Too many errors.

[Winks469]

I know this is probably a reoccuring problem, but I was just wondering how my left fielder commits 33 errors in 79 games. His defensive ratings may not be anything special, but you have to be pretty bad to make that many errors. Here is the guy: Dan Bryan (http://www-1.sportsmogul.com/bb/bg.pl?page=scouting&id=1908)

[takaplan]

He has a "D" rating for fielding. It makes sense to me...

Anyone below a C really shouldn't be playing in the majors...do you think someone with a "D" rating should not be making that many errors?

[dturkenk]

I don't think a D rating should involve that many errors. Look at Jeremy Giambi for example - he's rated as a terrible fielder by just about everyone. He has only 15 errors in over 230 games and almost 700 chances.

If below C means you shouldn't be a major leaguer, then I don't think you have enough breakdown above C to incorporate the difference between a Jeremy Giambi and a Rey Sanchez (or even a Jose Offerman and Rey Sanchez to equate positions)

[takaplan]

But when you say Jeremy Giambi is a poor fielder, you probably mean he has poor range. I disagree- if a B rated player makes 10 errors, and a D rated player makes 15 errors, doesn't that make the whole rating system pointless?

[Craigrrs]

The major league record is only like 30 errors in a season, I've seen players with a D rating with 200+ errors in a season. I don't care how bad you are, that's just insane.

[takaplan]

The 300 errors thing is a bug- we're working on that.

But having a "D" fielder make 50-70 errors makes sense; the reason you don't see that in the Majors is because anyone fielding that bad isn't on the field! In Mogul, a player that bad defensively shouldn't be in your lineup.

That's like saying an F contact hitter shouldn't hit .096 because no one in the majors hits that low. It's becasue anyone that bad wouldn't be getting playing time (except Greg Vaughn or Jeremy Burnitz, of course ;))

[Craigrrs]

ok 50 I could see, glad to hear that the insanely high errors bug is being worked on.

[dturkenk]

I agree with you about the F contact hitter batting under .100. However, I don't think fielding is the same sort of skill.

This could just be me, but I tend to think of range and fielding ability in the following manner - if you have good range, you normally field pretty well - if you field pretty well you may or may not have good range. I can run some numbers to check that at some point (or someone could do it for me - compare RF and fielding %) but that's how I always understood it.

And I might be able to understand 50 errors for a D fielding 2B or SS who have plenty of chances and where defense is more important - but I just don't see even a F fielder in a corner outfield position or at 1B having anywhere near that many.

Out of idle curiousity, if that's how the ratings are setup, where do you see the average amateur athlete - how far below F?

[dturkenk]

Ok, I ran those numbers I was talking about.

I took a look at players with over 200 chances a season after 1950 broken down by position. I then looked at the correlation between range factor ((A+PO)/G) and fielding pct ((A+PO - E)/ (A+PO)). Neither is a great defensive measure, but I think they're enough for what I'm trying to compare. Also, range factor would be more accurate if calculated as 9*(A+PO)/Innings but I don't have that data.

C: 2219 seasons worth of data - correlation= .304
1B: 1688 seasons worth of data - correlation = .466
2B: 1598 seasons worth of data - correlation = .404
SS: 1689 seasons worth of data - correlation = .121
3B: 1281 seasons worth of data - correlation = .494
OF: 3270 seasons worth of data - correlation = .193
I don't have the data to break out by outfield position which would have been more accurate too.


There's a few problems with my methodology I think - first that RF is largely era dependent. Since there are more strikeouts today, there are fewer chances for fielders than there were in the 60s. Fielding percentages do not depend on era, so this will lessen the correlation.

Anyway, my conclusions:

I was mistaken as to how strongly range and fielding ability were related. They are positively correlated, although less so than I would have expected. There's a fairly strong relationship between range factor and fielding percentage at 1B, 2B, and 3B but a relatively weak one at the other positions. The one that surprised me the most was the .129 correlation at the SS position.

