Accuracy

[LansdowneSt]

Originally Posted by Edward ohare

You didn't answer the question you invited. Why not just post a screen shot of the stats including the requested years? That's probably less effort than typing out a post.

Brad,

In post #6, he says it's a "Ratings recalculation base: 5 years" game so career is the better measure.

If you still own the game, just post your results.

Knock yourself out.

[Rain King]

Default Replay Mode. Real Lineups & Transactions

Ray Knight, 1979: .318/.360/.454/.814/133 (Real) vs. .325/.373/.463/.837/139 (OOTP)
Ray Knight, 1982: .294/.344/.402/.746/108 (Real) vs. .310/.346/.419/.765/113 (OOTP)
Ray Knight, 1987: .256/.310/.373/.683/81 (Real) vs. .264/.325/.382/.707/87 (OOTP)

Andres Galarraga, 1987: .305/.361/.459/.821/114 (Real) vs. .305/.349/.474/.823/114 (OOTP)
Andres Galarraga, 1996: .304/.357/.601/.958/140 (Real) vs. .298/.352/.604/.956/130 (OOTP)

Miguel Cabrera, 2006: .339/.430/.568/.998/153 (Real) vs. .355/.436/.593/1.029/161 (OOTP)
Miguel Cabrera, 2011: .344/.448/.586/1.033/182 (Real) vs. .365/.464/.599/1.063/190 (OOTP)
Miguel Cabrera, 2015: .338/.440/.534/.974/167 (Real) vs. .315/.397/.549/.946/157 (OOTP)
Miguel Cabrera, 2019: .282/.346/.398/.744/98 (Real) vs. .243/.297/.332/.629/68 (OOTP)

Evan Longoria, 2010: .294/.372/.507/.879/138 (Real) vs. .281/.372/.498/.870/136 (OOTP)
Evan Longoria, 2014: .253/.320/.404/.724/105 (Real) vs. .241/.297/.376/.673/90 (OOTP)
Evan Longoria, 2019: .254/.325/.437/.762/100 (Real) vs. .252/.318/.453/.771/102 (OOTP)

[jcard]

Originally Posted by LansdowneSt

Brad,

Not implying a position on the substantive point in contention in this thread, but I was wondering this exact same thing.

[BaseballMan]

At what point are stats accurate?
I prefer having accurate rosters and starting lineups
but i'm not sure i would want Hank Aaron to hit 755 home runs every sim.
However i wouldn't like him hitting only 200 homers either.
I try not to get to the point where i concentrate too much on getting perfect accurate stats
and don't enjoy the league history of the sim.

For the most part i usually see the good teams in mlb history being the good teams in my sims.
Same with players.
I will give credit to all the work done on the historical database and lineups for that.

There are times when teams and players do a little better or a little worse but isn't that the
point of playing a sim? To see what will happen?
If you want everything to be exactly as real life then whats the point of playing? Might as well read Total Baseball.
I can understand not liking Pete Rose getting 3,000 hits with no injuries but if he gets only 4,100 hits i won't be too concerned with it.

[Edward ohare]

Originally Posted by LansdowneSt

Brad,

In post #6, he says it's a "Ratings recalculation base: 5 years" game so career is the better measure.

If you still own the game, just post your results.

Knock yourself out.

Rain King understands the discussion is about replay mode with one year recalculation. See his posts.



And you're being silly.

[Edward ohare]

Originally Posted by Rain King

Default Replay Mode. Real Lineups & Transactions

Ray Knight, 1979: .318/.360/.454/.814/133 (Real) vs. .325/.373/.463/.837/139 (OOTP)
Ray Knight, 1982: .294/.344/.402/.746/108 (Real) vs. .310/.346/.419/.765/113 (OOTP)
Ray Knight, 1987: .256/.310/.373/.683/81 (Real) vs. .264/.325/.382/.707/87 (OOTP)

Andres Galarraga, 1987: .305/.361/.459/.821/114 (Real) vs. .305/.349/.474/.823/114 (OOTP)
Andres Galarraga, 1996: .304/.357/.601/.958/140 (Real) vs. .298/.352/.604/.956/130 (OOTP)

Miguel Cabrera, 2006: .339/.430/.568/.998/153 (Real) vs. .355/.436/.593/1.029/161 (OOTP)
Miguel Cabrera, 2011: .344/.448/.586/1.033/182 (Real) vs. .365/.464/.599/1.063/190 (OOTP)
Miguel Cabrera, 2015: .338/.440/.534/.974/167 (Real) vs. .315/.397/.549/.946/157 (OOTP)
Miguel Cabrera, 2019: .282/.346/.398/.744/98 (Real) vs. .243/.297/.332/.629/68 (OOTP)

Evan Longoria, 2010: .294/.372/.507/.879/138 (Real) vs. .281/.372/.498/.870/136 (OOTP)
Evan Longoria, 2014: .253/.320/.404/.724/105 (Real) vs. .241/.297/.376/.673/90 (OOTP)
Evan Longoria, 2019: .254/.325/.437/.762/100 (Real) vs. .252/.318/.453/.771/102 (OOTP)

Thank you. I'm gratified to see some of your BA data i outside the +19 and -20 point range mentioned by Garlon. I don't recall that ever happening on annual data he posts. Good job.

