Data update only for 26

[Brad K]

Only update data for 26. Use the extra year to go way back and redo not just patch.

1. Make the rating system accurate. A league of historical players with historical lineups and historical transactions, rated on their historical production, should produce realistic results, both for the players and the league, without the use of LTMs. The current need for not only LTMs but LTMs tested against the league and then adjusted shows the rating system is not accurate.

2. A historical what if league where the talent is different from the historical league should perform according to the talent present and not be hammered by LTMs into producing historical stats.

3. If retire according to history is not selected then some players should have their careers shortened to compensate for those who play longer.

4. Real stats should be neutralized against the game's park factors or, if they already are as some suspect, then that should be confirmed by devs.

5. Choices that don't apply to historical games should be grayed out. Effect of coaching, development budget, and psychobabble etc shouldn't apply to historical players' performance. They should perform in the ballpark of how they did historically based on randomness and the selection of TCR.

6. Changes in ratings due to the adjust/weaken feature should be kept within the group of players adjusted or weakened.

7. Get rid of the hit and run slider for manager strategy. It appears to be not connected. In watching games I've never observed a situation where an AI manager started a runner. Actually, having it not working makes sense. There isn't any data on H&R plays that can be used by the game and the results are already reflected in data on base runners taking an extra base, SB, CS, and double plays.

8. The fact that minor league players can't be neutralized isn't a reason not to include them in the neutralized data base. Their ratings are all a guess anyway, the stats they produced in the minors are depreciated for their MLB ratings, so the minor changes made by neutralization are lost in the other changes.

9. Treat a game with fictional players as a separate different game from one with historical players. For example, the idea of pitch types affecting a fictional player's development and ultimate effectiveness is fine for games with fictional players but pitch type ratings for historical players, who cares? It's all made up stuff by going backwards from the real life stats.

10. Lose the randomness purity ring. It's been stated that the game engine helps prolong both slumps and hot streaks but not enough that it's noticeable, presumably referred to anecdotal observation. So put in floors and caps to prevent completely unreasonable performances. Surely that can be done without making in noticeable! Besides, with the slump and hot streak thing, and altering real stats to lower home run totals, the randomness vow is already broken!


MS built on patched DOS for a long time and eventually had to start anew. It's time for a new OOTP engine.

For a long time a PC would show a notice "EMS memory is being used to simulate XMS memory..." which was a patch on a patch and none of it was actually real memory. OOTP is well beyond that.

[Brad K]

11. Make it clear which features are show only and shouldn't be considered when making decisions. As examples, spray charts, exit velocity, and hard out, all of which have been mentioned recently in this forum as the basis for decisions, And think about the subliminal usually incorrect message being sent to players who play out games by the randomly varying crack of the bat on the ball?

[Garlon]

1. The rating system is accurate. The league totals modifiers calibrate the league so that every parameter that can be modeled from historical stats also turns out accurately.

2. You can turn off the league totals modifiers if you want.

3. If there are only 25 roster spots then sone players will not play anyway.

4. You can do a test if this yourself with a neutral park factors file and load the same season and see if the ratings are any different.

5. The game is a sandbox and sone people may want to use those options.

6. This is how it works currently.

7. This is true that we do not have historical data on H & R. Whether or not the sliders work needs to be confirmed. They may also stack against the league setting or perhaps not be able to go higher or lower than the maximum or minimum settings.

8. This may be due to the fact that the game is pulling the neutralized stats from a different file and since the minor leagues do not have neutralized stats the game cannot pull from both files simultaneously. A possible workaround if they wanted to support this would be to copy all of the minor league stats over as a neutralized file so that everything can pull from the same set of files.

9. Not exactly sure what you are suggesting. The database does have real pitch types for many historical pitchers and those pitches get ranked and rated to produce the Stuff for pitchers.

10. The game engine does not generate or prolong streaks or slumps artificially. Streaks and slumps occur naturally from probability. When a player is hot or cold in the game you are just seeing a report on how they have been performing recently. This has no affect on how they will continue to perform.

[coolfish2]

Originally Posted by Garlon

.10. The game engine does not generate or prolong streaks or slumps artificially. Streaks and slumps occur naturally from probability. When a player is hot or cold in the game you are just seeing a report on how they have been performing recently. This has no affect on how they will continue to perform.

I think It's handled by 'Handle Success' and 'Handle Failure' which you can find in editor.

