Pre-Defined Draft Values

[actionjackson]

Originally Posted by Spritze

The alphabet thing is merely because all players in modded db's back in those ancient OOTPx days of yore had a draft value of NULL. They were ALL tied for the top rung. Nothing made anything Null and Void. There just weren't no data there at all which is considered a NULL. Voiding is what women do to men after they have emptied the mans bank accounts. That then leaves the feller with a $0.00 balance which is NOT null. Zero means you don't have any, null means you never did.

Ah Spritzie, ever the linguist/english language inventor.

Actually in my OOTPX Master.csv file there were draft values, along with historical pitch ratings and positional experience ratings. The funny thing was the pitch ratings and positional experience ratings transferred perfectly into the game, while the draft value preferred to recite the alphabet. I'm sure if I had played with the historical database.odb that everything would've been peachy keen, but then I wouldn't be playing it my way would I?

I know Mr. Loaf made a killing from the whole "2 out of 3 ain't bad" line and while I don't think this is bad, I just find it weird that 2 of the 3 things introduced into the game worked with modded databases, but the third was seemingly non-transferable. Had all 3 not worked or all 3 worked I wouldn't have been surprised (probably disappointed in the first case, but not surprised), but to have 2 of the 3 work was strange. Please don't take this rant personally as it's not about pointing fingers, but rather it's my long winded statement of bewilderment and my way of saying: "Huh? WTF? How did that happen?"

[Spritze]

3 out of 3 worked for me in my modded db's. I never used the ODB file at all. Just goes to show you never know you know.

[Charlie Hough]

Originally Posted by Spritze

Ya know, roughly if a player is in the top 4,000 he is most likely a first round draft pick, top 8,000 = second round etc. The Koufax example is still a first round pick or close and he only had 5 good years and the rest pretty much stunk.

Koufax went in the FOURTH round of a rookie draft, not the first round. Are you actually running historical rookie drafts to see the results from these values? Are you looking at the Master.csv and doing a descending sort by draft value? Are you sorting the draft value field and looking at where some of these players fall far below another player who never even made it into the majors beyond a single call-up from the minors?

There aren't enough players for the math to work out in the way that you're thinking. There are fewer than 18,000 players in the database. In many historical seasons, there may be enough players for some 70 to 80 to be in in the rookie draft pool based on their debut dates. Those players are spread out based on these erroneous draft values, and so there are good players that end up going in the 2nd, 3rd, 4th or even 5th rounds. And then we see players who literally did not even play one full season in MLB going two rounds ahead of Sandy Koufax. Sorry, that does not compute.

There is no possible way that Garlon's formulas or anyone else's have any merit if the end result is Jimmy Zinn being ranked just one spot outside the TOP 100 PLAYERS IN ALL OF BASEBALL HISTORY. So he will go in the 1st round of a draft while Sandy Koufax, Jeff Reardon and Robin Ventura may go in the 4th or 5th round. Obviously these players are from different years, but every time I run a rookie draft for any season in baseball history, I'm seeing multiple bad picks by the AI that are based on draft values being too low and nonsensical for certain players.

So I'd like to know whether you're looking at the same thing that I am. Right now you're not giving me any confidence that we're going to see a legitimate fix, because you seem convinced that there isn't much of a problem.

But this isn't a matter of my personal taste. I can show you the draft logs and show you the comparisons to the draft values, and you will see a direct correlation that explains why great players are being taken in later rounds after absolute scrubs.

[Charlie Hough]

Just to add further evidence, I just created a new 1966 historical game, simulated the season, and then held the rookie draft afterwards.

The top 10 or so picks were reasonable, except Pat Kelly went #7 overall. Other than longevity, he never had much going for him and was a mediocre player. But the bigger problems start, as usual, as we get later in the first round and into the remaining picks.

Suddenly we see Luis Alcaraz go #12 overall. A career .192 hitter over four partial seasons, Alcaraz goes at this slot ahead of Jerry Koosman, Aurelio Rodriguez, Joe Niekro, Mike Torrez, and Mike Marshall to name a few. And those are just some of the players taken behind Alcaraz in the first round, let alone the later rounds.

But, wait, there's more. Clarence Jones goes in the last pick of the 1st round. He had 159 career at bats with 2 HR, 16 RBI, and a .248 average. Jones goes at the end of the 1st round, while Doug Rader, future standout starter Pat Dobson, future All-Star catcher Ray Fosse and Sparky Lyle go later in the 2nd round.

Meanwhile, I'm not even bothering to review every player and provide details. These are just the most obvious problems. The real scope of this is even a bit worse. As the 2nd round unfolds, a number of other crap players with basically no MLB careers are selected ahead of the players mentioned above.

