Player Accuracy

ARockieInMaine · 05-23-2009, 05:04 PM

After I download the game, am I supposed to download a separate player database in order to get accurate player data? My players are wildly inaccurate. I do not mean the results are not identical to how the season actually played out. I mean the player data is so inaccurate as to make realistic results well outside the realm of probability. A few examples are Willie Taveras, E. Y. Jr. and Corey Wimberly.

Willie T. was setting or threatening records for infield hits and bunt hits in 2008, from the right-handed side of the plate for heaven's sake. My OOTP "Willie T." is nothing like this at all. His OOTP rating for bunt hits is 48 out of 250. He had an above average arm for a center fielder and fielded the position well with great range. His OOTP ratings are Outfield Range 158 out of 250, Outfield Arm 155 out of 250, Outfield Error 146 out of 250 and a resulting Outfield Rating of 139 out of 200. (He did have 7 errors in 2008, an out of charateristic number, probably due to the size of center field in Coors Field combined with Willie's great range. Just a guess.) His Stealing Ability is 165 out of 250. He actually stole 68 bases and was caught just 7 times, a 90.67% success rate. In the three previous years, he stole 101 and was caught 29 times, a 77.67% success rate. His strengths have not just been diminished, they have been made into weaknesses.

The cases are similar with E.Y.J. (2008: 46 sb, 16 cs, 74.19%; 2006-2008: 306 sb, 65 cs, 82.70%) and Wimberly (2008: 59 sb, 16 cs, 78.67%; lifetime: 181 sb, 54 cs, 77.02%). Their OOTP ratings are way off: EY, Jr. Running Speed 11, Stealing Ability 13; Corey Wimberly Running Speed 7, Stealing Ability 6!

My point is this, I cannot expect lifelike results from these ratings. It has nothing to do with the interplay of probabilities and individual trials. But, if these players are this far off, how am I supposed to know how far off the ratings are for all the other players in the game's universe? Why should I even have to have this question in mind as I play the game? What I think I "know" about players is bound to steer me wrong.

Is there any easy fix for this?

If not, my opinion of OOTP goes from 221 to 100. (That is 4.5 stars to 2 stars for those keeping track at home.)

Thanks.
Sincerely,
Disappointed

Thanks.

Kelric · 05-23-2009, 06:48 PM

First, understand that OOTP isn't really created to exactly replicate real life. It's bread and butter is fictional league play. Always has been. It is best to leave pre-conceived notions of players at the door when you're playing OOTP. Nobody is going to be exactly like they are in real life.

Second, you can always edit these players yourself. Or go to the Mod forum or check out Padres Fan's place.

Wimberly, a career minor leaguer, isn't going to be high on the roster makers' priority list to get right. Same with Eric Young.

OOTP is incredibly customizable. If you don't like something, you can usually change it yourself.

ARockieInMaine · 05-23-2009, 07:53 PM

You have got to be kidding. Baseball is all about the numbers. Of course one wants one's opinions of real players to be relevant in a baseball sim that uses real player names. I can't edit 10s of thousands of players ratings in this lifetime. That is what computers are for. Surely, OOTP produces its ratings by running an algorithm on stats. My question is, "Why don't they have a better algorithm?"

Or, am I confused. Do they not use real stats? If they use real stats, then there is no reason for inaccuracies in ratings. I cannot believe a bunch of baseball fanatics are so blase about their favorite players and teams. Someone must have addressed this problem. It is not that hard. Get the stats. Run algorithm that converts 75% steal success rate to 188 out of 250, or whatever is appropriate given the underlying game engine (not necessarily a linear scale). Stick it in a database. Etc.

I was under the impression that the algorithm actually used was much more sophisticated, taking into account the sample size (number of steal attempts), historical weighting (taking into account the last three years to smooth out anomalous years), player aging, etc. But, why then does it screw up so badly?

By the way, my idea of running a franchise is all about developing players, so minor league players are critical.

You say:

Wimberly, a career minor leaguer, isn't going to be high on the roster makers' priority list to get right. Same with Eric Young.

You are awfully quick to doom EY to the oblivion of the minor leagues at this stage of his career. But, more importantly, the whole point is that no one should have to put a minor leaguer on a roster priority list. All one has to do is crunch all the minor league stats and, presto, all the minor leaguers are correctly rated, including EY.

