THE BOOK--Playing The Percentages In Baseball

In para 1, do you mean honest mess vs. tidy lie?

Not sure what you mean about “sets their positional defensive baseline as the negative of their offensive baseline.” Can you explain?  I have a vague idea what you mean; you’ve talked about this before ...

Ouch, corrected.

The theory: If the average SS is -14 offensive runs relative to all players, then you would make the average SS +14 defensive runs relative to all players. If Ozzie Smith is +26 runs, defensively, against all SHORTSTOPS, then he would be +40 runs, defensively, against all players.

The implication of this theory is that, offense + defense, the average 1B = average 2B = ... average RF.

This has no basis in reality.  No team pays like this.  No managers acts like this.  No fan thinks like this.  Only people with “noses in their spreadsheets” do this.

This is even more clearly not true in high school.  There’s no way in the world that the average SS = average 2B in high school.  It’s almost certain that the average SS is both a better hitter and better fielder than 2B, in high school. And if you look at first round picks over the last, I dunno, 20 years, and count the number of SS and number of 2B chosen, I’d bet you can count on one hand the number of 2B chosen.

The avg QB is not equal to the avg DT.

There are countless examples.

While you might be able to say that the replacement level QB = replacement level DT (they both get paid zero dollars), you can’t even say that for SS/2B, since the replacement level 2B will be a former SS.

I basically reject the entire notion of equality of average across positions.

I’m completely with you when it comes to second base vs shortstop.  Certainly 1B/RF/LF vs. DH as well. Not so sure on IF vs OF, though I understand what the money says on this issue.

I do think that average offensive performance has to be considered along with data from position switchers.  Regarding 1B and corner outfield, it could be that in certain time periods teams valued defense at 1st base more than in left field.  Looking at the NL today, left field looks at quick glance to have inferior defenders than 1B.

If the defensive talent gap between OF and 2b/3b/ss is only 2-3 runs, and we’ve got a 10-12 run gap in their offensive performance, we should be moving some 4th outfielders to infield and improving our team’s talent on the field.

If we can’t find enough outfielders that can actually do this, that tells me the the 2-3 run estimate is wrong, and teams are not paying their infielders properly.

Yes, if we can agree that the entire difference is strictly on the IF v OF population (and by IF, I mean 2b,ss,3b), we can move forward on researching and discussing this.

Dan Rosenheck has done a bunch of work that may be relevent to this discussion - see the HOM threads at BBTF. It follows up on some work of Nate Silver.

Over the time period in question he has 2nd base replacement level as being significantly above SS-which would agree with your conclusion that an average SS is worth more than a average 2B (given that the hitting averages are approximately the same).

Dan posted here a few months ago, and we had somewhat similar results, IIRC.

I’m slow and not getting the distinction.  Let’s take Michael Schell’s recent book (which I’ve been rereading), Baseball’s All-Time Best Sluggers.  In it, he ranks everyone by a linear weights-type system with lots of adjustments, including position played.

To make the position adjustment, he reviews the relative production of each position by era (like Rally’s data) and uses that to adjust the offensive linear weights.  That way, he comes up with a “position-adjusted offensive performance”.

Do you have an issue with that?  Or is your issue more with how analysts fold fielding contributions into the mix?

Apologies if that’s a dumb question.

Studes,

If players at one position are better overall than players at another position, doesn’t this put players at the weaker position at an unfair advantage if you do such an adjustment?

In other words, what’s the purpose of those position-adjustments?  It’s to try to put all positions on an even playing field, right?  Because some positions are harder to play, defensively, than others?  If there weren’t differences among position defensive requirements, you’d stick your best offensive team on the field all the time.  So the assumption is that shortstops deserve a boost because they’re better defensive players than first basemen.