Then there's the 3rd round. Here we find Ken Brett, a serviceable starter for many seasons, and Gary Nolan, another solid starter for a number of seasons. Yet they go far after Alcaraz, and Nolan even goes a number of picks after Glen Clark, who had a total of 4 at bats in one MLB season and never got a hit.

Alcaraz has a draft value of 4532, which when you sort by descending values makes him the 796th most valuable player in the historical database out of more than 17,000 players.

And guess what? That's a higher value than Koosman (4426), Niekro (4053), Rodriguez (4278), and everyone else down the line. Not surprisingly, Glen Clark's value of 1977 is higher than Nolan (1622), and barely below Brett's (2089) as Brett went just a few picks ahead of him.

The AI is basically drafting according to these draft values, with very few exceptions, and it's the values that are wrecking these drafts and leading teams to waste picks on Alcaraz and Clark instead of legitimate players. Now, once the good players are already drafted, then it's a free-for-all. Let the AI GM's draft Alcaraz and Clark all they want.

But I submit to everyone reading this thread that this is not a matter of one gamer's peronal taste or preference. This is just plain STUPID. And if we can't expect to see a MAJOR improvement here, then let's get a new project started so we can use WAR rankings or some other means to produce draft values for OOTP. This is nothing against Spritze or anyone else personally. Hopefully Spritze will run some new numbers, and this will be solved. But if the prevailing view is that what I'm seeing is somehow okay, then obviously I will have to find an alternative if all else fails.

[Charlie Hough]

As a final note, I just want to say that I'm convinced that this is really down to an issue with the formulas or the data sets and how they may have been calculated or prepared for dumping into the CSV. It may be that this all comes down to something fairly simple, and once the data gets updated, we'll be fine.

Based on many of the draft values, I think the formulas may be fine, but something has gone wrong with the data that they should have produced. But I just want to make sure that no one is suggesting that if the formulas give Alcaraz the 796th best value among 17,000 players, this is somehow acceptable.

[Grafton19]

How about drafting the 1st 1-10 rounds yourself and letting the AI do the rest? You point here is valid and it sucks we have to use a work around to fix it, but this game is too fun not to give some other ways around stuff a try.

[Scoman]

Thats always been a problem with every version of OOTP ,getting players in later rounds in the historical drafts that should not be there,really takes the challenge away from building a team. I always end up letting the AI draft for me so I dont get an advantage and thats never a whole lot of fun. I was hoping that had changed in 11. Seems to me a players draft value should be heavily based on his peak years. Hopefully this will be fixed.

[Spritze]

Fixed.

Leastwise better. Don't think we will all ever agree on the values.

1955 draft values

Bunning #1
Clemente #2
B.Robinson
Colavito
K.Boyer
L.Jackson
L.McDaniel
Pagliaroni
Ramos
E.Howard #10

Koufax #20 behind a lot of players Rough Tough Charlie Hough probably won't like
Moford #63 but that does not mean Moford cannot get drafted ahead of Koufax, just depends on what it depends on.
Westlake is #137, bottom feeder.

This run is a little bit different than it was previously due to a few of those pesky divide by zero challenges and some playing time weirdnesses in the Lahman data. Those have been adjusted adjudicationally.

[Charlie Hough]

I've been trying to work on some alternative formulas that are somewhat similar to a runs created and runs saved approach. One I have for pitching puts Koufax among the top four pitchers in the 1955 draft pool, but the issue is longevity. That tends to disrupt things too much when you factor in longevity. Hopefully I can find a way to factor the longevity in such a way that it still preserves the peak seasons value of players.

[Spritze]

Originally Posted by Charlie Hough

I've been trying to work on some alternative formulas that are somewhat similar to a runs created and runs saved approach. One I have for pitching puts Koufax among the top four pitchers in the 1955 draft pool, but the issue is longevity. That tends to disrupt things too much when you factor in longevity. Hopefully I can find a way to factor the longevity in such a way that it still preserves the peak seasons value of players.

Draft Value in OOTP is NOT peak season value.

The tree up which you are barking is not even in the correct forest. In fact if you cut down the tree up which you are barking after filling it with bells the sound of the fall won't even make the teeny tiniest hint of a noise in the correct forest.

[Spritze]

I have sent Markus a draft value update for the next patch.

I did find an error in my procedure that did not start the process with a clean slate (which it should have) as well as a couple of Lahman challenges and some challenges I had dealing with that pesky "divide by zero = infinity" programming falderal. Those kick my programatic sphinxter from time to time as I seem to never notice them until AFTER I finish testing.

So watch for an adjustment in the first patch.

