much time to look at each player and give them the ratings that best represent his major league abilities. Adding 10 million minor league players, Negro league players and it looks like we are getting Japanese players too, might play into this.

[pstrickert]

Originally Posted by Markus Heinsohn

What changed were the defensive ratings in our database, they used to be buggy, but now aren't The results are now a bit different, but overall more realistic in terms of defense, at least in theory, lol.

Anyway, the topic here is a completely different one, as this is about random debut league. I will run a few tests, maybe there is a ratings calculation bug for some players.

Thanks for weighing in, Markus. You may be right about “in theory;” in practice, though, something doesn’t seem right. :-\

[Markus Heinsohn]

Originally Posted by pstrickert

Thanks for weighing in, Markus. You may be right about “in theory;” in practice, though, something doesn’t seem right. :-\

Well, we discussed that in length already, and there is nothing we can do here, except for revisiting the fielding ratings in the DB, although they do look better right now IMO. The game will never be able to replicate the success / failure of teams vastly over/underperforming in real life, which was the core issue you complained about IIRC.

[Markus Heinsohn]

Originally Posted by David Watts

I know this is a different topic, but it was brought up, so.... If you play using the MLB Quickstart, you will see guys rated 100(or 80 if that scale is being used) at a position. If you play fictional you will see the same. Play historical and it's pretty much unheard of. A guy like Mark Belanger ends up on his teams bench because his glove is not valued enough to make him get the starting nod. I question whether the ratings in the database were changed to make the game more realistic or whether they were changed because time wise it took too much time to look at each player and give them the ratings that best represent his major league abilities. Adding 10 million minor league players, Negro league players and it looks like we are getting Japanese players too, might play into this.

I will revisit this during beta, maybe I can tweak the import of the ratings a bit and apply some sort of modifier top make the really good fielders better.

The other topic, I just simmed a random debut from 1946 to 1966 and all numbers looks fine to me, including individual seasons. No weird / unexplainable outliners. League averages were fine, too. This was in replay mode with ratings recalc, tomorrow I will run a league with career mode (and player dev enabled, recalc disabled), maybe this is a different story. What mode did you use?

[David Watts]

Originally Posted by Markus Heinsohn

I will revisit this during beta, maybe I can tweak the import of the ratings a bit and apply some sort of modifier top make the really good fielders better.

The other topic, I just simmed a random debut from 1946 to 1966 and all numbers looks fine to me, including individual seasons. No weird / unexplainable outliners. League averages were fine, too. This was in replay mode with ratings recalc, tomorrow I will run a league with career mode (and player dev enabled, recalc disabled), maybe this is a different story. What mode did you use?

Markus, I've ran a random using 1 year recalc and "import historical modifiers for accuracy" from 1947 through 2015. Though, I did see some instances, like the above mentioned Melky Cabrera, for the most part the numbers were in the realm of believable. League average, OBP and ERA were a little high, but as a whole the league produced good numbers. In fact, Hank Aaron retired with 756 home runs. How cool is that?

The method of play that I'm having issues with is something that works great using 16, but produces super seasons with 18. This a method that is a little more time consuming to test because you have stop each season prior to opening day to reset things to a season of choice.

What I do is to start in 1984 (or any other season). After the first season begins I uncheck the box for "import historical modifiers for accuracy." I play out the season. I then proceed through the offseason and preseason, stopping a day or two before the next season begins. I then place a check in the box for "import historical modifiers for accuracy" and above I change the year (since I started in 84 and played one season it will now say 1985) back to 1984. The game runs the calculations(status bar moves across top of screen. Once calculations are finished, I once again remove the check mark from the "import historical modifiers for accuracy" box and play out my season. Rinse and repeat year after year.

One thing I do when playing this way is to also remove the check from the players creation modifiers once the first season starts. That way, the players are always imported with the creation modifiers from 1984. I've wondered at times if maybe the game is overriding that lack of check mark and using current day player modifiers. But, I'm not sure if that would create the super seasons I'm seeing.

[David Watts]

Originally Posted by Markus Heinsohn

I will revisit this during beta, maybe I can tweak the import of the ratings a bit and apply some sort of modifier top make the really good fielders better.

The other topic, I just simmed a random debut from 1946 to 1966 and all numbers looks fine to me, including individual seasons. No weird / unexplainable outliners. League averages were fine, too. This was in replay mode with ratings recalc, tomorrow I will run a league with career mode (and player dev enabled, recalc disabled), maybe this is a different story. What mode did you use?