Don't get me wrong. This is a great game. No. It is a great, great game. It is truly impressive in scope and execution. But, I cannot believe it has been left to be crippled by GIGO.

No. I am sure I must be missing something, a download, a mod, a database, something.

TribeFanInNC · 05-23-2009, 08:04 PM

Quote:

Originally Posted by ARockieInMaine

Surely, OOTP produces its ratings by running an algorithm on stats. My question is, "Why don't they have a better algorithm?"

The algorithm's forum name is CubbyFan. He does an incredible job. But, one man cannot possibly rate a handful of ratings for 1000s of players correctly. It isn't just crunch stats in a formula, especially for the minor leaguers who have small sample sizes of stats from all different levels of competition. He takes projections using MLEs from multiple sources to come up with ratings. I would say it is an impossible task, yet he does as good a job as could be dreamed of.

I would agree with Kelric about the minor leaguers and 2008 was an anomolous season for Taveras, who's SB totals were in the 30s for the other seasons. He's not that great a player. If he was good, he wouldn't be on like his 4th team.

As Kelric pointed out, if you don't like the ratings, change them. It's your game.

Kelric · 05-23-2009, 08:52 PM

There will be fluctuations in OOTP as well. If you run the same season ten times, the guy might steal 20 bases or 70. Or get hurt and steal none.

ARockieInMaine · 05-23-2009, 11:12 PM

Quote:

Originally Posted by Kelric

There will be fluctuations in OOTP as well. If you run the same season ten times, the guy might steal 20 bases or 70. Or get hurt and steal none.

Thanks, but I am concerned with how the results are created, not with the results per se.

Spritze · 05-24-2009, 02:32 PM

Perhaps you are looking for something more like these as far as ratings go?
Taveras SH=219 Bunt Hits =250 (max) OF Range=243 OF Arm =133

I really think this set of numbers is way more lifelike. And they come from the same stats OOTP uses (Lahman). I just use my own and I think way more accurate AlGoreRythum.

Spritze · 05-24-2009, 02:37 PM

Here this feller is.

A computer generated these numbers for 17,468 players in just a few minutes.

Spritze · 05-24-2009, 02:47 PM

Note that Eric Young has the range but not the arm (OF arm = 70 avg =100 avg RF arm = 137) to play RF. Also note that his infield Arm is better than his OF Arm but still not enough for him to play much SS or 3B. 2B is the spot for him in the infield.

The AlGoreRythum is set for 100 to be the average score (1871-2008). I do allow the score to be greater than the 250 max but I simply output 250 as the score in that case during export.

SteveP · 05-24-2009, 06:08 PM

There's a lot going on in this thread, but I'll try my own tack on all of this:

1. If you are playing the MLB Quickstart game (the one that allows you to play the 2008 season) you cannot judge what you see based on stats from the actual 2008 season. That Quickstart was created prior to the start of the 2008 season. In essence you are getting 2007 players organized into teams as they were at the start of the 2008 season, playing the 2008 schedule. This is the way it has always been with games using actual player stats to play an upcoming season. I think it unlikely that this will be substantially different for OOTP10 for playing the 2009 season.

2. You can't draw definitive conclusions by looking at a specific player's ratings in isolation. The game uses the ratings in a relativistic way. The game is designed to produce reasonable league wide results (total hits, HRs, SBs, etc.) and, in a way involving probability, "distributes" those results among players based on how their ratings relate to the ratings of everyone else in the league. The only way you can really judge what's going on is by analyzing stats at the end of the season -- and then the only valid comparison would be the 2007 stats.

3. The game does not use actual minor league stats to generate any ratings because there are no minor league stats in the Lahman DB. A lot of stuff has been worked in (as alluded to in this thread) by people to try to fill that gap, but it's something you would have to investigate pretty carefully. I am under the impression that the comparable Quickstart game that ships with OOTP10 may have some significant improvements in this area, but I have no reason to believe that the basic fact will change: minor league stats won't be built into the game for producing ratings for minor leaguers.

4. OOTP has basically punting on bunt-for-hit (to mix sports metphors

). It's not worth getting worked up about. I really doubt that this will be any different in OOTP10 (there are some workarounds, but they are difficult and only marginally satisfactory -- won't get into that here).

