Fan vs. PECOTA Projections

[El Gordo]

Just finished reading Bill James Jeter vs Everett article. Very entertaining. Taking his final conclusion:

But first, no one is saying that Derek Jeter is a lousy player. Let’s assume that the difference between Derek Jeter and Adam Everett is 72 plays on defense. That’s huge, obviously; that’s not a little thing that you blow off lightly. But almost all of those 72 plays are singles. What’s the value of a single, in runs? It’s a little less than half a run. 72 plays have a value of 30, 35 runs.

That’s huge—but it is still less than the difference between them as hitters. Derek Jeter is still a better player than Adam Everett, even if Everett is 72 plays better than Jeter as a shortstop. (Jeter created about 105 runs in 2005; Everett, 61.)

And applying Tangotiger's response to RShacks question:

Quote:

Originally Posted by rshackelford

So, when a SS prevents what should have been a single, isn't that worth more than just half-a-run? Shouldn't it be worth half-a-run plus whatever an out is worth? How much is an out worth? It's gotta be worth something, doesn't it?

Right. It's around 0.75-0.80 runs. Anyone who tells you otherwise: don't listen.

Then Jeter rather than being 9 runs better than Everett(+44RC to - 35 D-runs) is actually 10 runs worse(+ 44RC's to 54 D-runs). So the best fielding SS at .634 OPS in '05 is 10 runs better than the best hiiting SS at .839 OPS in '05.:002_sbiggrin:

[Frobby]

I understand what you're saying, but I disagree with your conclusion. The problem with your conclusion is that it's the same circular argument that gets used a lot around here. It's circular because it starts from the premise that says "stats = baseball" and then insists that any effort to refute it must be based on baseball stats, and that anything else is somehow wrong. Now, regardless of your opinion about the value of D, you gotta admit that that's circular reasoning.

Well, I think you misunderstood my point. What I am saying is, that if you believe the available stats do a good job of measuring what is really going on in baseball, then you basically have to believe that offense trumps defense, because that is what those stats show. And if you don't believe the stats do a good job, then all you can really do is argue your gut feelings, because the available stats don't support you. So a "believer" is going to dismiss the opinion of the "non-believer."

My statement wasn't intended as a value judgment as to whether the stats do or don't do a good job of measuring what's going on. Call me an agnostic on the issue. Somewhere in the last 12 months I wrote a thread in the articles section called "It's a game, not an equation," that detailed my views on this subject. I suspect that thread is long gone, however.

[RShack]

So a "believer" is going to dismiss the opinion of the "non-believer."

My statement wasn't intended as a value judgment as to whether the stats do or don't do a good job of measuring what's going on. Call me an agnostic on the issue. Somewhere in the last 12 months I wrote a thread in the articles section called "It's a game, not an equation," that detailed my views on this subject. I suspect that thread is long gone, however.

I looked for the post you mentioned, but I couldn't find it. Dunno why, it should be there. Anyway... personally, I'm not an agnostic about stats, I'm a believer in their value for helping us model the game. I just don't trust the current "state of the art" as being adequate, that's all. (Just like with hurricanes.)

'Just curious: What's your unsubstantiated hunch about whether future-stats will indicate that current-stats undervalue D?

[Frobby]

I looked for the post you mentioned, but I couldn't find it. Dunno why, it should be there. Anyway... personally, I'm not an agnostic about stats, I'm a believer in their value for helping us model the game. I just don't trust the current "state of the art" as being adequate, that's all. (Just like with hurricanes.)

'Just curious: What's your unsubstantiated hunch about whether future-stats will indicate that current-stats undervalue D?

You may recall (if you were reading the site at the time) that back in April I started a daily thread where posters offered opinions on every notable defensive play in every Orioles game. And concluded, at the end of April, that there were about 4-5 plays per game for the O's where the quality of the defense arguably could have made a difference (if I recall). Those included outstanding plays, plays that sometimes aren't made, plays that might have been made on another night, and truly bad plays. It was a bit anectdotal, but I believe pretty accurate. That's the basis for any hunches I have about the value of defense.

Overall, I believe players have more opportunity to influence games with their offense than with their defense. Whether the current stats have it exactly right, I doubt.

[Sports Guy]

You may recall (if you were reading the site at the time) that back in April I started a daily thread where posters offered opinions on every notable defensive play in every Orioles game. And concluded, at the end of April, that there were about 4-5 plays per game for the O's where the quality of the defense arguably could have made a difference (if I recall). Those included outstanding plays, plays that sometimes aren't made, plays that might have been made on another night, and truly bad plays. It was a bit anectdotal, but I believe pretty accurate. That's the basis for any hunches I have about the value of defense.