[Brad K]

Originally Posted by Brad

6. Changes in ratings due to the adjust/weaken feature should be kept within the group of players adjusted or weakened.

Originally Posted by Garlon

6. This is how it works currently.

From This thread post 51

Originally Posted by Lukas

So, for example, if you set the 'adjust hitters with fewer than X At-Bat' very low, say at 50 ab's, this doesn't only affect the performance of players with less than 50 ab's.

Because in this case, you're ending up adjusting very few player ratings, it sets a higher overall talent level throughout the league, with players with even 200 ab's not having their ratings adjusted at all.

So this also affects the performance of the higher level players, because more lower level players have higher ratings and that takes more pieces of the pie of how many HR's are distributed league wide and distributes those to the lower level players, which as a result takes away from the pieces of the statistical pie the top players can expect to get.

So, Garlon, the changes in the ratings are not kept within the pool of players being adjusted. Doing so would be the correct way to do it. Presumably the too high and too low ratings cause by inadequate sample size are a net zero and there is no need to involve other players.

Originally Posted by Brad

10. Lose the randomness purity ring. It's been stated that the game engine helps prolong both slumps and hot streaks but not enough that it's noticeable, presumably referred to anecdotal observation. So put in floors and caps to prevent completely unreasonable performances. Surely that can be done without making in noticeable! Besides, with the slump and hot streak thing, and altering real stats to lower home run totals, the randomness vow is already broken!

Originally Posted by Garlon

10. The game engine does not generate or prolong streaks or slumps artificially. Streaks and slumps occur naturally from probability. When a player is hot or cold in the game you are just seeing a report on how they have been performing recently. This has no affect on how they will continue to perform

This thread post 9

Originally Posted by Lukas

I believe once you actually get the indicator, there's an extremely slight boost or drop to their ratings until you get out of the streak or slump but it's not dramatic enough you should be able to discern with the naked eye, just looking at on player's production.

So streaks, good or bad, are helped along by the game. The next question is when a pitcher starts to get shelled in a game, does OOTP help that along? Because AI managers appear to pull pitchers based on in game performance when, if it was random, they would make the decision on fatigue.

I decided to respond to just two because responding to all at once would result in too many subjects to discuss at one time. So let's go over these for now and then I will get to the others.

[Garlon]

The ratings are not really changed, but performance can change based on the pool of players and the league totals.

If you give every team in 1921 a copy of Babe Ruth you should not expect them all to hit 59 HR. In fact, having 16 copies of Ruth in the league would probably knock them each down to like 8 HR or something.

The system is dynamic and accounts for the talent available in the league in order to produce the historical rates.

There is no issue with how this works.

[Brad K]

Originally Posted by Garlon

The ratings are not really changed, but performance can change based on the pool of players and the league totals.

If you give every team in 1921 a copy of Babe Ruth you should not expect them all to hit 59 HR. In fact, having 16 copies of Ruth in the league would probably knock them each down to like 8 HR or something.

The system is dynamic and accounts for the talent available in the league in order to produce the historical rates.

There is no issue with how this works.

You're arguing terminology. If the LTM adjustments don't change ratings what DO they change? It's admitted they change the ratings of the adjusted players. The effect is to adjust the ratings of the non adjusted players. Please stop the word games. Remember, your argument here is against statements made by the devs.

As I observed, a league of players using historical lineups and historical transactions with players rating on their historical stats cannot produce output reasonably close to historical without LTMs that themselves don't work out of the box and have to be tested three times against the league to determine the final numbers to be used. There is no argument to be made that ratings are accurate.

And your comments on a bunch of Babes in a league hitting 8 HRs each, well, that shows that when talent varies from historical, as it will in games played by human GMs, with historical retirement off and development and TCR on, then the output will not match the talent in the league. Such a league is hammered into historical output. So much for alternative history, which is what human GMs want!
*
You didn't address the issue on hot streaks and slumps. Do you accept my conclusion?

[Sweed]

Ask all you want but a data update only is never going to happen.


As to the many "cracks of the bat"? Simple solution, make a copies of whichever sound you like, rename them to match the different cracks so all files are now the same. Voila you have only one sound for batted balls. Post it as a mod and I doubt your site's bandwidth will be able to keep up with the demand.


Hit and run is used by the AI . You will see it on hard ground balls to SS where the only play is to first base, because the runner was in motion.

[Syd Thrift]

I would like the game to keep and show stats for the strategies though. I’m pretty sure one of the teams in my league is doing hit and runs a lot because they have atrocious stolen base rates but I’d love to see the stats broken down.