5. The idea that OOTP is really about playing fictional is certainly news to me.

I guess that ignorance is bliss.

Kelric · 05-24-2009, 07:30 PM

Quote:

Originally Posted by SteveP

5. The idea that OOTP is really about playing fictional is certainly news to me.

I guess that ignorance is bliss.

I almost never play fictional (unless a MLB game goes past the DB years), but from everything I've learned over the years OOTP runs best when we aren't trying to make it conform to real life.

SteveP · 05-24-2009, 08:06 PM

Quote:

Originally Posted by Kelric

I almost never play fictional (unless a MLB game goes past the DB years), but from everything I've learned over the years OOTP runs best when we aren't trying to make it conform to real life.

Yeah, I think the last phrase is the key. I keep telling people that OOTP is alternative history, no matter how you cut it. OOTP succeeds for me when what I see is plausible alternative history. I believe OOTP10 will be more plausible than OOTP9, though it will still have holes (e.g., bunt for hit). I have analyzed fictional a little bit, and I could identify some significant issues there, but it's easier for people to live with because it's a completely fantasy world.

Afino · 05-24-2009, 08:08 PM

Quote:

Originally Posted by Kelric

I almost never play fictional (unless a MLB game goes past the DB years), but from everything I've learned over the years OOTP runs best when we aren't trying to make it conform to real life.

Which is also why I stick to the fictional. I'd get too upset when it doesn't conform to real life

Spritze · 05-24-2009, 08:42 PM

Quote:

Originally Posted by ARockieInMaine

I am sure I must be missing something, a download, a mod, a database, something.

No, I do not think you are missing anything.

My point is simply this. If you don't like what the game and/or the roster team produces for ratings Markus has made it easy for you to Export the Rosters (and Ratings), change them to whatever you would like, and then to Import your own ratings.

I find my own ratings play better for me than those made up by the roster team so that is what I use. Markus has given the user the tools to do this for himself any time he wants and for any reason. You can't ask for more than that. Choose your own AlGoreRythum and salley forth old sot!!!!!!

bigMatt · 06-03-2009, 03:00 PM

perhaps the OP needs to adjust the random rating. I was pretty frustrated with OOTP until I discovered this and dialed it way down. All of a sudden, Frank Robinson didn't suck anymore.

ARockieInMaine · 06-03-2009, 08:28 PM

Quote:

Originally Posted by Spritze

No, I do not think you are missing anything.

My point is simply this. If you don't like what the game and/or the roster team produces for ratings Markus has made it easy for you to Export the Rosters (and Ratings), change them to whatever you would like, and then to Import your own ratings.

I find my own ratings play better for me than those made up by the roster team so that is what I use. Markus has given the user the tools to do this for himself any time he wants and for any reason. You can't ask for more than that. Choose your own AlGoreRythum and salley forth old sot!!!!!!

Spritze,

Your responses intrigue me the most. But, I cannot find instructions on how to do what you suggest. In fact, somewhere in the forum I read that such manipulations couldn't be imported back into OOTP, which left me sort of confused on the point of generating them. I have also read that far from being easy "no one has been able to crack OOTP's database layout." (a rough quote from memory)

I would love to know how to do what you do. Please, tell me more (or, rather, tell me all).

I would also love to know how to start a sim before the most current season (before 2008 in OOTP 9, before 2009 in OOTP X). I want to be the general manager, to put together the team. It is seriously frustrating to start the sim on the day before the season starts, when all the other teams have maxed out their budgets, etc. Totally ties one's hands. Also, I want to sign Hoffman as a free agent.

Is there a way to start after a World Series (or whenever the "off season" begins), with the historical rosters and budgets of the time? I can only find how to set the year, not the date. If I set the year, I start out on 01/01, but there are no free agents.

I have thought about starting a sim earlier than I really wanted to play and letting it run to the appropriate date, but I think there would be a bunch of injuries and trades I would have to track down and manually undo. Is this correct?