Still, I just don't buy the argument that D defensive players shouldn't be major leaguers. I started a new game on the CD version and decided to see who had ratings of D+ and below. Here are some of the MLB position players who have fielding ratings of D+ and below and their errors and fielding %s from 2001 and 2002 in real life.


Player, Pos, Rating, 2001 E, 2001 %, 2002 E, 2002 %
Geronimo Gil, C, D+, 2, .985, 4, .991
Jacque Jones, LF, D+, 8, .983, 8, .982
Eric Karros, 1B, D+, 4, .996, 4, .995
Michael Barrett, C, D, 7, .993, 5, .990
Cliff Floyd, RF, D, 8, .972, 5, .983
Ray Lankford, CF, D, 6, .966, 4, .962


That's just a few of the players, but I think it's fair to say that D or D+ fielders can and do in fact play major league baseball with under 50 errors a season.

[takaplan]

Yeah, but your argument doesn't really make sense. The defensive ratings given to current players in the first year in the sim were set by the development volunteers, based totally on their opinion of the player's defense based on their statistics and watching them play.

So a D assigned to a player in the game cannot be compared to their performance in real life, as their evaulation is based on someone's opinion and research, and is not 100% fact. You might think Ichiro should have A+ contact, while the development volunteer might have thought it should be A-.

Maybe you have a point, that D fielders shouldn't be making 75 errors. But I think a D fielder with 50 errors is reasonable...I don't think it's that big of a deal, as anyone that bad wouldn't be starting anyways.

So what do you suggest the numbers be changed to? How many errors should a D fielder make for each position?

[dturkenk]

I'm confused by your point. I understand the ratings were given by volunteers and don't necessarily reflect completely accurately on the real world. But I'm not sure why you're saying that doesn't really affect the outcome. Maybe I'm wrong and that's not what you're saying.

To answer your question though, I think a D fielder (if we assume the db developers as a whole did a fairly good job of setting up the defensive ratings) should therefore have a fielding percentage somewhere between .920 and .990 (depending on position) or 1-8 errors per 100 chances.

Looking at the fielding stats for last year I think that for the positions, the absolute worst fielders should have:

1B - .975 fielding %, roughly 30 errors in a full season of work (rough estimates based on rough averages of the workloads of starters last year)
2B - .960, 28 errors
SS - .940, 42 errors
3B - .925, 27 errors
LF - .950, 13 errors
CF - .960, 14 errors
RF - .950, 15 errors
C - .970, 24 errors

And I do think this matters, because if you're getting more errors, you're letting more runners get on base and giving up more outs - so more runs are scoring.

[takaplan]

That's ridiculous. You're saying that a D rated player could be fielding .990?

My point is, you can't say: "Cliff Floyd made 8 errors in 2002, he has D fielding in the game, so therefore all D fielding outfielders should only make 8 errors"

Because maybe, in the game, Floyd should have B or B+ defense. Maybe the rating is inaccurate, which many might be as they were done by volunteers limited to resources like ESPN.com

And why your numbers are so high- because if a player in the Majors was a D fielder, he wouldn't be playing!

Answer this: would you start a pitcher with D control, and complain when he walks 12 batters per 9 innings? Would you tell me that in the majors, the worst control pitcher only walked 6 batter per 9 innings? that's because anyone with D control wouldn't be pitching in the Majors!

My question is, if a D fielder has a fielding % of .990, what does an B fielder have? .996? It doesn't make sense.

[dturkenk]

I admit the .990 is a little high, but I meant that for catchers based on the relatively few number of errors per catchers. And that playing time is an issue that throws difficulty into evaluating this situation.

However, I don't buy the argument that the ratings could be wrong though, unless they have no predictive value in the game.
If we can't match ratings and past performance to predict future performance how can we hope to understand what a player is going to do?