[Charlie Hough]

Originally Posted by Spritze

Draft Value in OOTP is NOT peak season value.

The tree up which you are barking is not even in the correct forest. In fact if you cut down the tree up which you are barking after filling it with bells the sound of the fall won't even make the teeny tiniest hint of a noise in the correct forest.

I never claimed that OOTP's draft values are peak season values. I stated that MY work is aimed, among other things, at capturing the impact value of certain players that had shorter careers but had such strong peak seasons that they should not be rated below other players merely because of longevity. Hence, Sandy Koufax is valued far more appropriately in my output than what OOTP provides. So is Rocky Colavito.

Here is an attached file that shows the draft values for the top 40 players in the 1955 draft pool based on formulas that I'm testing. I'd like the others who have been posting on this thread to review my numbers and compare them to these players' real life statistics and historical value. I think they will find that I'm barking up precisely the right tree if we're going to see OOTP draft results resemble something respectable rather than complete nonsense.

[Scoman]

Originally Posted by Spritze

Fixed.

Leastwise better. Don't think we will all ever agree on the values.

1955 draft values

Bunning #1
Clemente #2
B.Robinson
Colavito
K.Boyer
L.Jackson
L.McDaniel
Pagliaroni
Ramos
E.Howard #10

Koufax #20 behind a lot of players Rough Tough Charlie Hough probably won't like
Moford #63 but that does not mean Moford cannot get drafted ahead of Koufax, just depends on what it depends on.
Westlake is #137, bottom feeder.

This run is a little bit different than it was previously due to a few of those pesky divide by zero challenges and some playing time weirdnesses in the Lahman data. Those have been adjusted adjudicationally.

That looks pretty good,but Koufax at 20 still seems off,but like you say everybody has their own views on draft values. Guys like Koufax are a hard read.

Spritze,lets say you were drafting 1955,were would you draft Koufax at in your own personal view. In mine,the only 2 guys I would even think of drafting ahead on Koufax would be Clemente or Bunning.

In my historical draft perfect world,Id say that Koufax should always go in the top 5 easily,because you know you are going to get 5-6 years of Cy Young caliber pitching.

[Spritze]

Plus you get 5 or 6 years of below average pitching which drags his career numbers down, plus if Koufax played in Brooklyn his whole career his stats may have been effected negatively since he wouldn't have Dodger stadium to play in, which is why the Draft Values are based on NEUTRALIZED stats not real stats. In a draft he probably won't be a Dodger. That drags him down a little more. When you add up all the little things that drag him kicking and screaming back to the pack maybe 20 is reasonable for a draft value based on his entire neutralized career. Myownself there are obviously players I'd take him ahead of knowing what I know about the things I know about. But 20? I don't know, it isn't that far off. I won't wet my pants over it. The numbers are just numbers and they fall out however they fall out. I just had Marvelous Marv Breeding beat out Teddy Ballgame for the batting title in my sim. Just goes to show, you never know.

[Charlie Hough]

Originally Posted by Spritze

which is why the Draft Values are based on NEUTRALIZED stats not real stats.

And yet it seems like most people playing historical games are not using neutralized stats. So does it make sense to base draft values that affect games with real stats on neutralized stats?

I don't believe that one person's philosophy about using neutralized stats should be forced on the rest of the customer base by having draft values for all games based on neutralized stats. Ideally there should be two choices that are applied depending on whether the player ratings in a saved game will be based on real stats or neutralized stats.

My formulas use real career stats, so those could easily be applied to games using real stats. People who want the Garlon neutralized version could select that when playing games based on neutralized stats.

[actionjackson]

Originally Posted by Charlie Hough

And yet it seems like most people playing historical games are not using neutralized stats. So does it make sense to base draft values that affect games with real stats on neutralized stats?

I don't believe that one person's philosophy about using neutralized stats should be forced on the rest of the customer base by having draft values for all games based on neutralized stats. Ideally there should be two choices that are applied depending on whether the player ratings in a saved game will be based on real stats or neutralized stats.

My formulas use real career stats, so those could easily be applied to games using real stats. People who want the Garlon neutralized version could select that when playing games based on neutralized stats.

It's not about one person's philosophy though Charlie Hough. It's about what the most accurate method is and since that's what you appear to be striving for, I'm surprised you can't see that. Unless you're playing an exact replay complete with players playing the exact amount of games with the actual teams they played with (and I'm not sure why you would bother because as other forum members have stated there are better [and more expensive] baseball replay products on the market, but to each his own) then the only stats you should be looking at are the neutralized ones.

