THE BOOK--Playing The Percentages In Baseball

I don’t think Barry Larkin gets it.

They had to have someone “against WAR” on the segment, right? Maybe Larkin just drew the short straw.

To me it sounds like the only piece people ever object to with WAR is the use of the term “replacement” rather than average. You can sell people on WAA, even though the method isn’t much different. The answers are not the same, between WAR & WAA, but mainstream analysts and casual fans just like that one word better.

Which means we’re really just arguing over which set of players is going to be the comparison group. Not entirely trivial, but this is a smaller problem than “WAR is totally junk.”

WAA vs WAR is pretty close to trivial.  The rankings will be in the exact same order and separated by the exact same amount of wins.  I agree with some that comparing to average is a bit more intuitive.  Anyone positive is helping you be above .500, anyone negative is not.  But then you end up with solid major leaguers with negative WARs.  The average player is much better than most people realize.  The term ‘average’ is sometimes used with a pejorative connotation when in reality it is very difficult to find an MLB average player to fill a hole if you need to.  For that reason I prefer WAR.  Also it indicates when someone is performing at a sub major league level, which would not be obvious with WAA.  For the purpose of comparing MVP candidates or trade value or otherwise comparing players with each other the difference of comparison points is meaningless.  People who object don’t really understand.  The need for a comparison point like that is that you can’t use 0.  A team could be infinitely bad.  A high school team or a little league team would both win 0 games in a 162 game major league season but the high school team is much better. The little league team would likely have a .002 batting average where the high school team may bat .075.  Both would lose every game but .073 batting average points is a big difference so where do you put the 0 win level?

Schilling seems pretty smart in that clip.  Larkin not so much.  Doesn’t understand WAR at all.  That was a pretty generous interpretation of his comments.  Dancing around the obvious if you ask me - which is he really doesn’t get it at all.

Larkin had virtually no clue what WAR represents.  Since the purpose of the show was to discuss this stat, you would think that all the participants should/would familiarize themselves with it.  Maybe he tried and just couldn’t get it.  Who knows.

The only reason that “replacement” is the baseline, as far as I know, is to make it easier to value players for salary purposes, since the assumption is that you can get a replacement player at the major league minimum (which is essentially the definition of a replacement player)…

I’m fully on-board the WAR train. As is, most, if not all the audience here.

I’m not saying we switch to WAA instead of WAR, WAR is superior intuitively for the reason you state: lots of valuable MLB players would have negative values.

I wasn’t 100% sure the order of WAA & WAR would be exactly the same due to the replacement/average adjustments, so I fudged that a bit with the word “trivial.”

If you’re discussing WAR with someone and they get stuck at “replacement level,” you could still (probably) get them to concede to WAA. From that point on it should be easier to sell them on WAR because you’re just changing the comparison group. You could even show them that switching to WAR doesn’t change the order from WAA. A light bulb may just go off.

I’m not going to applaud Schilling or give the Joe Morgan treatment to Larkin, based on this one clip, TV is almost never what it presents itself to be.

"Replacement” is the baseline because a player who performs above replacement level but below ML average is adding positive, not negative value to his team. In addition to being worth more than ML minimum in terms of salary, a good player who has a few below average (but above replacement) years adds to, rather than subtracts from, his career value. A common mistake is to assume “average” players are readily available; the term does not sound impressive, but a simple thought exercise exposes the fallacy—rank all the regulars in MLB from 1-240 (8 per team), then remove all those ranked 121 to 240—what would happen to the average talent level? It would clearly drop, as they would be replaced by players ranked 241 to 360.

Obviously, a team needs above average players to win more than half their games. But championship teams can and do include below average players, and a player who provides 1 or 2 Wins Above Replacement is worth more than 0, -1 , or (horrors) Adam Dunn.

WAR and WAA will not have the same order when the players have different amounts of playing time.

WAR will give more credit to players with more playing time.

#7, I wouldn’t do it that way.  I don’t know about anyone else.

If a player plays 10 games (in a whole season), for WAR I simply give him the number of wins (or runs) above/below what a replacement player at that defensive position would do in 10 games.  For WAA, I give him wins/runs above an average player.  I think the rank order will always be the same.  I’d have to jot down some numbers to be sure.  What would someone else do?

MGL, take three hypothetical players.

Player A is +10 RAA in 160 games
Player B is 0 RAA in 40 games
Player C is -10 RAA in 160 games

if you use a replacement level of -20 RAA per 160 games, you get the following

Player A is +30 RAR
Player B is +5 RAR
Player C is +10 RAR

So the rank order changes.

One of the things that drives me nuts about the objections people bring up to “replacement level” is that people say that “there’s no such thing as a ‘replacement level’ player.” Perhaps that’s true, but there’s also no such thing as an “average” player.  It’s more a mathematical concept than a real thing.  It’s just that people have an easier time with “average” than they do with “a bit below average.” Blame the school system, I guess.

I find “replacement level” quite a bit easier to put a face to than “average”, without resorting to numbers.  Replacement players are guys like Mike McCoy and Drew Sutton who are continually bouncing up and down from AAA to fill in for injuries.