Overall, I believe players have more opportunity to influence games with their offense than with their defense. Whether the current stats have it exactly right, I doubt.

By WARP3 standards, ManRam is usually a 6-8 win player.

Considering that is partly based on rate I believe, that tells you how important the offense is...His rate is usually in the 80s...Terrible.

Yet, he is usually worth as many wins as someone like BRob, who is a solid all around player.

[El Gordo]

By WARP3 standards, ManRam is usually a 6-8 win player.

Considering that is partly based on rate I believe, that tells you how important the offense is...His rate is usually in the 80s...Terrible.

Yet, he is usually worth as many wins as someone like BRob, who is a solid all around player.

That's one way of looking at it. Another way is to say the negative value of his poor defense at the 2cd least important position on the field, is great enough to bring a 1.000 career OPS hitter to the level of a career .760 OPS hitter who is a slightly above average defender.

[Sports Guy]

That's one way of looking at it. Another way is to say the negative value of his poor defense at the 2cd least important position on the field, is great enough to bring a 1.000 career OPS hitter to the level of a career .760 OPS hitter who is a slightly above average defender.

And that is a great way of looking at it as well...I totally agree with your above point.

However, what I think that shows is that offense is valued much higher than things like base running and defense combined. That was my point.

However, as you said, the "other things", outside of offense, can bring a great hitter down to the level of a far inferior hitter but a hitter with a much better all around game.

You know, its funny, i always felt this board undervalued defense...And then, once DT came around, I feel many have started to overrate it.

[RShack]

Here you go:

http://www.insidethebook.com/ee/index.php/site/article/how_bad_a_hitter_can_you_be_to_justify_a_great_glove/

This is very interesting. That's what I said when I first looked at it yesterday. However, when I look at it again today, among the interesting parts I see a puzzling part that I don't understand yet. In there, you assign what appears to be a constant Win-value to each of the 8 positions. For SS, that value is +0.5 Wins. For 3B, that value is 0. I'm not sure what this is. When you apply your standards of what each position is worth, what does "worth" mean? I'm not arguing, I'm just confused about it.

Here's an example-question which I hope will make clear what I don't understand: Does this mean that if we take Player-X who is a mediocre 3B-man, and we move him to SS where he fields in a way that's half-a-win worse than he did at 3B, and if he hits exactly the same while playing worse-D, his value stays the same even though we moved him to a position where he plays worse? If so, why? Is it because we're saying that SS is half-a-win harder to play than 3B is? Or is it something else?

[RShack]

Somewhere along the line, I asked the question, "How much of a good fielder does LH have to be to justify playing him at SS, despite his crappy bat?" Well, for better or worse, I now have an answer. I'm not claiming that I have the world's best answer. I just have the best answer I could come up with. As with any answer to a question like this, it's based on certain assumptions about numerical equivalencies. The whole problem boils down to one issue: how-much-bat is equal to how-much-glove? It's impossible to get any kind of answer at all without using some kind of numerical basis for equivalence. So, I'll be explict about the ones I used, then everybody can be their own judge.

Here's what I used:

• RC = AB * OBP * SLG. I used 600 AB's because I figure we're talking about a fullt-time SS who's not at the top of the order.
  
• Unfortunately, this formula does not give you an automatic answer about how OPS translates into RC. This is because different combinations of OBP and SLG produce the same OPS but somewhat different RC.
  
• For LH, I used Bill James 2008 projection, which is .282 OBP and .296 SLG = .578 OPS. If you think you're smarter than Bill James, that's OK. I don't think you're smarter than Bill James, but there's certainly no law against you thinking you are. Based on what I say here, you can use your own spreadsheet to adjust the answer, based on your own expectations about how crappily LH will hit. Or, you can use the approximate factors I give below.
  
• For everybody else but LH, I calculated RC using a consistent proportion of OBP to SLG, based on something tangotiger said someplace. He said that he uses .340/.410./750 for league-average hitters (regardless of position). I figured if that's good enough for him, it's good enough for me. So, for comparison to other players' OPS, I used that same proportion between OBP and SLG. So, for example, if the OPS is.700, I used .317/.383/.700. If the OPS is .650, I used .295/.355/.650. It's just simple arithmetic based on the proportions of .340/.410/.750. Maybe that's not the best thing to do, but I had to pick something and that's what I picked.
  