Thanks for all your help.
Charles

05-23-2009, 05:04 PM	#1
ARockieInMaine Bat Boy Join Date: May 2009 Location: Portland, ME Posts: 19	Player Accuracy After I download the game, am I supposed to download a separate player database in order to get accurate player data? My players are wildly inaccurate. I do not mean the results are not identical to how the season actually played out. I mean the player data is so inaccurate as to make realistic results well outside the realm of probability. A few examples are Willie Taveras, E. Y. Jr. and Corey Wimberly. Willie T. was setting or threatening records for infield hits and bunt hits in 2008, from the right-handed side of the plate for heaven's sake. My OOTP "Willie T." is nothing like this at all. His OOTP rating for bunt hits is 48 out of 250. He had an above average arm for a center fielder and fielded the position well with great range. His OOTP ratings are Outfield Range 158 out of 250, Outfield Arm 155 out of 250, Outfield Error 146 out of 250 and a resulting Outfield Rating of 139 out of 200. (He did have 7 errors in 2008, an out of charateristic number, probably due to the size of center field in Coors Field combined with Willie's great range. Just a guess.) His Stealing Ability is 165 out of 250. He actually stole 68 bases and was caught just 7 times, a 90.67% success rate. In the three previous years, he stole 101 and was caught 29 times, a 77.67% success rate. His strengths have not just been diminished, they have been made into weaknesses. The cases are similar with E.Y.J. (2008: 46 sb, 16 cs, 74.19%; 2006-2008: 306 sb, 65 cs, 82.70%) and Wimberly (2008: 59 sb, 16 cs, 78.67%; lifetime: 181 sb, 54 cs, 77.02%). Their OOTP ratings are way off: EY, Jr. Running Speed 11, Stealing Ability 13; Corey Wimberly Running Speed 7, Stealing Ability 6! My point is this, I cannot expect lifelike results from these ratings. It has nothing to do with the interplay of probabilities and individual trials. But, if these players are this far off, how am I supposed to know how far off the ratings are for all the other players in the game's universe? Why should I even have to have this question in mind as I play the game? What I think I "know" about players is bound to steer me wrong. Is there any easy fix for this? If not, my opinion of OOTP goes from 221 to 100. (That is 4.5 stars to 2 stars for those keeping track at home.) Thanks. Sincerely, Disappointed Thanks.

05-23-2009, 06:48 PM	#2
Kelric Hall Of Famer Join Date: Mar 2002 Location: Greater Boston Area Posts: 3,992	First, understand that OOTP isn't really created to exactly replicate real life. It's bread and butter is fictional league play. Always has been. It is best to leave pre-conceived notions of players at the door when you're playing OOTP. Nobody is going to be exactly like they are in real life. Second, you can always edit these players yourself. Or go to the Mod forum or check out Padres Fan's place. Wimberly, a career minor leaguer, isn't going to be high on the roster makers' priority list to get right. Same with Eric Young. OOTP is incredibly customizable. If you don't like something, you can usually change it yourself. Last edited by Kelric; 05-23-2009 at 06:51 PM.

05-24-2009, 02:32 PM	#7
Spritze OOTP Historical Czar Join Date: Dec 2001 Location: Bothell Wa Posts: 7,253	Perhaps you are looking for something more like these as far as ratings go? Taveras SH=219 Bunt Hits =250 (max) OF Range=243 OF Arm =133 I really think this set of numbers is way more lifelike. And they come from the same stats OOTP uses (Lahman). I just use my own and I think way more accurate AlGoreRythum. Attached Images

05-24-2009, 02:37 PM	#8
Spritze OOTP Historical Czar Join Date: Dec 2001 Location: Bothell Wa Posts: 7,253	Eric Young Here this feller is. A computer generated these numbers for 17,468 players in just a few minutes. Attached Images

05-24-2009, 02:47 PM	#9
Spritze OOTP Historical Czar Join Date: Dec 2001 Location: Bothell Wa Posts: 7,253	Sorry but I did not rate the esteemed Mr. Wimberly Note that Eric Young has the range but not the arm (OF arm = 70 avg =100 avg RF arm = 137) to play RF. Also note that his infield Arm is better than his OF Arm but still not enough for him to play much SS or 3B. 2B is the spot for him in the infield. The AlGoreRythum is set for 100 to be the average score (1871-2008). I do allow the score to be greater than the 250 max but I simply output 250 as the score in that case during export.