But “average”, I have to think a little bit to figure out who an average shortstop is in the AL this year, for example.  Jhonny Peralta? No, he’s above average.  Alcides Escobar? No, he’s below average.  I’d have to list out all the shortstops and rank them.

Mathematically, average is much easier and more straightforward to compute than replacement level, but as a simple baseball concept, replacement level makes a ton of sense.

Mike, OK got it.  Wow, I didn’t realize that it would change things…

Best way to think about it is from the pitcher perspective.  You have a(n effectively) .450 pitcher who pitches 180 innings (20 full games).  So, his equivalency would be as a 9-11 pitcher. This pitcher is either -1 win relative to average (10-10), or +1 win relative to replacement (8-12).

You can have an average pitcher who just doesn’t throw alot of innings (injuries or whatnot).  So, a .500 pitcher, who pitches 90 innings (10 full games).  That makes him a 5-5 pitcher.  He’s either +0 relative to average of +1 relative to replacement (of 4-6).

How you choose to view these two pitchers will tell you which of the two methods you prefer to look at.

There’s no wrong answer.  Just a question of which glasses to wear.

#9, very good Mike, I was planning to post a follow up about playing time. WAA is simply wrong whenever you compare players with different playing times. And good point about replacement level faces (#11)—the Reds have cut players like Danny Herrera and Jeremy Hermida, other teams pick them up, that’s replacement level.

"There’s no wrong answer.  Just a question of which glasses to wear. “

Here, I disagree with Tango. Take an extreme example, say Player A plays 160 games and is 1 run below average (roughly 19 or so runs above replacement). Say player B plays one game all season, but it’s one good game (3 for 4 with a HR, good defense). WAA is going to tell you that player B was the more valuable player over the season, which is clearly wrong.

The concept of “replacement” level is completely antithetical to a major league player’s core values.  You have a guy who has oriented his whole life about always giving 110% and now you have a system which says that the 601st best baseball player in North America isn’t worth anything.  All those long hours, sweat, tears, etc. don’t mean anything?  Are you crazy?  I honestly think it’s that simple.  For them, the baseline is not the AAA/MLB barrier.  It’s going home and never playing another inning or getting another plate appearance.  The baseline is zero.

There is woefully little discussion on the role of psychology and the acceptance of advanced statistics.  It’s not just they’re complicated and never appeared on the back of a baseball card.  It’s that they cross an invisible emotional barrier that makes people feel devalued in a world where the axe is always looming.

Smitty/14: if you are using one pair of glasses, you can’t then say that using the other pair of glasses to see what the first pair wanted is wrong!

If you JUST said: hey, this pitcher is 8-12 and that pitcher is 5-5 and just leave it RIGHT THERE, then we have no issue whatsoever.

The entire problem is turning the two dimensions of wins-losses, or runs and outs, or runs and PA, or runs and IP, into ONE dimension.

We both agree that one guy is 8-12 and we both agree that the other guy is 5-5.  Then, you figure out what your question is FIRST, and then I can tell you if WAR or WAA is the better method.

There is no wrong answer.  Or rather, each is an answer to a particular question.

Statistics like these (WAR, WAA) are improperly used when you use them to answer questions they are not designed to answer.  The example that Smitty gives is a good case in point. Player A is 1 run AA and player B is, say 2 runs AA.  Does that mean that player A was more valuable?  No!  Who says that WAA (or WAR) tells us how “valuable” a player is?  It might or it might not, depending on the context and depending on your definition of value.

WAA answers one question and one question only:  If an average player played in place in that player, how many theoretical runs would he have created (using some kind of lwts formula) as compared to that player?  Period.

You can do whatever you want with that answer, but please don’t assume that it translates to “value.”

Same with WAR.  It simply tells you the same thing, but instead of an average player, it uses a replacement player for comparison purposes.  Again, you can do whatever you want with that answer, but WAR, like WAA, does not necessarily tell you how much “value” a player has produced.

And even if either stat did, you are still left with the question of, “What does ‘value’ mean?” in that context…

16 - a replacement player does have value. On average, about half of the players at replacement level (not below) are on major league rosters making a half million dollars a year. Most of the rest are in Triple-A. That’s better than being 27 and still in Double-A, or being released at age 19, or never being signed, or being cut in a college walk on attempt.

for me, the rubber hits the road when you evaluate HOF credentials and want to compare Dick Allen v Tony Perez or Sandy Koufax v Jamie Moyer.  I’d argue that WAA is a more appropriate metric when you’re looking at the trade-off between brilliance and health/longevity.  My justification is that in the long term, the star will be replaced by an average-ish player. 

That said, I’m not sure the rank orders change by an amount that’s outside of the margin of error of the metric.  So, it may be a mostly theoretical argument/distinction that’s being made.

I thought the clip was ok.  They didn’t really get into WAR too much, just using it as a launching point.  It was ok for that, and, really, it would be a tough thing to talk about with 4 people in 3 minutes.  I don’t think anyone was embarrassed here, and everyone was respectful.

***

Jon said that he’s going to tackle BABIP next.  I’m not sure when that’ll happen.  Maybe tonight?  Maybe 10:30?  We exchanged several emails, and Jon really gets it.  I’m hoping it’ll make some impact, especially with Schilling there.