• We need a value for calculating how much a defensive play at SS is worth, Fortunately for us, SS's are easy in this regard. If we were talking about 3B, things would be complicated because we'd have to worry about whether the hit he took away was a single vs. a double. But SS's mainly deal in singles, so it's easier. If you recall, tangotiger said that taking away a single is worth somewhere between .75 and .8. Evidently, the values change slightly year-by-year, based on what actually happens. Since 2008 hasn't happened yet, we don't know what it will be. So, I used the numbers from 2007, according to which a single is worth .479 runs, and an out is worth .303 runs. When a 2007 SS takes away a single, it's worth .782 runs in his favor, and when he gives up a single that he shouldn't have, it's worth .782 runs against him.
  
• From what people say, it seems that a league-average SS is worth 63 runs. Since we're comparing SS's to SS's, I don't see how that really matters, but I'll throw that in anyway, just to not upset anybody needlessly.
  
• I gather that tangotiger figures that anybody who plays SS is worth 0.5 wins because he's playing SS. The reasons for this are not clear to me, but I don't mind doing it. Since we're comparing SS's to SS's, it doesn't really matter. We can give them all an extra half-a-win or not, it doesn't make any difference in the end. I left that part out, but nothing changes is we add it in.
  
• To convert runs to wins, I used 10 runs per win simply because it seems that's what people do.
  
• To equate Runs Created to Defensive Runs, I used the same constant that Bill James uses in Win Shares. He doesn't divvy things up 50/50, he does it 52/48, such that a Defensive Run (aka a "run prevented") is worth a bit more than is a Run Created. Here's why I did this: Win Shares is basically an accounting method for considering a team's performance vs. its league, and for divvying up the credit among the various aspects of team performance. While it gets all blurry and if-fy about the specifics of a given guy's contribution, it's main job is to look at the big picture of how a team did and why. When looking at this, the actual numbers forced Bill James to conclude that a run-prevented is worth slightly more than a run-scored. AFAIK, nobody knows the exact details of how he came up with 52/48. But we do know that it's an empirical value, in the sense that he looked at everything else and determined that those were the best constants to use. We also know that it's based on an average team, and the exact numbers change slightly for good teams and bad teams, but the 52/48 values are close enough that Bill James uses them for doing Win Shares for everybody. What this means is that when we look at RC's, we have to mark them down by 4%, and when we look at DR's we have to mark them up by 4%. Now, if you think this is the wrong thing to do, then you're fight is with Bill James, not with me. I'm quite convinced that Bill James didn't start out wanting to do this, and that he would not have done this unless the actual numbers convinced him to do it.

The bullet points above fully describe the basis on which I calculated everything, so you can do it yourself if you want to. The answers I get are 100% reproducible, you can double-check them if you want to. Given all this, including the 4% adjustments to both RC's and DR's, here's how it turns out:

• On the defensive side, the results are constant: 12.3 plays at SS is worth 10 runs = 1 win.
  
• On the hitting side, it varies a bit. This is because 50 point of OPS is not worth a constant amount of runs. Moving up from an OPS of .600 to .650 is worth 8.9 runs, while moving from an OPS of .750 to .800 is worth 11.1 runs. If we focus on the range from .600 to .800, then it averages out to about 10 runs per each 50 points of OPS, but it's more for higher OPS and less at lower OPS. But, just to keep it more-or-less simple, we could say that 50 points of OPS is worth about 10 runs or about 1 win.
  
• So, 12.3 defensive SS plays is worth about-50 points of OPS.
  
• Using the more exact values for OPS instead of the average, gives us the following values for what LH would have to do in order to earn his keep, assuming he hits like Bill James thinks he's gonna:
  • If the Other Guy's OPS is .650, then LH will have to make 15 more plays at SS than the Other Guy to make the same net contribution.
    
  • If the Other Guy's OPS is .700, then LH will have to make 26.9 more plays at SS than the Other Guy to make the same net contribution.
    
  • If the Other Guy's OPS is .750, then LH will have to make 39.7 more plays at SS than the Other Guy to make the same net contribution.
    
  • If the Other Guy's OPS is .800, then LH will have to make 53.3 more plays at SS than the Other Guy to make the same net contribution.
    
  • If the Other Guy's OPS is .850, then LH will have to make 67.8 more plays at SS than the Other Guy to make the same net contribution.

According to this, those people who want the O's to get a SS whose OPS is .650 are betting that LH isn't 15 defensive-plays better than whoever that guy is.