Why you ask? Todd Helton's a very good hitter and would probably be so even without the pre-humidor Coors boost, but if you use real stats I guarantee you, you will be inflating his draft value higher than it should be. Same thing with pitchers Don Drysdale, Don Sutton, Orel Hershiser and yes even the one you're most concerned with Koufax. You're also going to deflate the draft value of Biggio and Bagwell for all those years they spent in the dead zone otherwise known as the Astrodome. Probably I shouldn't be looking at the greats because their fluctuations will be less severe than the players with shorter careers who benefited from their environs or suffered due to their environs. Neutralized stats level the playing field and allow for better comparisons between players, which is essentially what you want for the AI when it comes to draft decisions. A better comparison process theoretically should lead to better draft logic by the AI, which is what we all want.

My suggestion is to wait and see what Spritze has come up with in the next patch. Whether you take that with an entire mine full of salt or not is up to you.

[Charlie Hough]

Neutralized stats are ideal. The only problem is that the neutralized databases that the modders have developed go beyond simply neutralizing for ballpark factors. They also fill in missing seasons for players who missed seasons in real life. And I don't know if OOTP also incorporates any of their other database features, but they allow players to debut based on their minor league debut dates and not their major league debut dates. Their databases also include Negro league and foreign players that never played in MLB.

Again, I don't know exactly which features are included and which are not, but I'm not willing to give up that much historical accuracy in exchange for park neutralized stats. I don't want seasons filled in or anything else modified. Besides, the park factors should be coded to handle this on their own anyway. So if you move Biggio and Bagwell to Wrigley Field, then their performance should be adjusted by the ballpark factors. We shouldn't need a modified database to make this work.

And, on a final note, I agree that it's not about one person's philosophy. And it shouldn't be about two people's philosophy either. I should not be forced to use database values developed based on how two particular people choose to play OOTP. Markus should provide proper support for reasonable and realistic draft values for BOTH real and neutralized stats.

[Spritze]

1) The neutralized database in OOTP includes ONLY Major League players,
2) there is a check box to allow for seasons to be skipped as in real life in the Historical Game Setup function,
3) every player debuts in his rookie major league year and at no other time,
4) every player can retire after his last major league season via a checkbox on the Historical tab in Game setup,
5) there are no negro league nor any foreign players in OOTP.

6) The amount of erroneous information you are working from is really quite ...well, I can't think of a good word for it. I think perhaps it might improve your understanding of how the game plays and what the options actually are if you spent a little bit of time with the manual. It contains a wealth of information that might quiet some of your anxieties a little bit. Until you have a solid foundation of actual information to work from your assumptions will remain ever so slightly flawed.

You just need to put some extra time into knowledge gathering. It is mostly all in the manual. Instructions and directions can be a positive thing. Many people have worked hard at creating a useful and easy to understand manual. You can access it in game under Help.

[Charlie Hough]

Originally Posted by Spritze

6) The amount of erroneous information you are working from is really quite ...well, I can't think of a good word for it. I think perhaps it might improve your understanding of how the game plays and what the options actually are if you spent a little bit of time with the manual.

I appreciate your condescending post, but I am fully aware of all historical features in the game and the manual as well. I have been playing historical games in OOTP since 2001. I have also used the manual extensively and have helped provide content and proof read it. I have also quoted the manual numerous times on these forums to explain features to new users.

The one thing that the manual does not address, however, is the full extent of features in the neutralized database. The manual has always contained only two brief references to the neutralized stats database, and it mentions only its capacity to neutralize stats for park effects. The manual has never explained whether or not other features of the database are used, even though these features have always been mentioned in the threads about the database, and the manual has never explained whether using the neutralized database would override other historical settings.

This is why I stated in my post that I don't know which features of the database are included or not. For all I know, using the neutralized database could have overridden other settings such as player debuts, missing seasons, etc. Absolutely none of this is addressed in the manual, so that is why I used the phrase "I don't know" in my post.

But don't worry. I have already started modifying my own database with proper draft values, and I've had to step in run drafts myself to make sure that the AI does not continue to make absolutely suicidal draft picks. I can't continue to have my historical games ruined by the current state of things with draft values, and I'm not going to wait weeks for a patch that may not even address them to an appropriate level.

I'm also not going to have my draft values determined by neutralized stats with no option to base them on real stats.

I appreciate that you and Garlon have monopolized the market on draft values for use in OOTP and that neither of you are interested in addressing my concerns. So I'll deal with it on my own. I just find it ironic that you don't even use historical drafts and yet you are responsible in large part for setting the draft values in OOTP.

[Scoman]

Hey Charlie,have you checked out the draft values with the new patch. It is suppose to be fixed. I havent had a chance to check it out yet.