I think it's fair to say, here is Cliff Floyd, he started full time because he was such a good offensive player, but in the game he is rated as D+ defensively. I look at his player card, and see he has 8 errors last season. Since I know Cliff Floyd was not hurt nor at an age when players would traditionally suffer a decline, I would predict he would perform somewhat according to his previous stats. So I would predict a D+ defensive outfielder would have somewhere between 4-12 errors. Basing that logic on Cliff Floyd specifically is probably not the greatest move, but if the pattern holds through multiple players - which it seemed to, then that's how I establish my predicted level of performance

This has been my biggest complaint with BBM - that you cannot use realistically predict future performance based on historical patterns even if you take into account scout's ratings. I understand that sometimes players will have outlying seasons (see Bret Boone, 2001) but generally players will improve until they peak around 27 (both offensively and defensively, I'm not sure about pitching) and then decline until a somewhat precipitous decline around 32. Granted some players don't fit that trend (see Barry Bonds, et al.) but that curve has been pretty well understood for about 20 years now. Very few players experience the kinds of year-to-year ups and downs you see in BBM, which, at least to me, takes a great deal of the fun out of the game. If choosing players based on their past performance and likely future performance doesn't work, than why should I bother trying building what seems to be a good team. I've had plenty of teams go from 75 wins to 90 wins to 70 wins in 3 successive seasons with very little change (and they weren't all old teams).

I know that Football Mogul and large Baseball Mogul bugs are the main priority now, and I understand and appreciate that fact. But I have yet to hear anyone associated with Sports Mogul acknowledge that this is in fact an issue that should be dealt with despite repeated posts from long timers such as myself, Khan, Schaefling and others.

Ok, so this was a bit of a rant and as such is probably in the wrong section. But I just wanted to try to state what has been really my only frustration with an otherwise great game. I am definitely appreciative of all the hard work that has been put into this so far, and just hope it will continue to improve in the future.

[Craigrrs]

I asked James about Floyd a few days ago... it's an error in the database... he shouldn't be a D for fielding... the person that did his ratings forgot to give him predicted defensive stats.


Craig

[dturkenk]

Ok, so Floyd was a bad example. Thanks for letting me know. But that just illustrates my point, - how should I predict his play? Based on his historical stats, or based on his ratings? If it's based on historical stats, then why have ratings? If it's on ratings, that discounts any sort of empirical analysis and suggests you should spend anything to be number 1 in scouting. If it's some mix of the two, then what happens what ratings and performance are so varied (which happens in a lot of circumstances beyond the mistake in the DB for Floyd's defense)?

[takaplan]

No-
the only time historical stats and ratings do not go hand-in-hand are in the first years of the sim, when the real players still exist.

You see, any rating we give a real player is based on our evaluation of him, which may be inaccurate. However, these inaccurate ratings will indicate what happens in the sim.

So, if Cliff Floyd has good defense in real life, but was rated with D defense, then he will perform in the sim with his D defense.

Remember, once the computer generates rookies, this isn't an issue- the ratings you see determine all their stats, so historical stats are generated by the game engine, making it accurate.

The only problems are when the historical stats are not generated by the game engine, as with the real players, so their historical stats may not match with what they are ranked.

It's tough- I know. You occasionally see this happen in other situations...like where a guy has historical averages of 20+ HR but has like a C- for power and can barely put up 10 HR.

[dturkenk]

Ok. I didn't realize that was the case. It's makes it hard for the first couple seasons, but then it should be better off in the future. Thanks for the explanation. I've still noticed a great deal of year-to-year inconsistency, but apparently that's a different situation.

[takaplan]

Great- glad we got everything straightened out.

Talk about year to year inconsistency...
I just had Brian Giles hit .330 with 55 HR and 130 RBI in 2002, only to hit .230 with 15 HR and 55 RBI in 2003. :eek:

It is harder early on...but the later years are hard too. They are challenging as scouting really matters since you don't know the players from real life.

[malpits]

My problem is I picked up a starting pitcher with good stats and he was only 28 the pervious three seasons he had a below 3 ERA, the first seaons for me he only have a 2 ERA but after this for the next season and on wards he had and ERA of 5+ and I was under the view that most pitchers are normally at there best better the age of 25 to about 35 (and i know a few develop later or early) if they are lucky. so why did this go so mad, he was only pitching in Texas so if there would have been a problem it should have happed in his first.