For those who say the O's should get a guy who's OPS is .700, you're betting that LH won't be 27 plays better than whoever that guy is.

As one point of reference, the Bill James article about Jeter vs. Everett credits Everett as being 73 plays better than Jeter in 2005.

Using Jeter's 2005 OBP and SLG like the article did, and giving Jeter credit for the extra AB's above 600 that he actually had, then LH would have to make 70 plays more than Jeter.

On a final note, I would like to say that almost all of this can be blamed on Dan Bricklin at least as much as on Bill James. Because if Dan Bricklin hadn't invented the spreadsheet, none of us would be thinking like this. Too bad he didn't get rich from spreadsheets the way other people have gotten rich from his invention. I have no news of any kind about Dan Bricklin's personal finances, but I sure hope he got rich from something.

[Frobby]

If the Other Guy's OPS is .800, then LH will have to make 53.3 more plays at SS than the Other Guy to make the same net contribution.

If the Other Guy's OPS is .850, then LH will have to make 67.8 more plays at SS than the Other Guy to make the same net contribution.

According to your calculations, this describes what he'll have to do to be an overall improvement over Tejada. Basically, once every 2-3 days he has to make a play that Tejada would not have made.

On a gut level, that sounds a bit low. Tejada's OBP is going to be about 90 points higher, which in 650 PA's means 60 times that Tejada will get on base when Hernandez would have made an out. That by itself would negate most of the 67.8 extra plays made by LH, without even taking into account the differential in extra bases (probably 80 additional bases or so).

Well, I will leave it to the greater minds to explain why my gut instinct is wrong.

[RShack]

According to your calculations, this describes what he'll have to do to be an overall improvement over Tejada. Basically, once every 2-3 days he has to make a play that Tejada would not have made.

On a gut level, that sounds a bit low. Tejada's OBP is going to be about 90 points higher, which in 650 PA's means 60 times that Tejada will get on base when Hernandez would have made an out. That by itself would negate most of the 67.8 extra plays made by LH, without even taking into account the differential in extra bases (probably 80 additional bases or so).

Well, I will leave it to the greater minds to explain why my gut instinct is wrong.

I think it mostly comes down to whether you buy the idea that taking away a single is worth .75-.80 runs. I think that's most of it, right there.

[Sports Guy]

Ton Tango's research seems about right.

If LH can field at an Adam Everett level and give us a 609 OPS, he can be a ML average SS. That sounds about right to me.

However, if he is just above average and has a 550-580 OPS, as I would expect him to do, he is going to be well below average.

[square634]

I think it mostly comes down to whether you buy the idea that taking away a single is worth .75-.80 runs. I think that's most of it, right there.

Perhaps I'm not thinking this through very clearly, but I don't see why preventing a single should be worth more than hitting a single. Using those numbers (a single is worth .479 runs, and an out is worth .303 runs), I understand the theory that a single prevented is worth .782 runs.

However, when a SS hits an additional single beyond what the replacement SS would have hit (I'll call it a "marginal single") shouldn't the same .303 be added? For example, if Tejada reaches base 60 more times than LH, he also makes 60 fewer outs, so for each "marginal single", LH is costing his team .303 runs because he would have produced an out. That doesn't even take into account differences in slugging percentage. Does any of that make sense, or am I not thinking about this correctly?

[RShack]

Ton Tango's research seems about right.

If LH can field at an Adam Everett level and give us a 609 OPS, he can be a ML average SS. That sounds about right to me.

However, if he is just above average and has a 550-580 OPS, as I would expect him to do, he is going to be well below average.

The diff between the .609 OPS and Bill James' LH projection works out to about 5 runs over the course of the season. Big deal.

[RShack]

Perhaps I'm not thinking this through very clearly, but I don't see why preventing a single should be worth more than hitting a single. Using those numbers (a single is worth .479 runs, and an out is worth .303 runs), I understand the theory that a single prevented is worth .782 runs.

However, when a SS hits an additional single beyond what the replacement SS would have hit, shouldn't the same .303 be added if the replacement SS would have gotten out? For example, if Tejada reaches base 60 more times than LH, he also makes 60 fewer outs, so LH is costing his team .303 runs each of those times, while Tejada's singles give an additional .479 runs. That doesn't even take into account differences in slugging percentage. Does any of that make sense, or am I not thinking about this correctly?

The problem is that's not how anybody seems to estimate offensive value. Rather, the estimate is based on OBP and SLG. I'm not saying you're wrong, just that you're suggesting a method for estimating hitting value that's completely different, that's all.