Ok, in current MLB, second basemen and center fielders hit about the same, with a slight edge to center fielders.  And yet, center fielders are also better defenders than second basemen (based on Tango’s data, at least if they’re right-handed, the average center fielder would be a better than average-second baseman), making them a more talented group of players overall.  So why should we give second basemen as much of a boost, if not a greater boost, relative to what we give to center fielders?
-j

Studes/7: if you add up the total linear weights (off+def+pos), the total will equal to exactly zero for each position, as you are describing it.

And yet, the average 2B is paid far lower than the other positions.

You wouldn’t make the average QB = average DT, would you?

OK, suppose the average SS is -14 relative to all players, and the average 2B is -10.  Your argument is, why assume the SS are +4 better fielders than the 2B?

But then I reply, they can’t be worse.  If they were worse, you’d move your 2B to SS and finish with a profit. 

So the SS must be AT LEAST +4 better fielders than the 2B.  Which seems to be consistent with your argument.

Now, consider the 31st best SS.  Compare him to the 30th best 2B.  Suppose again that the SS offense is -4 compared to the 2B.  Why not move the SS to 2B?  You’ll lose 4 runs offense, but since we have proven that SSs are MORE than +4 defensive runs better than the 2B, you’ll make another profit!

It seems to me that if SS are collectively X offensive runs worse than 2B, they must be *close* to X defensive runs better than 2B.  If it’s more, you’d move some SS to 2B.  If it’s less, you’d move some 2B to SS.

Doesn’t there have to be an equilibrium?

(Of course, I say “close” above because learning a new position isn’t trivial, and defense may not be fungible that way.  But it’s probably reasonably close.)

Studes/7: if you add up the total linear weights (off+def+pos), the total will equal to exactly zero for each position, as you are describing it.

No, in my example there are no defensive linear weights.  Only position-adjusted offensive weights, based on the relative performance of all players at all positions.

In your case at the top, shortstops and second basemen would have equivalent offensive baseline adjustments.  Do you object to this for calculating relative offensive contributions?  If so, why?

Phil/10: There should be an equilibrium, especially at 2B/SS.  But for a 14-yr time period, they were virtually identical as hitters.  You can’t tell me that defensively, the 2B/SS over such a long time period were equals?

And the same applies for the 1B and corner OF.  There should be an equilibrium, but over a THIRTY year period, they were equals offensively.  But where do old OF go to die?  Well, ask Yaz, and Stargell, and all other corner OF: they go to 1B.  Why?  Because their fielding competition is so much worse that they can survive as 1B.

There should be equilibrium, but there is not.

Studes/11: To me, only considering half the equation, doesn’t help.  It couches what’s really happening.

But, yes, I object.  The offensive average for a DH is lower than that of 1B.  Why compare Ortiz to a lower offensive level than Pujols is compared to?

And what about in the late 40s when the offensive level of the CF was *higher* than 1B?  What purpose does it serve to compare the offensive output of your CF to the average CF’s offensive output?  So, you’ll say “my CF is +20 runs above the average CF, and my 1B is +20 runs above the average 1B, offensively...”

But, so what?  How does this help?  You need to know that the average CF is so much better fielding-wise than the average 1B.  Why would you care if someone is a CF or 1B, if you are not going to introduce fielding accomplishments in the mix?

You might as well compare your starter to the average starting pitcher and your reliever to the average relief pitcher.  It will similarly tell you nothing.

OK, I see what you’re saying.  Now, why don’t you think there’s equilibrium?  Is it because there’s too much correlation between hitting and fielding skill, so that if someone hits as well as a LF, he probably fields as well as a LF?  Is it because players have trouble switching positions?  Is it something else? 

For instance, if LF/1B are offensively equal, why don’t some teams fire their 1B and find the 31st-best LF to play first?  Offense won’t change, but the defense should improve immensely, right?

Just a thought: If declining outfielders move to 1B, than they bring down the average for 1B, assuming that the decline in hitting runs parallel to the decline in fielding. Same might be the case for the move 1B to DH. Maybe one should remove former OF of players past a certain age when establishing the baseline for 1B and DH. I realize this does not solve the 2B/SS problem.