I know this was a lot of work due to the non functionality of sortable stats. They should allow the display of in game stats and historical stats side by side.

[Edward ohare]

Originally Posted by BaseballMan

At what point are stats accurate?
I prefer having accurate rosters and starting lineups
but i'm not sure i would want Hank Aaron to hit 755 home runs every sim.
However i wouldn't like him hitting only 200 homers either.
I try not to get to the point where i concentrate too much on getting perfect accurate stats
and don't enjoy the league history of the sim.

For the most part i usually see the good teams in mlb history being the good teams in my sims.
Same with players.
I will give credit to all the work done on the historical database and lineups for that.

There are times when teams and players do a little better or a little worse but isn't that the
point of playing a sim? To see what will happen?
If you want everything to be exactly as real life then whats the point of playing? Might as well read Total Baseball.
I can understand not liking Pete Rose getting 3,000 hits with no injuries but if he gets only 4,100 hits i won't be too concerned with it.

Bingo.

[Rain King]

Originally Posted by Edward ohare

I know this was a lot of work due to the non functionality of sortable stats. They should allow the display of in game stats and historical stats side by side.

Nah, it is pretty easy to run a simulation and look at/record stuff.

Do you not have any comments about "accuracy" since that is why you seemingly specifically requested this?

[Edward ohare]

Originally Posted by Rain King

Nah, it is pretty easy to run a simulation and look at/record stuff.

Do you not have any comments about "accuracy" since that is why you seemingly specifically requested this?

You removed my comments about accuracy in your quote and now you ask me if I have any comments about accuracy. I suppose you can hope the fast readers won't notice that deception.

[Rain King]

Originally Posted by Edward ohare

Thank you. I'm gratified to see some of your BA data i outside the +19 and -20 point range mentioned by Garlon. I don't recall that ever happening on annual data he posts. Good job.

This is the part I didn't quote, I honestly wasn't sure what you meant by the BA stuff so didn't want to further confuse things..and the Garlon comment was irrelevant.

Here are your quotes from earlier in the thread I'm wondering if you are planning to clarify.

Originally Posted by Edward ohare

OOTP is not designed to optimize the accuracy of individual player performance. It is designed to optimize the accuracy of league wide performance.

Originally Posted by Edward ohare

"Actually, the game is designed to produce accurate results for players and for the league."

Accurate player results? Over what time span? Certainly not for a season. Unless they repealed the law of large numbers or put program restrictions on randomness.

What is "accurate" to you and how does OOTP not prioritize that for player results?

[twins_34]

Originally Posted by Edward ohare

You removed my comments about accuracy in your quote and now you ask me if I have any comments about accuracy. I suppose you can hope the fast readers won't notice that deception.

I have seen Garlon give you tons and tons of proof of how good OOTP does with historical accuracy. You threw out names, he threw them same names at you with the numbers he got. I mean, Brad, if you have these questions, run the sims yourself. You have been here long enough to know how OOTP works and how to set up games, yet you are in here being a troll to everyone that answers the question. Either prove them wrong with your proof or just stop the trolling.

[Edward ohare]

Following is BA data on my 1995 replay. It includes all the players with 512 or more ABs and they are all from the same save. All the players. One save. Verifiable by the screenshots.

512 ABs was selected because that was the number I used in a question. The answer to the question provided a BA range of -20 and +19. The pdf is a spreadsheet showing OOTP BA, actual BA, and the difference. Differences of +19 or more or -20 or more are marked with an X. XX indicates a difference of at least twice +19 or -20.

[Rain King]

What mode/settings were used for your simulation?

There are 61 players in your sheet. You have X's for 9 of them (i.e. 14.8%) and XX's for 3 of them (4.9%).

Why did you choose the ranges you chose? Are you trying to make a statement about "accuracy" of some kind with this information?

To me, with some understanding of how statistical variance works, this all seems to be very good?

[Reed]

Originally Posted by Edward ohare

Following is BA data on my 1995 replay. It includes all the players with 512 or more ABs and they are all from the same save. All the players. One save. Verifiable by the screenshots.

512 ABs was selected because that was the number I used in a question. The answer to the question provided a BA range of -20 and +19. The pdf is a spreadsheet showing OOTP BA, actual BA, and the difference. Differences of +19 or more or -20 or more are marked with an X. XX indicates a difference of at least twice +19 or -20.