[Brad K]

Originally Posted by Syd Thrift

I would like the game to keep and show stats for the strategies though. I’m pretty sure one of the teams in my league is doing hit and runs a lot because they have atrocious stolen base rates but I’d love to see the stats broken down.

I think you're right about the low SB % being an indication.

When I called H&R myself I had a SB% well under 50% in the mid 60s. I only called it with a fairly decent SB guy on and a good contact and avoid K hitter at the plate.

Now that I have it on manager control I have 1 hit and run attempt in 41 games. Thread in the 24 general discussion forum.

[Brad K]

Originally Posted by Sweed

Ask all you want but a data update only is never going to happen.

Hey, if they want to hire more staff instead of taking a revenue hit, that's their choice. I just figured they'd be more apt to take the revenue hit.

[Brad K]

Originally Posted by Sweed

Hit and run is used by the AI . You will see it on hard ground balls to SS where the only play is to first base, because the runner was in motion.

Please see my thread in the 24 general discussion forum and explain how I make this happen. Thank you.

[Garlon]

The function of the league totals modifiers is to modulate the probability of outcomes. They do not change the ratings of players. The modifiers tune the game to a given season.

[Brad K]

Semantics. LTMs change the performance of the players. Whether it changes the actual ratings or alters the output in some other way is irrelevant. You're arguing the terms I'm using to describe how it's done when the issue is what is done.

And the point remains the game can't run right without the LTMs changing the ratings, or their output, or whatever you want to call it. Which means, as I've stated, the ratings fashioned from historical data cannot alone produce historical outcomes. That means the ratings aren't accurate.

All these ratings below, created from historical performance data, cannot produce historically accurate output in a save with historical lineups, historical transactions, no development, and TCR turned off, without being tested three times against another file and adjusted after every test.

[Brad K]

But these ratings can produce historically accurate outputs.

Kind of embarrassing, isn't it?

[Garlon]

Please play out 162 games with a card and dice game for all 30 teams in the league (2430 games) and report the following parameters and compare them to how they match up with historical known values from the given season. We await your results.


As a reminder:

New for OOTP25

1. New and more accurate Fielding Range ratings for all players in history.

2. New OF Arm ratings based on advanced historical fielding stats.

3. New Framing ratings for all catchers.

4. New Pitcher Hold Baserunner ratings based on advanced historical statistics.

5. New Baserunning Skills ratings based on advanced historical statistics.

6. Pitcher Splits since 1914 (not yet in the game though). We have Batter Splits since 1901.

7. New Modifier to model the distribution of Airouts between the Outfield and Infield for all seasons.

8. New Modifiers for Sac Bunting to model positions player rates, pitcher rates, and success rates each season.

9. New and improved pitcher BABIP ratings.

10. More accurate R/G each season.




These are the various elements that OOTP models per season and per player.


