THE BOOK--Playing The Percentages In Baseball

One question is where the “every day” notion came from. I suspect it came from the players themselves. More than once, I have seen players (including pitchers) state that the real leaders on a team have to be non-pitchers, because they’re out there every day. Real or not, that’s how the players’ effect on chemistry is perceived internally.

So, if someone is going to look at leadership as a tiebreaker, maybe the everyday player should get the edge.

I think your comparison works well for DH vs AL pitcher.  But you are not giving Morneau credit for fielding.  He fields every inning, every batter.  So on defence he is working against every batter that Santana faces, and also against all of the batters that all the other Twins pitchers face.

Morneau doesn’t field 4500 balls in a season.  Depending on your position, he’ll field 2 to 5 balls per game.  At least half of those are “routine” balls that someone has to pick up (like a catcher).  You want to give a SS credit for an extra 300 plays, cool.  Want to give a 1B an extra 100 plays, cool too.  Want to give the P an extra 30-50 plays as a fielder.  Good.

Santana has faced 250 more batters than Morneau.  Seems to me, that this all works out in the wash.

Why does the pitcher get full credit for the defense?  Doesn’t he only contribute about half and the fielders contribute the other half?  That would cut Santana’s contribution to about 450 batters and advancing towards 15 wins or losses.  Even if you don’t give any credit to Morneau for fielding that puts him on top.

No, that’s not how it works.  Read this first, and then we can talk:
http://www.tangotiger.net/winshares.pdf

Tango, thanks for that I think it makes a little more sense to me now.  Let’s see if I understand it.

Santana has faced 892 batters.  He gets full credit for K + BB + HR (311) and half credit for balls in play (581 / 2 = 290.5) for a total of 601.5.  That looks pretty similar to Morneau’s plate appearances.

Now for defense.  I don’t know what Morneau’s actual stats are, but say he gets the feilding credit for 250 outs.  Give the pitcher half the credit for those outs and that leaves him with 125.  To add these to his PA’s we need to convert from outs to PA’s, so multiply by something like 4 / 3.  That gives him another 167 PA equivalents.  Add the batting and fielding credits and compare: Santana has about 80% of the playing time credit that Morneau has.

Now compare him to a shortstop who gets fielding credit on 500 outs.  That comes out to 333 PA equivalents.  Now Santana is down to about 65% of the playing time credit of a SS (with 600 PA).

To me that says that you can compare an SP like Santana straight up with a DH for playing time, but you have to give some significant extra credit to fielders.

What do you think?

One more thing.  On the Fangraphs site the WPA for batters and pitchers adds up to (W-L)/2 for a team.  (For Minn: batters: 1.88, SP 2.24, RP 9.88 for a total of 14.) That means that the fielders are not getting any credit.  I assume that is true for the WPA+ and WPA- as well.

In that implementation, right, no credit for fielders.

The important thing to remember is that what we care about is wins above some baseline level.  Right now, Santana is 4 wins above average and Morneau is 5.  If you take replacement level, they get even closer.

The whole playing time issue, on its own, is really completely irrelevant, since it’s already incorporated into the baseline level.

My point about the playing time issue is the “100% of contribution 20% of the time” is along the same lines as “20% of contribution 100% of the times”.

You still can’t do the “splitting up” the way you are doing it, but it’s better than nothing.

Let’s go back to WPA.

Santana has 17 WA and 13 LA (win advancement and loss advancement).  In there, we gave him full credit (and debit) for the actions of his fielders.  But, let’s say we give 100% of the credit to his fielders on every single ball in play.  What would happen?  It would knock out about two-thirds of the game advancement.  His 17-13 record would collpase to 7 WA and 3 LA.

How does that compare to Morneau’s hitting of 15 WA and 10 LA? 

For that, we need to know about replacement level.  Giving Santana total batters faced, a replacement level pitcher would have had something like 2.5 WA and 5.0 LA.  (Remember, all balls in play now go to the fielders.)

So, Santana is +4 WPA and the replacement is -2.5 WPA, for a total net of +6.5.

And if we had given all balls in play back to Santana?  His 17/13 record would be compared to 13.5 WA and 16.0 LA for the replacement, and again, a total net of +6.5.