I am glad you posted your results. Are you using Garlon’s suggested settings?
5 year recalc. In my experience, Garlon’s recommendations are fantastic for “career stats”. The only con is that it flattens out individuals outlier seasons. Example would be almost impossible for Maris to hit 61 HRs.
1 year recalc.. Again in my experience. With 1 year recalc. you are much more likely to see a player’s up/down seasons but you are also more likely to see extreme outlier seasons for players. For example it would be easier for Maris to hit 65+ HRs using 1 year recalc.. Career stats may not be as reliable as using Garlon’s settings.
3 year recalc.. To me this is a compromise between 1 year and 5 year recalc.. It is still possible for Maris to hit 61 HRs. Not as likely as using 1 year recalc but more likely than using 5 year recalc.. Career stats maybe not as good as 5 year recalc but probably better than 1 year recalc..

Again, this is just my experience and someone else’s experience may be different. There are pros and cons to each IMO so pick the one you want and TEST IT OUT.

[Matt Arnold]

Originally Posted by Rain King

What mode/settings were used for your simulation?

There are 61 players in your sheet. You have X's for 9 of them (i.e. 14.8%) and XX's for 3 of them (4.9%).

Why did you choose the ranges you chose? Are you trying to make a statement about "accuracy" of some kind with this information?

To me, with some understanding of how statistical variance works, this all seems to be very good?

Given that a pure binomial calculation would have around 30% outside those ranges, the fact that we come in with less than 20% is arguably too accurate.

[Rain King]

Originally Posted by Reed

I am glad you posted your results. Are you using Garlon’s suggested settings?
5 year recalc. In my experience, Garlon’s recommendations are fantastic for “career stats”. The only con is that it flattens out individuals outlier seasons. Example would be almost impossible for Maris to hit 61 HRs.
1 year recalc.. Again in my experience. With 1 year recalc. you are much more likely to see a player’s up/down seasons but you are also more likely to see extreme outlier seasons for players. For example it would be easier for Maris to hit 65+ HRs using 1 year recalc.. Career stats may not be as reliable as using Garlon’s settings.
3 year recalc.. To me this is a compromise between 1 year and 5 year recalc.. It is still possible for Maris to hit 61 HRs. Not as likely as using 1 year recalc but more likely than using 5 year recalc.. Career stats maybe not as good as 5 year recalc but probably better than 1 year recalc..

Again, this is just my experience and someone else’s experience may be different. There are pros and cons to each IMO so pick the one you want and TEST IT OUT.

One of the true beauties of OOTP on display in this post. There isn't a game that comes anywhere close to giving you the number of options/settings for customizing up your historical/replay world.

[Edward ohare]

Originally Posted by Reed

I am glad you posted your results. Are you using Garlon’s suggested settings?
5 year recalc. In my experience, Garlon’s recommendations are fantastic for “career stats”. The only con is that it flattens out individuals outlier seasons. Example would be almost impossible for Maris to hit 61 HRs.
1 year recalc.. Again in my experience. With 1 year recalc. you are much more likely to see a player’s up/down seasons but you are also more likely to see extreme outlier seasons for players. For example it would be easier for Maris to hit 65+ HRs using 1 year recalc.. Career stats may not be as reliable as using Garlon’s settings.
3 year recalc.. To me this is a compromise between 1 year and 5 year recalc.. It is still possible for Maris to hit 61 HRs. Not as likely as using 1 year recalc but more likely than using 5 year recalc.. Career stats maybe not as good as 5 year recalc but probably better than 1 year recalc..

Again, this is just my experience and someone else’s experience may be different. There are pros and cons to each IMO so pick the one you want and TEST IT OUT.

I used out of the box replay settings.

[Edward ohare]

Originally Posted by Matt Arnold

Given that a pure binomial calculation would have around 30% outside those ranges, the fact that we come in with less than 20% is arguably too accurate.

Of course, small sample size. However posting data from lots of players from one save is more informative than a lot of what has been posted in the past.

[Syd Thrift]

Originally Posted by Edward ohare

Of course, small sample size. However posting data from lots of players from one save is more informative than a lot of what has been posted in the past.

Right, but your supposedly illumunative small sample is literally producing the opposite of what you’re claiming. I realize that what you actually want but probably won’t say with the new ban evading account is that OOTP takes hits away from players who are over performing their historicals but only a bad simulation would do that, so you are once again stuck writing your own (bad) game if that’s what you want.

[uruguru]

Originally Posted by Edward ohare

If there is a player who has 135 hits in 512 ABs for a .264 BA and hit 20 doubles, 11 HRs, and walked 69 times what is the greatest amount he would deviate from that, plus and minus, in your game?

Plus or minus 50 points in batting average should not shock you.

Sometimes you will have a guy who hit 25 homers in 100 games played end up playing a full season and hitting 60 homers.

There is a tremendous amount of RNG in this game.