In regards to fielding. In the fast sim random debut league I did from 47-2015, I noticed there seemed to be a lot less repeat winners of the Great Glove award when I compared it to my current random that's using OOTP16, currently in the 18th season. Made me wonder if great gloves were much more random, because no players are rated high enough to stand out above the crowd? Therefore, it's much more common for a different player to win each year.

[pstrickert]

Originally Posted by Markus Heinsohn

Well, we discussed that in length already, and there is nothing we can do here, except for revisiting the fielding ratings in the DB, although they do look better right now IMO. The game will never be able to replicate the success / failure of teams vastly over/underperforming in real life, which was the core issue you complained about IIRC.

IIRC, I began seeing in OOTP18 results I had not seen in previous versions of the game when simming the 1974 season (and other seasons). Instead of being a virtual lock to win the AL West, for example, the ‘74 A’s struggled to win the division even half the time. So, I began investigating. Turns out that the A’s defensive ratings took a major hit in v18 compared to earlier versions. Instead of getting comparable wins and losses (and ERA) compared to real life (as in pre-OOTP18 versions), the A’s generally underperformed in OOTP18. I could cite other examples, but this one was the most noticeable and disturbing. I have a hard time believing that the game is producing more accurate results than it used to.

[Markus Heinsohn]

Originally Posted by pstrickert

IIRC, I began seeing in OOTP18 results I had not seen in previous versions of the game when simming the 1974 season (and other seasons). Instead of being a virtual lock to win the AL West, for example, the ‘74 A’s struggled to win the division even half the time. So, I began investigating. Turns out that the A’s defensive ratings took a major hit in v18 compared to earlier versions. Instead of getting comparable wins and losses (and ERA) compared to real life (as in pre-OOTP18 versions), the A’s generally underperformed in OOTP18. I could cite other examples, but this one was the most noticeable and disturbing. I have a hard time believing that the game is producing more accurate results than it used to.

Yes, that is the example team we discussed already, the real life A's 1974 probably simply overachieved relative to the basic stats that we use for rating calculation.

Let's revisit this during beta. I'll examine the fielding ratings import again and maybe Spritze can do a little tweak to the ratings DB. We'll see

[Markus Heinsohn]

Originally Posted by David Watts

What I do is to start in 1984 (or any other season). After the first season begins I uncheck the box for "import historical modifiers for accuracy." I play out the season. I then proceed through the offseason and preseason, stopping a day or two before the next season begins. I then place a check in the box for "import historical modifiers for accuracy" and above I change the year (since I started in 84 and played one season it will now say 1985) back to 1984. The game runs the calculations(status bar moves across top of screen. Once calculations are finished, I once again remove the check mark from the "import historical modifiers for accuracy" box and play out my season. Rinse and repeat year after year.

One thing I do when playing this way is to also remove the check from the players creation modifiers once the first season starts. That way, the players are always imported with the creation modifiers from 1984. I've wondered at times if maybe the game is overriding that lack of check mark and using current day player modifiers. But, I'm not sure if that would create the super seasons I'm seeing.

The player creation modifiers are not used for real players, only fictional.

Regarding your mode of play, I think what you do confuses the game. What is it that you are trying to accomplish? I am sure there is a better way to do it.

[pstrickert]

Don’t want to wear on your nerves again, Markus. Sure looks to me, when examining the real life A’s, that they were the cream of the crop in the AL. Up until OOTP 18, they were likewise the best team in the AL as a rule (when simming). But I’ll give it another go during beta.

[pstrickert]

If I’m not mistaken, I believe Markus’ stance is that the real life ‘74 A’s overperformed. Previous versions of the game allowed them to overperform. OOTP18, however, revealed the truth: the A’s were not really as good as their real life performance led us to believe.

[David Watts]

Originally Posted by Markus Heinsohn

The player creation modifiers are not used for real players, only fictional.

Regarding your mode of play, I think what you do confuses the game. What is it that you are trying to accomplish? I am sure there is a better way to do it.

The idea is to keep or tie stats to a particular season, in this case 1984 forever. This method in the past has always ensured you didn't get a gradual upward creep in offense that you get when you simply leave the box unchecked. Actionjackson started his random debut in 1901. He is currently up to 1962 or thereabouts. He has used 1984 for a base year for the entire history of his league. In this thread, he has posted screenshots of several of his leaderboards showing the results we are used to with OOTP16.

I would love to find out that it's me causing the problems.