05-23-2009, 07:53 PM	#3
ARockieInMaine Bat Boy Join Date: May 2009 Location: Portland, ME Posts: 19	You have got to be kidding. Baseball is all about the numbers. Of course one wants one's opinions of real players to be relevant in a baseball sim that uses real player names. I can't edit 10s of thousands of players ratings in this lifetime. That is what computers are for. Surely, OOTP produces its ratings by running an algorithm on stats. My question is, "Why don't they have a better algorithm?" Or, am I confused. Do they not use real stats? If they use real stats, then there is no reason for inaccuracies in ratings. I cannot believe a bunch of baseball fanatics are so blase about their favorite players and teams. Someone must have addressed this problem. It is not that hard. Get the stats. Run algorithm that converts 75% steal success rate to 188 out of 250, or whatever is appropriate given the underlying game engine (not necessarily a linear scale). Stick it in a database. Etc. I was under the impression that the algorithm actually used was much more sophisticated, taking into account the sample size (number of steal attempts), historical weighting (taking into account the last three years to smooth out anomalous years), player aging, etc. But, why then does it screw up so badly? By the way, my idea of running a franchise is all about developing players, so minor league players are critical. You say: Wimberly, a career minor leaguer, isn't going to be high on the roster makers' priority list to get right. Same with Eric Young. You are awfully quick to doom EY to the oblivion of the minor leagues at this stage of his career. But, more importantly, the whole point is that no one should have to put a minor leaguer on a roster priority list. All one has to do is crunch all the minor league stats and, presto, all the minor leaguers are correctly rated, including EY. Don't get me wrong. This is a great game. No. It is a great, great game. It is truly impressive in scope and execution. But, I cannot believe it has been left to be crippled by GIGO. No. I am sure I must be missing something, a download, a mod, a database, something.

05-23-2009, 08:52 PM	#5
Kelric Hall Of Famer Join Date: Mar 2002 Location: Greater Boston Area Posts: 3,992	There will be fluctuations in OOTP as well. If you run the same season ten times, the guy might steal 20 bases or 70. Or get hurt and steal none.

05-24-2009, 06:08 PM	#10
SteveP Moderator Join Date: Aug 2008 Posts: 3,109	There's a lot going on in this thread, but I'll try my own tack on all of this: 1. If you are playing the MLB Quickstart game (the one that allows you to play the 2008 season) you cannot judge what you see based on stats from the actual 2008 season. That Quickstart was created prior to the start of the 2008 season. In essence you are getting 2007 players organized into teams as they were at the start of the 2008 season, playing the 2008 schedule. This is the way it has always been with games using actual player stats to play an upcoming season. I think it unlikely that this will be substantially different for OOTP10 for playing the 2009 season. 2. You can't draw definitive conclusions by looking at a specific player's ratings in isolation. The game uses the ratings in a relativistic way. The game is designed to produce reasonable league wide results (total hits, HRs, SBs, etc.) and, in a way involving probability, "distributes" those results among players based on how their ratings relate to the ratings of everyone else in the league. The only way you can really judge what's going on is by analyzing stats at the end of the season -- and then the only valid comparison would be the 2007 stats. 3. The game does not use actual minor league stats to generate any ratings because there are no minor league stats in the Lahman DB. A lot of stuff has been worked in (as alluded to in this thread) by people to try to fill that gap, but it's something you would have to investigate pretty carefully. I am under the impression that the comparable Quickstart game that ships with OOTP10 may have some significant improvements in this area, but I have no reason to believe that the basic fact will change: minor league stats won't be built into the game for producing ratings for minor leaguers. 4. OOTP has basically punting on bunt-for-hit (to mix sports metphors ). It's not worth getting worked up about. I really doubt that this will be any different in OOTP10 (there are some workarounds, but they are difficult and only marginally satisfactory -- won't get into that here). 5. The idea that OOTP is really about playing fictional is certainly news to me. I guess that ignorance is bliss.

06-03-2009, 03:00 PM	#15
bigMatt Hall Of Famer Join Date: Dec 2005 Location: Kansas Posts: 2,378	perhaps the OP needs to adjust the random rating. I was pretty frustrated with OOTP until I discovered this and dialed it way down. All of a sudden, Frank Robinson didn't suck anymore.