It’s not how many batters you face, or how many games you play.  It’s the amount of impact you have, more than the bench player will have.

What if you had two pitchers who each contributed 4 WPA, but one did it in 200 IP and the other did it in 100 IP.  Which pitcher is more valuable, or are they equal?  I think the 200 IP guy is more valuable, because you get the same WPA plus you get an extra 100 IP at league average, which ought to be worth something.

You need replacement level.  That’s why I keep bringing up WA and LA.  So, you are talking about a guy with say 14 WA and 10 LA, and another is 8 WA and 4 LA.

The replacement level pitcher would have been say 11 WA and 13 LA in the first case, and 5.5 WA and 6.5 LA in the second case.

+6 wins above replacement for the first, and +5 for the second.

Got to admit that I don’t understand why win and loss advancements resolve the pitcher/batter argument.  The same objection would still hold: even though pitchers face more batters, or have more advancement opportunities, their impact is still limited to a subset of games.

I’m not saying that I necessarily agree, just that I don’t think WPA helps resolve the issue.

17 win advancement spread over 30 games or spread over 150 games is the same thing.  You’ve added 17 wins of win probability to the team.

If it’s not clear, make the distinction this way.  Allowing (or hitting) 50 HR is different if you do it over 30 games or over 150 games.  These 50 HR will not have the same win value.

But, the win value is the win value.  Adding +.50 wins in one game, or over 10 games, is still +.50 wins added.  This is the point of using win values, in that it adjusts for the context, so that if you allow a ton of HR and hits in one game, that each of those events are now less costly, since they occur in a blowout, and have less added win value.

Once we are on the same currency, wins, which is the final currency we care about, we now no longer need to do a conversion.

(Replacement level may still interest you though.)

OK, I can see the logic in using WPA for this purpose, but not total “game advancement,” which I understood to be your argument.  A player could, theoretically, incur most of his “game (win + loss) advancements” in just one or a few games and wind up with a small WPA but a large total absolute advancement.

Fangraph’s implementation of WPA does affect the issue by not allocating WPA to fielders.  I don’t think this would impact Santana very much, but it could impact some (non-strikeout) pitchers a lot.

I do plan to refer to replacement-level WPA in the Annual, and use it a bit.  The thread in which you laid out the methodology was very helpful.

Tango,

Thanks for your patience in explaining all this stuff. I realize that you are probably right because you helped define all this stuff and you have been working with these concepts in detail far longer than I have been casually reading about them.  Nevertheless, I am not quite convinced by your arguments.

I am happy to agree that 100% contribution 20% of the time is worth 20% contribution all of the time.  And I think that the SP’s contribution is a lot closer to the contribution of an everyday batter/fielder than it is to 20% of that, but I still think that the SP’s contribution < regular position player.

Why do we need to look at WSAB (or WPA over replacement) over WS?  WSAB excludes the contribution that a hypothetical bench player would have made replacing someone.  Why should a starting pitcher get more credit just because the hypothetical replacement pitcher is worth fewer runs than the hypothetical batter/fielder.  I can see where he might be paid more per win because he is a scarce resource, but does he really contribute the same number of wins? 

I’m looking at THT’s WS pages and if I sort by WS over the whole of MLB I see that Beltran leads with 37.  The first pitcher that shows up is Santana with 24 (ranked 24ish).  Santana moves up the rankings to 10 when we look at WSAB but he’s still way behind Beltran and Pujols.  On the Fangraph site Santana is also the best pitcher with 3.86 WPA.  But the best batters are again way above the SP (led by Pujols with 8.48).  To me it still seems like SPs are not worth as much as position players (at least in an MVP context).

2005 on THT looks similar to 2006 in that the top batters (Pujols, ARod) are way above top pitchers (Willis, Clemens).  Same again in 2004 (but more so with Bonds vs Santana).  It doesn’t seem to matter whether you look at WS, WSAB, or WPA the top SPs aren’t close to the top position players.

I never said that they were.  All I said is that the playing field is determined by win contribution, and not by “he played 150 games”.  And I only compared Santana to Morneau, not to Ortiz or Pujols or the other heavyweights.

I’ve long advocated that all free agent starting pitchers are terribly overpaid.

Therefore, I don’t disagree with your post.