League Elements Each Season Based on Historical Rates

1. Batting Average

2. BABIP

3. Doubles per non-HR Hit

4. Triples per non-HR Hit

5. HR per AB

6. BB per PA

7. HBP per PA

8. OBP (indirectly by modeling the above elements accurately)

9. SLG (indirectly by modeling the above elements accurately)

10. OPS (indirectly by modeling the above elements accurately)

11. Strikeouts per AB

12. SB per times reaching First Base

13. SB success rate

14. Sac Bunts per Out (technically per Out – K)

15. Sac Bunt rates for position players

16. Sac Bunt rates for pitchers

17. Sac Bunt success rate

18. Sacrifice Flies per Out (technically per Out – K)

19. Double Play Frequency

20. GIDP frequency as a subset of all Double Plays

21. Runs per Game

22. IPouts per Game Started

23. IPouts per Relief Appearance

24. GO% (essentially GO:AO ratio)

25. Distribution of AO between the outfielders and infielders

26. Wild Pitches per Inning

27. Balks per Inning

28. Passed Balls per Inning

29. Pickoffs per Reched First base

30. XBT% (extra bases taken by players on the basepaths)

31. League FLD%

32. P FLD%

33. C FLD%

34. 1B FLD%

35. 2B FLD%

36. 3B FLD%

37. SS FLD%

38. LF FLD%

39. CF FLD%

40. RF FLD%

41. OF Assist rate

42. Ballpark Factors for BABIP, 2B, 3B, HR including LHB/RHB splits for most seasons.




Player Elements

Batting and Offense Based on Historical Player Statistics

1. BABIP

3. Rate of 2B per non-HR Hit

3. Rate of 3B per non-HR Hit

4. HR per AB

5. Strikeouts per AB

6. BB per PA

7. HBP per PA

8. BA (indirectly by modeling the above elements accurately)

9. OBP (indirectly by modeling the above elements accurately)

10. SLG (indirectly by modeling the above elements accurately)

11. OPS (indirectly by modeling the above elements accurately)

12. Baserunning Skills (frequency of taking extra bases)

13. Stealing Aggressiveness (stealing attempts per reaching first base)

14. Stealing Ability (stolen base success rate)

15. Sac Bunting frequency

16. Batting Splits vs RHP/LHP since 1901




Pitching Based on Historical Player Statistics

1. K per BF

2. HR Allowed per BF

3. BB per BF

4. HBP per BF

5. Balks per BF

6. Wild Pitches per BF

7. GO%

8. Pitcher BABIP against

9. Stamina (IPouts per Game Started or IPouts per Relief Appearance)

10. Hold Baserunners (preventing stolen base attempts)

11. Pitching Splits vs RHB/LHB since 1914 (coming soon to historical games)




Fielding and Defense Based on Historical Player Statistics

1. Infielder Range (adjusted for Team Outs in Play, Team GO%, Team Defensive Efficiency, Team Home Ballpark BABIP factor, Team LHP, and Team BABIP against)

2. Infielder Error Rate

3. Infielder Turn DP Rate (adjusted for expected team double plays based on opportunities)

4. Outfielder Range (adjusted for Team Outs in Play, Team GO%, Team Defensive Efficiency, Team Home Ballpark BABIP factor, and Team BABIP against)

5. Outfielder Error Rate

6. Outfielder Arm (assist rate adjusted for opportunities to make an assist)

7. Catcher Blocking (PB and Error rate)

8. Catcher Framing rate

9. Catcher Arm (CS rate)

10. Positional Eligibility

[Sweed]

Originally Posted by Brad K

Please see my thread in the 24 general discussion forum and explain how I make this happen. Thank you.

Not going to waste a lot of my time trying to diagnose your game going back and forth, especially going back to v24. I will ask before leaving...

From your thread am I to take it that the AI turned a DP on every ground ball hit to ss or 2b with a runner on and less than two outs? For the most part there are only two situations where a DP won't be turned (if the fielders are competent). Ball hit to slow to get two. AND the runner was in motion, ie hit and run. IOW if on a gb the SS or 2B is taking the fielder's choice at 1b rather than going for the lead runner it is probably because of a H&R, with the exception of a slowly hit ball.

... and comment on my game.

I don't control my baserunning while playing out games. I think allowing the manager from the dugout to tell a runner to advance or not is one of the most unrealistic things in OOTP, and it's why I hired base coaches **. So I don't set baserunning to user control. That doesn't mean I can't override it and attempt steals, H&R, or R&H. I do send runners or attempt steals as a manger can do from the dugout. With this setting the AI will at times send runners when I haven't.

In all versions I can recall (includes v24 and now 25) the AI has kept me out of DP's by sending my runner.
I've had times where I called H&R and fouled off the pitch. I don't spam the button so if it doesn't work after two tries I go to a normal swing. After two unsuccessful attempts my player will hit a GB. My first thought is "sure now he hits it fair" knowing a DP is coming. BUT will you look at that.. my runner is in motion AND I didn't send him. If it wasn't me that sent him that only leaves the AI.

** That doesn't mean I think it should be removed from the game. If some users like and use it, more power to them.

[Syd Thrift]

ITT: hey devs, next version only work on things I want you to work on

[Brad K]

The discussion on request 6, about handling the redistribution of ratings for adjusted ratings drifted into a discussion about request 1, the inability of the ratings to produce historical output in a perfect historical environment. The move of the discussion to stating all the improvements admits there was a problem. And the fact the game still using triple tested LTMs in 25 shows the problem still exists. So clearly latest feeble defense shows the allegations of problems in requests 1 and 6 are true.

The issue of request 10, hot and cold streaks being prolonged by the game, has been ignored after a post was made quoting a dev saying the game prolonged hot and cold streaks. Unaddressed is whether this deviation from the commitment to randomness (and the alteration of real stats to reduce the peak randomness for HRs) means the principle is moot and the OOTP has no principle to stand on to resist the implementation of floors and ceilings to minimize outlier performances.

So requests 1, 6, and 10 have been shown to be valid. It's time to move on to discussion of the other points.

[Syd Thrift]

lmao what If I told you that 10 is a feature, not a bug