Other than Gary Sheffield and catchers, no one should have trouble switching to 1B.  No SS should have trouble switching to 2B or 3B.

The IF to OF (and more problematically, OF to IF) requires a different skillset, so I can see trouble there.

Sometimes, you don’t have equilibrium because players aren’t very fluid.  Look at Pujols.  Dude should be at 3B, but if you have Scott Rolen (who should be at SS at some point), you’re not going to trade one of those guys to get equilibrium.

Plus switching positions comes at a cost of a few runs of learning curve.  So, you can’t constantly move players, especially if they are important to you.  You won’t risk it.  So, the equilibrium takes a long time to reach.  And as we can see with the 2B/SS, it might take decades.

Also, sometimes, you get special players who simply can’t move anywhere else.  Take ARod.  If he’s already a great hitter, and can field SS, there’s nowhere else for him to go.  If he’s a +60 hitter as a SS, that, by itself, automatically boosts the league average by +3 runs at SS.  That’s huge!  There’s no good reason to take him off SS.

And in the 16-team league, if you’ve got a Willie Mays in CF, that might be +4 runs in offense, just by himself (64/16).  If he were a bad fielder, you’d want to move him to the corners (which brings up the offensive baseline in the corners and lowers at center, to more typically conform to what we expect).

OK, I see where you’re coming from, and I think it’s a great point.  The 1B/DH comparison is particularly useful for me.

But I think there is room to consider other systems that aren’t set up like a linear weights above/below average system.

In particular, I see this issue differently than you do:

Why would you care if someone is a CF or 1B, if you are not going to introduce fielding accomplishments in the mix?

There are systems out there, like Schell’s, that look at position comparisons because of the relative supply of talent at a position.  They aren’t considering fielding accomplishments, only the relative scarcity of hitting talent at a position.

I think this is a particularly legitimate approach for catchers.

Definitely for catchers.  But, 2B is not a “position” independent of SS.  Virtually all 2B are a former SS.

The pool of players is:
Pool 1: catchers
Pool 2: Pool 1 + 2B+SS+3B
Pool 3: Pool 2 + OF
Pool 4: Pool 3 + 1B
Pool 5: Pool 4 + DH

You can’t look at 1B in isolation, without looking at the other guys that could play 1B (your corner OF, your 2b, your 3B, not to mention your Mark Belangers, and Darin Erstads).

This is why I reject the entire idea of looking at single positions.  It is not reality.

You cannot have a situation where the offensive accomplishments of your average SS is equal to that of your average 2B.

This approach would fail miserably in high school.

One way or another MLB teams are, and have been, doing something wrong.

Case 1:  Defensive talent gap between IF/OF is small, as Tango’s position adjustments reflect.  Since outfielders are unquestionably better hitters, they are being paid correctly. 

But teams continue to move infielders to the outfield, and very rarely move players in the other direction.  Some of the greatest outfielders in history were once infielders.  Teams made a mistake in not leaving Henry Aaron and Tim Raines at second base, and not leaving Mickey Mantle at shortstop.

Case 2:  The defensive talent gap is large, and/or the skills needed for each position are not transferable. 

Since you need 3 infielders just as much as you need 3 outfielders, teams are making mistakes by overpaying their outfielders and underpaying their infielders.

Not sure which case is actually true.

OK, a combination of position scarcity and ease of transition between positions, such as your outline, makes a lot of sense to me.

I did make it through high school, but it wasn’t easy.

I meant with High School Baseball Players (or College ball).  Not us as students!

Yoicks.  I’m glad you never saw me play baseball in high school.  There’s a reason I switched to basketball!

I just checked the stat website for my high school.  Last year their top 2 pitchers were also 2 of the 3 best hitters on the team.

I suspect this is fairly common at the high school level.

I would bet so.  The biggest player on the team would be the pitcher, no?